SEVENTH ANNUAL
WASTE TESTING
AND
QUALITY ASSURANCE
SYMPOSIUM
JULY 8-12,1991
GRAND HYATT WASHINGTON
WASHINGTON, D.C.
PROCEEDINGS
Volume II
-------
VOLUME
II
THE SYMPOSIUM IS MANAGED BY THE AMERICAN CHEMICAL SOCIETY
pi inted on recycled paper
-------
TABLE OF CONTENTS
Volume II
Paper Page
Number Number
ORGANICS
44. Methods for the Determination of Volatile Organic Compounds in Soil Samples. T. A. Bettor, U - 1
J. W. Eichelberger
45. Concentration of Water Soluble Volatile Organic Compounds From Aqueous Samples by n - 2
Azeotropic Microdistillation. M. L. Bruce, R. P. Lee, M. W. Stephens
46. Pollution Reduction in the Laboratory Through the Use of Smaller Initial Sample Size. n - 17
D. J. Bencivengo, B. N. Colby, P. W. Ryan
47. Extraction of Phenolic Compounds From Water Samples Using Styrene-Divinylbenzene SPE n - 27
Disks. C. G. Markell, D. F. Hagen
48. Comparison of Alternative Methods for Analysis of Volatile Organic Contaminants. n - 38
J. E Ryan, T. C. Voice
49. Evaluation of Sample Preparation Methods for Solid Matrices. V. Lopez-Avila, J. Milanes, n - 40
N. Dodhiwala, J. Benedicto, W. F. Beckert
50. Analysis for Selected Appendix DC Compounds in Environmental Matrices by High II - 53
Performance Liquid Chromatrography/Particle Beam Mass Spectrometry. J. L. Cornell,
J. C. Lowry, M. D. Tilbury
51. The Implementation of HPLC/Post-Column Techniques for Rugged Carbamate and n - 66
Glyphosate Analysis. M. W. Dong, M. V. Pickering
52. Determination of Low-Level Explosive Residues in Water by HPLC: Solid Solid-Phase n - 68
Extraction vs. Salting-Out Solvent Extraction. M. G. Winslow, B. A. Weichert, R. D. Baker
53. Reduction of Azo Dyes to Aromatic Amines for Environmental Monitoring. R. D. Voyksner, E - 82
J. T. Keever, H. S. Freeman, W. N. Hsu, L. D. Betowski
54. Hazardous Waste Component Identification Using Automated Combined GC/FTIR/MS. n - 84
R. J. Leibrand
55. Environmental Applications of Multispectral Analysis. J. M, McGuire H - 96
56. Sample Preparation Using Supercritical Fluid Extraction Methodology. W. F. Beckert, n - 97
V. Lopez-Avila
57. The Research Status of Supercritical Fluid Extraction for the Analysis of PCBs In Incinerator n - 106
Ash. R A. Pospisil, M. A. Kobus, C. R. Hecht
58. Supercritical Fluid Extraction (SFE) of Total Petroleum Hydrocarbons (TPHs) With Analysis H - 117
By Infrared Spectroscopy. Af. L. Bruce, R. P. Lee, M. W. Stephens
-------
59. Application of Supercritical Fluid Extraction of Dioxins/Furans From Soil and "PUF". n - 125
J-PHsu, J. C. Pan, K. Villalobos, G. P. Miller
60. The Application of Supercritical Fluid Chromatography to the Analysis of Herbicides and n - 126
Pesticides in TCLP Extracts. C. R. Hecht, P. A. Pospisil, M. A. Kobus, M. F. Marcus
61. Problem Solving in the Organic Extractions Laboratory: Herbicides. />. Smith, J, Doeffinger, n - 139
T. Wittwer, J. Giannella, C. Lott, L. Stanton, E. Alverson, S. Fitzgerald, K. Klinger
62. Infrared Microsampling for the Qualitative Analysis of Organics Extracted from Soil Samples. n - 152
M. R Putter, F. J. Weesner
63. A Performance Comparison Study of Different Types of Devices for Solid Phase Extraction. n- 153
K /. Lee, E. N. Amick, J. A. Berges, G. L. Robertson
64. Standard Reference Spectra for MS/MS Quality Assurance, Performance Evaluation, and n- 154
Proficiency Testing: X[rf]Q Tandem Mass Spectrometers. R. I. Martinez
65. Improved Techniques for Formaldehyde Analysis by HPLC Using Automated Sample n- 165
Preparation and Diode Array Detection. B. Goodby, S. Vasavada, J. Carter, L. Schaleger
66. An Interlaboratory Comparison Study of Supercritical Fluid Extraction for Environmental n- 179
Samples. T. L. Jones, T. C. H. Chiang
67. An Analytical Manual for Petroleum Products in the Environment. M, W, Miller, H - 181
M. M. Ferko, F. Genicola, H. T. Hoffman, A. J. Kopera
68. Evaluation of Liquid/Solid Extraction for the Analysis of Organochlorine Pesticides and PCBs n- 182
in Typical Ground and Surface Water Matrices. A. D. O'DonneU, D. R. Anderson,
J. T. Bychowski, C. G. Markell, D. F. Hagen
69. Improving the Analysis of Semi-Volatile Pollutants. C. Vargo, N. Mosesman, G. Barone n - 195
70. Electrospray Combined with Ion Trap Mass Spectrometry for Environmental Monitoring. n - 203
R. D. Voyksner, H-Y Lin
71.* Recent Advances in the Use of Supercritical Fluid Extraction for Environmental Applications. n - 205
/. M. Levy, A. C. Rosselli, D. S. Boyer, M. Ashraf-Khorassani
72. Using Supercritical Fluid Extraction to Separate Diesel From Soil Matrices. C. A. Craig, U - 206
S. Prashar, J. Cunningham, B. E. Richter, A. Rynaski
73. Creative Review of 'Tentatively Indentified Compound" Data Using the Retention Index. n - 217
W. R Eckel
74. High Efficiency GPC Cleanup of Environmental SamplesColumn Optimization. n - 232
G. J. FalUck, R. Cotter, R. Foster, R. L. Wellman
75. Efficient Aqueous Extraction Using an Emulsion Phase Contactor. K. R Kelly, L. C. Schrier, n - 242
K. C. Kuo
76. SFE Practical Applications for Environmental and Industrial Samples. L. J. D. Myer, n - 249
J. Tehrani, P. K. Mignon
INORGANICS
77. Microwave Sample Preparation Methods for Environmental Analysis. H. M. Kingston, n - 253
F. A. Settle, M. A. Pleva, L. lassie, P. Walter, J. Petersen. B. Buote
-------
78. Comparison of Procedures for TCLP Extract Digestion; Conventional vs. Microwave. EL - 255
V. L. Verma, T. M. McKee
79. Sample Decomposition in Closed Vessels with a Pressure Controlled Microwave Oven. n - 265
E Panholzer, G. Knapp, P. Kettisch, A. Schalk
80. State-of-the-Artof Microwave Digestion Methods for Environmental Analysis. M.E.Tatro n-266
81. The Application of X-Ray Fluorescence Spectroscopy for Rapid Hazardous Waste n - 271
Classification and Screening. P. A. Pospistt, H. Van Kley, M. F. Marcus, C. Taylor,
N. K. Shah, E. S. Tucker
82. Semi-Quantitative Determination of the Inorganic Constituents in Specific and Non-specific n - 282
Categorical Solid Waste Using EDXRF. K. Fennell, R. M. Olbrot, T. G. Howe
83. Identifying Sources of Environmental Contamination through Laser Sampling ICP-Mass n - 284
Spectrometry. K. J. Fredeen, M. Broadhead
84. ICP/MS Analysis of Toxic Characteristic Leaching Procedure (TCLP) Extract: Advantages H - 298
and Disadvantages. M. G. Goergen, V. F. Murshak, P. Roettger, I. Murshak, D. Edelman
85. Chromium VI: An Overview of Its Relevant Environmental Occurrence, Analytical Methods n - 312
of Quantitation, and Report on Recent Ion Chromatography Methods Development and
Validation Activities. L. B. Lobring
86. Rapid High Performance Microwave Digestion. R. Rubin, M. Moses n - 314
87. The Performance of a Low Cost ICP-MS for the Routine Analysis of Environmental Samples. n - 315
R. C. Seeley, T. M. Rettberg, P. D. Blair
88. Robotics for Automated Digestion of Environmental Samples. A. C. Grillo, C. Balas n - 317
89. Application of Laser Sampling ICP-Mass Spectrometry to Environmental Analysis. n - 318
E. R. Denoyer, K. J. Fredeen, R. J. Thomas
REGULATORY COMPLIANCE
90. Status of Developing Land Disposal Restrictions for Superfund Soils. R. Troast, C. K. Ofrutt, n - 321
J. O. Knapp
91. Certification Protocol for Meeting Organic Treatment Standards for Incineration Ash. n - 337
W. R. Schofield, J. W. Kolopanis, T. S. Johnson
92. Factors Affecting the Admissibility and Weight of Environmental Data as Evidence. n - 349
J. C. Worthington, K. G. Luka
93. Review of Groundwater Monitoring Requirements at RCRA Sites. W. G. Stek n - 350
94. The Paperless Environmental Laboratory: A Plan for Realization. /. C. Worthington, n - 361
G. A. Duba
95. Data Management Issues in the Hazardous Waste Industry. G. A. Austiff, J. Krecisz n - 362
Affi/GROUNDWATER
96. New Directions in RCRA Ground-Water Monitoring Regulations. /. R. Brown, V. B. Myers, H - 379
A. E. Johnson
-------
97. De-Mystifying the Problem of Filtered Vs. Unfiltered Samples. It D. Brown H - 394
98. Determination of Target Organics In Air Using Long Path Spectroscopy. R. D. Spear,
P. D. Greenlaw, R. J. Bath H - 409
99. Measurement of Toxic Organic Compounds in Landfill Gas Samples Using Cryogenic n - 410
Trapping and Full Scan GC/MS. S. D. Hoyt
100. The Determination of the Heat of Combustion and Water Content of Incinerator Feeds Using n - 411
Near Infrared Spectroscopy. N. K. Shah, P. A. Pospisil, R. A. Atwood,
D. L. Wetzel, A. Eilert
101. Source Sampling and Analysis Guidance: A Methods Directory. M. D. Jackson, n - 422
L. D. Johnson, K. W. Baughman, R. H. James, R. B. Spafford
AIR & GROUNDWATER POSTERS
102. A Field Investigation of Groundwater Monitoring Well Purging Techniques. V Maliby, n - 430
J. P. Unwin
103. Analysis of Polychlorinated Biphenyls in Water and Stack Emissions by High Resolution Gas n - 4SO
Chromatography / High Resolution Mass Spectrometry. E. A. Marti, J. Amin, H. S. Karam,
T. J. Yagley, A. F. Weston
104. Continuous Analysis of VOCs in Air Using a New, Phenyl-Methyl Silicone Stationary Phase n - 465
for High-Resolution Capillary GC. R. P. M. Dooper, N. Vonk, H. J. Th. Bloemen
GENERAL
105. Developing a Uniform Approach for Complying with EPA Methods. J. Parr, P. Sleevi, n - 479
D. Loring, N. Rothman
106. Performing TCLP Analyses to Get Meaningful Data. K Dolbow, J. Price n - 490
107. Total Cyanide By Photolysis. J. Gutierrez H - 492
108. Ammonia and Total Kjeldahl Nitrogen Determinations Using Flow Injection Analysis with n - 493
Gas Diffusion. J. P. Calvi, B. P. Bubnis, J-A. Persson
109. An Objective Criterion for Terminating Permeability Tests. M. S. Meyers n - 508
110. Sampling and Analysis Plans to Evaluate the Performance of Lead-Based Paint Abatement n - 521
B. S. Lint, J. Schwemberger, R. Cramer, B. Buxton, S. Rust, G. Dewalt, B. Lordo, J. McHugh
111. Further Evaluation of the Cage Modification to the TCLP. P. White n - 532
112. Comparative Study of EPA TCLP and California W.E.T. for Metals in Different Matrices. H - 533
G. S. Sivia, M. S. Iskander, J. T. Coons
AUTHOR INDEX
-------
ORGANICS
-------
44 Methods for the Determination of Volatile Organic Compounds in Soil Samples
Submitted by:
Thomas A. Bellar, James W. Eichelberger
Organic Chemistry Branch, Chemistry Research Division
Environmental Monitoring Systems Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
26 West Martin Luther King Drive
Cincinnati, Ohio 45268-1564
513-569-7512, FTS 684-7512
Grace Plemmons-Ruesink
Technology Applications, Incorporated
Methodology for the determination of volatile organic compounds (VOCs) in
soil and some other solid matrices has traditionally been fraught with
problems. Sample integrity is jeopardized when samples are manipulated to
introduce internal standards or surrogates, or when the sample is exposed to
the atmosphere while being transferred to the extraction device. There has
also been a problem with the incomplete extraction of the VOCs from the solid
matrices. Recently, instrument manufacturers have developed analytical
equipment specifically designed to efficiently extract VOCs from a variety of
solid matrices, while preserving the integrity of the original sample.
Evaluations of two such units, the Dynatech PTA-30 W/S and the Tekmar Model
7000 Equilibrium Headspace Analyzer, are described for the determination of a
broad spectrum of organic compounds contained in several soil types. For
comparison purposes, similar analyses were performed with both systems
according to Method 8260. Problems such as excessive amounts of water vapor
interfering with the reproducibility of the gas chromatographic retention
times are addressed. For both evaluations, several types of matrices were
fortified and analyzed. The same gas chromatograph equipped with a wide-bore
capillary column, and the same ion trap detector were used for separation and
measurement in both studies. The features of each instrument, accuracies,
precisions, and method detection limits are discussed for representative VOCs.
M-1
-------
45 CONCENTRATION OF WATER SOLUBLE VOLATILE ORGANIC COMPOUNDS FROM
AQUEOUS SAMPLES BY AZEOTROPIC MICRODISTILLATION
Mark L. Bruce, Richard P. Lee and Marvin W. Stephens
Wadsworth/ALERT Laboratories, Inc.
4101 Shuffel Dr. N.W.
North Canton, Ohio 44720
ABSTRACT
Methanol and other similar volatile organic compounds in zero headspace extracts and
other aqueous matrices can be analyzed by azeotropic microdistillation, followed by gas
chromatographic separation and detection. The method detection limits for methanol,
1-butanol and 2-methyl-l-propanol are at least an order of magnitude below the current
Land Disposal treatment standards using the Toxicity Characteristic Leaching Procedure
(TCLP).
A microdistillation system was developed to address the limitations of direct sample
injection, purge-and-trap and other azeotropic distillation systems. Sample volume
requirements range from 10 to 40 ml. The concentration factors range from 90 to 250
(depending on the analyte) with a 40 ml sample. The total distillation time is
approximately five minutes. Typical detection limits are between 5 and 15 ng/1 when the
distillate is analyzed by gas chromatography with flame ionization detection.
Aliquots of zero headspace extraction fluid and ground water were spiked with methanol,
1-propanol, 2-methyl-l-propanol, 1-butanol, 1,4-dioxane, acetonitrile, propionitrile,
acrolein, acrylonitrile and ethyl acetate at 0.10 mg/1 and 0.75 mg/1. Each aliquot was
distilled and analyzed in duplicate during a 10-day period. Accuracy and precision were
determined. System bias for most compounds was less than 15% (i.e., the average
percent recovery was between 85-115%). The relative standard deviation for percent
recovery for most compounds was also less than 15%. The microdistillation was most
effective for the alcohols.
INTRODUCTION
The Hazardous and Solid Waste Amendments of 1984 amended RCRA by banning all
land disposal of untreated hazardous waste within 5*/2 years after passage on
May 8, 1990. The basic purpose of the land disposal restrictions is to discourage
activities that involve placing untreated wastes in or on the land when a better treatment
or destruction alternative exists. Under the land disposal restrictions (40 CFR part
268.41) for spent solvents, methanol has a treatment standard of 0.25 mg/1 for
wastewaters containing spent solvents and 0.75 mg/1 for all other spent solvent wastes in
the waste extract using zero headspace extraction (ZHE). To date there are no EPA-
approved methods for methanol that have detection limits below these treatment
standards. The effect of this situation is that residues from the treatment of solvent
wastes and multi-source leachate wastewaters cannot presently be certified to meet the
corresponding treatment standards and thus cannot be landfilled.
This paper presents the development of an aqueous sample concentration, cleanup and
analysis method with a detection limit lower than the spent solvent treatment standards
1-2
-------
for methanol. The total sample handling time from the start of distillation to the
completion of analysis is less than one-half hour. The initial experimental parameters
were derived from a method for the azeotropic distillation of water soluble volatile
organic compounds (1,2). This method is based on the fractional distillation of
compounds which form azeotropes with water.* When distilling a 40 ml aqueous sample,
or ZHE extract, total distillation time, including warm-up, is five minutes. GC run time is
approximately 17 minutes. The distillate is free from nonvolatile organic and inorganic
interferences. These nonvolatile components may degrade gas chromatographic
performance and shorten the life of the GC column.
^INSTRUMENTATION, EQUIPMENT AND SUPPLIES
Gas Chromatograph/Data System
Hewlett Packard 5890 equipped with a flame ionization detector, Macintosh Ilci
(Apple) with Lab View (National Instruments) and GC Integrator & Workmate
(WillStein) software.
Gas Chromatography Columns
Quantitation: DB-Wax, 30 m X 0.53 mm I.D., 1.0 micron film thickness
Confirmation: DB-1, 30 m X 0.53 mm I.D., 1.5 micron film thickness
Hardware
Wadsworth MicroVOC^ Systemฎ, Shamrock Glass (see Figure 5.)
Round bottom flask, 100 ml, 14/20 joint
Fractionation column, 14/20 joint, 1.6 cm O.D., 1.3 cm I.D., 60 cm in length,
Shamrock Glass (see Figure 3.)
Pipe insulation, polyurethane foam, lV2" O.D., 5/s" I.D., 55 cm in length
Glass beads, 5 mm O.D.
Keck clamps, for 14/20 ground glass joint, Shamrock Glass
Glass reducing union, 14/20 ground glass joint to 6 mm O.D. tube, Shamrock Glass
(see Figure 4.)
Stainless steel reducing union, Vi6" to l/4"
Air condenser, Teflonฎ tubing, Vie" O.D., 1/32" I-D. (40 cm in length)
GC autosampler vials
Autosampler vial inserts, 100 |jl, calibrated
Graduated cylinder, 50 ml
Support stand with rod, 1 meter
Three-finger clamp
Heating mantle, Glas-Col, 115 volts, 230 watts, STM 400
Temperature controller, Glas-Col PL115-Cordtrol, 115 volts, 600 watts
Porous carbon boiling chips, VWR cat # 26397-409
Reagents and Standards
Ethanol, Everpure, 200 proof
Methanol, B&J Brand, purity 99.9%
1-Propanol, Baxter, purity 99%
2-Methyl-l-propanol, Aldrich, purity 99.9%
1-Butanol, Aldrich, purity 99.8%
1,4 Dioxane, Aldrich
Acetonitrile, Aldrich, purity 99.9%
* Note: Methanol does not form an azeotrope with water; nevertheless it can be
effectively distilled with this method.
1-3
-------
Propionitrile, Aldrich, purity 99%
Acrolein, Aldrich, purity 97%
Acrylonitrile, JT Baker, purity 99%
Ethyl Acetate, Aldrich, purity 99%
Reagent water, deionized
Zero headspace extraction fluid: Refer to Method 131 1 of the Federal Register Vol 55
No. 126, Friday, June 29, 1990, pg 26986-26998
The goal was to develop a sample preparation/introduction system which when combined
with GC-FID analysis would provide methanol method detection limits (MDL) below
0.1 mg/1, use less than 100 ml of sample and require less than 30 minutes of sample
preparation. Reaching the MDL goal would require a concentration factor of
approximately 30. Concentration factor is the ratio of the analyte concentration in the
collected distillate fraction to that in the original sample.
Many physical parameters were investigated, such as the sample volume, boil/reflux rate,
total distillation time and volume of distillate collected. The physical design
characteristics of the distillation system itself were investigated. Several
distillation/condenser designs were used: a commercial modified Nielson-Kryger and two
miniaturized Nielson-Kryger (Peters) systems. Several alternate overflow systems were
studied: the straight tube, notched, flared, wick and hoop systems (3). In addition, a
completely redesigned capillary condenser was developed. The capillary condenser
system was later refined into a more rugged form, the Wadsworth Micro VOC3. Two
chemical parameters were also studied: analyte concentration and matrix. Table 1 lists
the parameters that were studied.
Table 1. Distillation Parameters Investigated
Physical
sample volume
boil/reflux rate
distillation time
distillate volume collected
10 to 1000 ml
2 to 7 ml/min
5 to 120 minutes
2 ill to 20 ml
Physical design
Fractionation column
Distillation system design
collection chamber volume
condenser height/cooling surfaces
overflow design
overflow tube inside diameter
overflow tube height
capillary condenser
Wadsworth MicroVOC3
Vigreux, glass bead, sand, glass wool,
Rashig ring, spinning band
1 to 20 ml
15 to 60 cm, cooling coil, baffles
Peters/Dow, straight, notched, flared,
side drain, wick, hoop
2 to 10 mm
2 to 35 mm
Chemical
analyte concentration
matrix
0.025 to 10 mg/1
ground water, ZHE extract
11-4
-------
Modified Nielson - Kryger condenser
Initial experiments employed a commercially available modified Nielson - Kryger (N-K)
condenser from Ace Glass (3). Its design was similar to that described by Peters (2)
except with larger dimensions and sample removal through a stopcock was used rather
than a syringe (Figure 1). The N-K collection chamber volume was larger; 20 ml vs 1 ml.
Factorial design experiments* indicated that 70% recovery and an estimated detection
limit in the mid ppb range could be obtained for methanol with a distillation time of one
hour using a one liter sample aliquot. Azeotropic distillation appeared to be the right
process. However, this large scale system was not practical because of the long
distillation time and large sample volume requirement. Many miniaturized condenser
overflow systems were investigated (3). Most miniaturized systems were more practical
than the modified Nielson-Kryger system, but none produced a concentration factor
greater than 20. Examination of fundamental distillation principles led to a radical
change in condenser design.
t
E
E
o
c
t
-Cooling Water
Condensed Steam'
m Chamber
(20ml)
Overflow Tube
(4 mm ID)
Overflowing Water
\
Vigreux Column
2 L Flask
Heating Mantle
Figure 1. Modified Nielson-Kryger Condenser Distillation System
ANALYSIS
EPA SW-846 Method 8015 (modified) was used for analyzing the concentrated aqueous
samples. The analytical conditions are summarized in Table 2.
The instrument detection limit (IDL) was calculated to be 0.15 ng using 10 2 |il injections
of a 0.10 mg/1 standard. IDL = (tn-i,99%)(Std Dev) = (2.821)(0.0528) = 0.15 ng. A
special note of caution regarding the GC temperature program is in order. Even though
methanol and most other analytes elute relatively early, the GC column temperature must
be ramped high enough and held long enough to remove all water from the capillary
column. Retention time shifts may result if the water is not eluted from the column.
* Note: Factorial design is a statistical procedure which facilitates optimization of several
parameters at the same time. Precision estimates can also be obtained.
1-5
-------
Since methanol is a common laboratory solvent it is difficult to obtain methanol-free
water. One deionized water system contaminated reagent water with methanol. Also,
airborne methanol can be absorbed by water in open containers.
Table 2. Analysis Parameters
quantitation column
confirmation column
instrument calibration range
response factor %RSD
continuing calibration
response factor %D
injection volume
injection type
injection port temperature
temperature program
DB-Wax
DB-1
0.2 to 2000 ng
2 |il
splitless
180ฐC
external & internal standardization
earner gas
carrier gas flow
detector
detector temperature
hydrogen flow
airflow
make-up gas
make-up gas flow
helium
2.5 ml/min.
FID
230ฐC
37 ml/min.
426 ml/min.
nitrogen
30 ml/min.
30ฐC for 5 min., 5ฐC/min. to 70ฐC, 20ฐC/min to 150ฐC
PROTOTYPE VOC3
Previous N-K distillation systems had not
met the goals described above. A new
condenser design improved both the
concentration factor and the simplicity of
the distillation system. The capillary
condenser, an early prototype of the VOC^,
is shown in Figure 2. The fractionation
column and condenser were very simple
and inexpensive to make. The system
consisted of a sample flask, fractionation
column packed with glass beads (35 cm
length), capillary column (0.53 mm I.D.
and 35 cm length) and microcollection vial.
The capillary tube was normally water-
cooled. The first 10 to 100 ^1 of distillate
were collected in the micro vial. When
100 p.1 of distillate were collected a
concentration factor of 80 was achieved in a
7-8 minute distillation. The methanol
absolute recovery was 20%. The method
detection limit of methanol in reagent water
was 0.018 mg/1.
t
Air or Water Cooled
Capillary Column
Micro vial
Glass Bead Fractionation Column
Flask
Figure 2. Capillary Condenser
Various types of fractionation columns
were studied. Glass and Teflonฎ tubes
were packed with sand, glass wool, Rashig rings and glass beads. A spinning band
fractionation column was also studied. Small increases in distillation efficiency (relative
to glass beads) were found with some fractionation column types, but the columns were
either more difficult to clean or mechanically complex. Thus, the glass bead fractionation
11-6
-------
column was chosen as the best compromise between ease of use and distillation
efficiency.
WADSWORTH MTCROVOC3
The Wadsworth Micro VOC^ is a rugged version of the capillary condenser system
constructed from standard glass, stainless steel and Teflonฎ components. VOC^ is an
acronym for Volatile Organic Compound Concentration and Cleanup. The glass bead
fractionation column is constructed from glass tubing with standard 14/20 ground glass
joints (Figure 3). The air condenser consists of three parts (Figure 4): a custom glass
reducing union which converts from the ground glass joint to 1/4" glass tube, a stainless
steel reducing union which joins the V4" glass tube to a Teflonฎ tube and a Teflonฎ tube
(Vie" O.D., Vsz" I-D.) which was substituted for the 0.53 mm capillary column used in
the prototype. The complete system is shown in Figure 5. The total system cost is about
$300 with glassware comprising less than $70.
This microdistillation system more effectively concentrates methanol (and other alcohols)
than the previous prototypes. The concentration factors range from 100 to 250 depending
on analyte. The absolute analyte recoveries range from 20% to 60% (Table 3). The
microdistillation system is more effective than purge-and-trap or other azeotropic
distillation systems even though the absolute percent recovery is significantly less than
100%. Relative recoveries, calculated by using standards which are also distilled,
average 99%.
Table 3. Analyte Concentration and Recovery
Analyte
Methanol
1-Propanol
2-Methyl- 1 -propanol
1-Butanol
1,4 Dioxane
Acetonitrile
Propionitrile
Acrolein
Acrylonitrile
Ethyl acetate
CAS number
67-56-1
71-23-8
78-83-1
104-51-8
123-91-1
75-05-8
107-12-0
107-02-8
107-13-1
141-78-6
Typical
Concentration
Factor*
140
240
250
250
150
200
200
100
100
100
Typical
Absolute
Recovery*
35%
60%
63%
63%
38%
50%
50%
20%
20%
20%
Average
Relative
Recovery
100%
92%
86%
89%
100%
101%
96%
99%
116%
114%
* When a 40 ml sample aliquot is used and the first 100 ul of distillate are collected.
Method Summary
The azeotropic microdistillation method is summarized in Figure 6. A 40 ml aliquot of
sample is transferred to a round bottom flask. Boiling chips and internal standard(s) are
added to the sample. Matrix spike compounds are added when appropriate. The
distillation apparatus is assembled using Keck clamps at both ground glass joints after
insuring that the fractionation column and air condenser are completely dry. The sample
1-7
-------
14/20 Ground Glass Joint
Glass beads
Indentations
14/20 Ground Glass Joint
Figure 3. Fractionation column
Stainless Steel
Reducing Union
OD Teflonฎ Tube
1
1
T'oe
Glass Reducing Union
6 mm OD Tube
14/20 Ground Glass Joint
Figure 4. Air condenser
I-8
-------
Collection Vial
\
Reducing Unions
Keck Clamp
Fractionation Column
Insulation
Keck Clamp
100 ml flask
Heating Mantle
Figure 5. Micro distillation system
I-9
-------
Measure sample aliquot
I
Transfer sample to flask
I
Add I.S., spikes and boiling chips
I
Assemble distillation system
is heated to the boiling point (2-3 minute warm-up) and held at
a boil for 2 minutes. The first 100 |il of distillate are collected
in a microvial for analysis by GC-FID. All calibration
standards are distilled in the same manner as samples to
compensate for system bias since the absolute recoveries of
analytes are typically 50%. This calibration procedure is
analogous to purge-and-trap calibration procedures.
TECHNIQUE COMPARISONS
Four sample introduction/preparation techniques were
compared to this microdistillation for low molecular weight
alcohols such as methanol (Table 4). Direct sample injection
did not provide adequate analyte detection limits because there
was no concentration step. In addition, direct sample injection
deposited nonvolatile sample constituents in the
chromatographic system, which degraded performance. This
was particularly true for zero headspace extracts. Purge-and-
trap sample introduction did not meet the detection limit
requirements because the analytes were very water soluble and
thus difficult to purge. Absolute analyte recovery was very
low, (typically <1%) and highly variable.
Two modified Nielson-Kryger (N-K) azeotropic distillation p. (-
systems have been used. A large scale N-K system (5) did M A*?^
provide adequate analyte detection limits but required one liter Metnoa summary
of sample and a one hour distillation. The one liter sample requirement was problematic
since ZHE extraction produced only a few hundred milliliters. Also, the precision of
methanol recovery was poor (40% RSD). A small scale N-K system (3) did not meet the
detection limit (concentration factor) requirement. However, only a small sample aliquot
was required and the distillation time was short relative to the large scale N-K system.
The microdistillation system presented in this paper has the highest actual concentration
factor and lowest analyte detection limits of these five sample introduction/preparation
techniques. Sample volume requirements and equipment cost are low and preparation
time is short.
METHOD VALIDATION
A method validation study following the guidelines specified in the EPA Test Method
Equivalency Petitions guidance manual (4) was performed. A data summary of the
aqueous matrix study for samples spiked at two concentration levels is presented below.
Two sample matrices were studied: ground water and ZHE extraction fluid. The ground
water was taken from a residential drinking water well. It was high in calcium,
magnesium and iron content. The ZHE extraction fluid was prepared from reagents with
low methanol content. Appropriate amounts of each matrix were spiked with each of the
compounds listed in Table 5. The spiking concentrations were 0.10 mg/1 for the low
concentration spike and 0.75 mg/1 for the high concentration spike. Both matrices
contained low concentrations of methanol. Unspiked aliquots of each matrix were
11-10
-------
processed and analyzed to allow the percent recoveries to be corrected for the "native"
analyte concentrations.
Each spiked matrix was subsequently shaken briefly (with minimal headspace) to
homogenize it. Each spiked and un spiked matrix was divided into sample aliquots and
stored in glass 40 ml VOA bottles with Teflonฎ lined caps at 4ฐC with zero headspace.
Each day for 10 consecutive working days each matrix was distilled six times: two
unspiked samples, two low concentration spikes and two high concentration spikes. A
total of 12 "samples" were distilled each day. All calibration standards were distilled in
the same manner as the samples to automatically compensate for system bias. An internal
standard (ethanol) was used to improve precision. Analysis of Variance (ANOVA) was
used to estimate method accuracy (bias) and precision.
Table 4. Technique Comparison
Method* Theoretical
Concentration
Factor
Direct sample
injection
Purge-and-trap
Nielson-
Kryger(5)
Nielson-
Kryger(3)
Wadsworth
MicroVOC3
1
2500**
350
8
400
Absolute
%
Recovery
100
1
40
100
50
. . Method Sample ,
Actual Detection Preparation ^lQ
Concentration LimU Time Volume
Factor mg/1 minutes ml
1
25
150
8
200
2
0.1
0.05t
0.3
0.01
0
10
60
10
5
0.002
5
1000
40
40
Notes and Equations:
* Assume a 2 ul injection into the GC-FID for comparison purposes.
** The purge-and-trap TCP assumes a 2 ul final sample volume to be consistent with the injection volumes
used by the other techniques.
t The methanol recovery precision is low so the method detection limit is not improved as much as expected
based on the actual concentration factor.
, ,. . . ,. f /-V-,T*\ original sample volume
Theoretical concentration factor (TCP) = & i- -
final prepared sample volume
Absolute % Recovery = amount of analyte in prepared "sample" .
amount of analyte in original sample
Actual Concentration Factor = TCF (Absolute %Recovery / 100)
Direct Inject DL
Method Detection Limit (estimated) =
Actual Concentration Factor
' Sample prep precision
The method detection limits (MDL) for both matrices are shown in Table 5. The
detection limit was calculated from two different data sets. The one-day detection limit
was derived from seven replicate analyses performed on the same day. The 10-day
11-11
-------
detection limit was derived from the equivalency study data and consisted of 20 replicates
spread over 10 days. The one-day MDL is often much lower than the 10-day MDL. This
is expected since day-to-day reproducibility is usually not as good as same-day
reproducibility. In addition, the distillates from the one-day ground water detection limit
study were analyzed on a less sensitive GC. The detection limits remained essentially
unchanged. This indicates that in this study the precision of the distillation is the limiting
factor for method detection limits. Thus, using a less sensitive detector will not
necessarily raise the method detection limit. Regardless of the GC used for analysis, the
methanol, 2-methyl-l-propanol and 1-butanol method detection limits are well below
current land disposal treatment standards.
Table 5. Target Analytes
Analyte
Methanol
1-Propanol
2-Methyl- 1 -propanol
1-Butanol
1,4 Dioxane
Acetonitrile
Propionitrile
Acrolein
Acrylonitrile
Ethyl acetate
Method Detection Limit* (mg/1)
Ground Water THE Fluid
Iday1 Iday2 10 day1 1 day1 10 day1
0.008
0.007
0.005
0.002
0.007
0.004
0.002
0.012
0.010
0.011
0.014
0.005
0.007
0.009
0.012
0.005
0.005
0.021
0.020
0.021
0.017
0.029
0.018
0.026
0.022
0.037
0.082
0.10
0.11
0.008
0.018
0.004
0.004
0.018
0.030
0.011
0.019
0.014
0.015
0.028
0.024
0.029
0.027
0.042
0.027
0.029
0.080
0.092
0.089
* Microdistillation with modified 8015 analysis.
1 GC number 1, nominal instrument detection limit 0.
2 GC number 2, nominal instrument detection limit 0.
lmg/1
5tol.0mg/l
The method may be extended to 2-butanone, 2-propanol and acetone, but these
compounds were not included in this study.
The results of the equivalency study are summarized in Tables 6 and 7. No outlying data
points were found in any of the data sets. The day effect was significant for some data
subsets. Day effect is statistically significant when the precision within days is much
better than the precision between days. This is a normal situation for analytical
procedures. The bias column in Tables 6 and 7 shows the 95% confidence interval of
analyte fraction recovered. A value of 1 corresponds to 100% recovery. The lower
bound for precision is the lower limit of the 95% confidence interval of the true variance
(of analyte recovery). The EPA has used 0.25 as an example maximum (4).
Figures 7 and 8 graphically present the bias data of Tables 6 and 7. The 95% confidence
intervals (CI) of analyte percent recovered are plotted for both low and high spike levels.
Most 95% CIs are small and near 100% recovery. This indicates that both accuracy
(bias) and precision are good.
Overall the method is very effective for concentration and cleanup of the two aqueous
matrices studied. The alcohols exhibited excellent accuracy (bias) and precision. The
nitrile results were also quite good. The method is not as effective for acrolein,
1-12
-------
acrylonitrile and ethyl acetate although it may be adequate for some uses. Ethanol was
not a good internal standard for these three compounds. A more appropriate internal
standard may solve most of the precision problems associated with these compounds.
Table 6. EPA Equivalency Study-Analysis of Variance (ANOVA) Results for THE Fluid
Analyte
Methanol
1-Propanol
2-Methyl- 1 -propanol
1-Butanol
1,4 Dioxane
Acetonitrile
Propionitrile
Acrolein
Acrylonitrile
Ethyl acetate
Low Concentration
Outliers Day Bias
effect
No
No
No
No
No
No
No
No
No
No
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
1.02-1.12
0.92-1.04
0.81-0.92
0.82-0.95
0.95-1.10
1.04-1.17
0.94-1.05
0.89-1.28
1.06-1.40
0.97-1.41
Lower
bound for
Precision
0.007
0.006
0.008
0.007
0.017
0.007
0.008
0.059
0.082
0.072
High Concentration
Outliers Day Bias
effect
No
No
No
No
No
No
No
No
No
No
Yes
No
No
No
Yes
No
No
No
No
No
0.97-1.06
0.87-0.92
0.77-0.87
0.83-0.90
0.97-1.06
0.90-1.00
0.83-0.97
0.70-1.03
0.84-1.21
0.87-1.28
Lower
bound for
Precision
0.003
0.002
0.007
0.003
0.003
0.007
0.015
0.080
0.096
0.123
Table 7. EPA Equivalency Study-Analysis of Variance (ANOVA) Results for Ground Water
Analyte
Methanol
1 -Propanol
2-Methyl- 1 -propanol
1-Butanol
1,4 Dioxane
Acetonitrile
Propionitrile
Acrolein
Acrylonitrile
Ethyl acetate
Low Concentration
Outliers Day Bias
effect
No
No
No
No
No
No
No
No
No
No
No
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
0.90-0.98
0.87-0.95
0.80-0.94
0.82-0.91
0.92-1.01
0.99-1.10
0.93-1.12
1.00-1.39
1.18-1.68
1.07-1.63
Lower
bound for
Precision
0.004
0.003
0.008
0.003
0.006
0.005
0.013
0.063
0.098
0.117
High Concentration
Outliers Day Bias
effect
No
No
No
No
No
No
No
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
0.91-0.98
0.88-0.94
0.82-0.96
0.85-0.95
0.94-1.04
0.87-0.97
0.81-0.98
0.71-0.89
0.82-1.06
0.77-1.08
Lower
bound for
Precision
0.003
0.003
0.008
0.004
0.004
0.004
0.010
0.015
0.024
0.032
LIMITING FACTORS
The method is most effective for water soluble compounds having a boiling point low
enough that the distillate is enriched in the target compounds relative to the original
sample. The precision with which the distillate is collected significantly affects overall
method precision. The distillation rate also affects method performance. System and
reagent contamination must be kept to a minimum. The type and condition of
fractionation column affect the recovery of the target analytes. Although the
microdistillation removes many nonvolatile and semivolatile interferences, it does not
remove interferences from nontarget water soluble volatile organic compounds. Absolute
analyte recovery ranges from 20 to 65%. Although the recovery is significantly less than
100%, the bias is consistent and the results can be corrected to account for this limitation
11-13
-------
using internal standards and calibration procedures similar to the purge-and-trap
technique.
140 --
130 --
T3
2 120
CD
I 110
CD
tr
+ฑ 100
CD
y
90 --
80 --
70
0 0.10 mg/1 spike concentration
ฐ 0.75 mg/l spike concentration
I
^ o^
Methaflol 1-Propanol 2-Methyl-1- 1-Butand 1,4Dbxane Acetonitnle Propionitrile Acrolan Acrytonitrile Ethyl acetate
propand
Figure 7 EPA Equivalency Study-Bias Results for Zero Headspace Extraction Fluid
170 '
150 -
CD
I 13ฐ
CD
EC
100
90 -
70
o 0.10 mg/l spike concentration
D 0.75 mg/l spike concentration
o1?
Methanol 1-Propanol 2-Methyl-1- 1-Butanol 1,4Dioxane Acetonitrile Propionitrile Aaolein Acrylonitrile Ethyl acetate
propanol
Figure 8 EPA Equivalency Study-Bias Results for Ground Water
The target analyte must be sufficiently volatile to be distilled from the aqueous sample.
Significant enrichment of the analyte in the distillate (relative to the original sample) only
occurs when the vapors released from the boiling water have a higher analyte to water
ratio than the original sample. This happens when the analyte forms an azeotrope with
water which is > 50% analyte. If the azeotrope is < 50% analyte no enrichment of the
vapors will take place in the fractionation column. Some low boiling analytes such as
11-14
-------
methanol do not form an azeotrope with water but still are effectively concentrated by
this system. In general this method is most effective for compounds that have boiling
points below that of water. However, some butanols have boiling points higher than
water but form azeotropes that boil at less than 100ฐC. This method appears to work for
such compounds. It does not work for compounds such as 2-ethoxyethanol which form
an azeotrope that is predominantly water.
The precision with which the distillate is collected significantly affects method precision.
We recommend collecting the FIRST 100 jj.1 of distillate. This seems to be a reasonable
compromise between maximum concentration factor and ease of handling. The first few
\i\ will contain the highest concentration of analyte; however, it is very difficult to
manually collect this small fraction in a reproducible manner. Larger volumes such as
1 ml can be collected. However, this significantly reduces the method concentration
factor. The volume collected should be 100 ฑ 20 fj.1. An internal standard (added prior to
distillation) should be used to help compensate for these small variations in volume in the
same manner that an internal standard compensates for purge-and-trap, chromatographic
and detection variations.
The distillation rate also affects method performance. If the distillation rate is
significantly higher than 2 ml/minute the fractionation column may not function
efficiently. The analyte enrichment in the distillate may be reduced. If the distillation
rate is too slow the distillate will not reach the air condenser or the distillation may take
too much time.
System and reagent contamination must be kept to a minimum. The specific maximum
contaminant concentration varies according to the quantitation limit required. Methanol
and acetone are common contaminants in deionized water, reagents and laboratory air. If
either of these compounds are target analytes, special laboratory practices may be
necessary. Some water deionizers actually increase the amount of methanol and other
potential target compounds in the laboratory water system. High purity reagents may also
be necessary, particularly in the preparation of ZHE extraction fluid.
The type and condition of fractionation column affect the recovery of the target analytes.
For best reproducibility and efficiency, the fractionation column, reducing unions and air
condenser must be completely dry before use. Only 50 p.1 of water in the condenser can
seriously reduce the analyte concentration in the distillate. Therefore, the entire
distillation apparatus should be oven-dried before use.
Although the microdistillation removes many nonvolatile and semivolatile interferences,
it does not remove interferences from nontarget water soluble volatile organic
compounds. Nonvolatile sample components will not be distilled and thus will not be
introduced into the GC. Most semivolatile components will also be eliminated or greatly
reduced. This greatly reduces contamination of the injection port and GC column. Many
water soluble volatile organic compounds may be collected in the distillate. Some of
them may be difficult to resolve chromatographically. For example methanol, 2-butanone
and 2-methyl-2-propanol elute very closely on a polyethylene glycol stationary phase
(J&W DB-Wax). Such interferences may require different GC columns and/or detectors
to resolve.
The method bias due to low analyte recoveries can be corrected by using an internal
standard and distilling all calibration standards. This is similar to purge-and-trap
11-15
-------
procedures except that the microdistillation system is not directly interfaced to the GC at
present.
CONCLUSION
Methanol and other water soluble volatile organic compounds in zero headspace extracts
and other aqueous matrices can be analyzed by azeotropic microdistillation, followed by
gas chromatographic separation and detection. The method detection limits for methanol,
1-butanol and 2-methyl-l-propanol are much less than the current land disposal treatment
standards.
This microdistillation system (Wadsworth MicroVOC^) addresses the shortcomings of
direct sample injection, purge-and-trap and other azeotropic distillation systems. Small
sample aliquots are required (40 ml). Analyte concentration factors are about two orders
of magnitude when a 40 ml sample aliquot is used. The total distillation time is five
minutes. Typical detection limits are between 5 and 15 (ig/1 when the distillate is
analyzed by gas chromatography with flame ionization detection. The cost of the
complete system is less than $300 with glassware comprising less than $70 of the total
cost.
REFERENCES
1) Report to EPA: Measurement of Polar. Water - soluble. Nonpurgeable VQCs in
Aqueous matrices by Azeotropic Distillation - Gas Chromatography / Mass Spectrometry,
Midwest Research Institute, September 30,1989.
2) Peters, Steam Distillation Apparatus for Concentration of Trace Water Soluble
Organics, Anal. Chem.. 52 (1), pp. 211 - 213,1980.
3) Bruce, M.L., Lee, R.P., Stephens, M.W., A Method for the Concentration and Analysis
of Trace Methanol in Water by Distillation and Gas Chromatography, Sixth Annual
Waste Testing and Quality Assurance Symposium Proceedings Vol n, p. 93,1990.
4) Test Method Equivalency Petitions. A Guidance Manual. EPA/530-SW-87-008,
OSWER Policy Directive, No. 9433.00-2, February 1987
5) Cramer, P.H., Wilner, J., Eichelberger, J.W., Azeotropic Distillation-A Continuing
Evaluation for the Determination of Polar, Water - Soluble Organics, Sixth Annual Waste
Testing and Quality Assurance Symposium, July, 1990.
11-16
-------
46 POLLUTION REDUCTION IN THE LABORATORY THROUGH
THE USE OF SMALLER INITIAL SAMPLE SIZE
by
Dante J. Bencivengo, Ph.D.
Bruce N. Colby, Ph.D.
Philip W. Ryan, Ph.D.
Pacific Analytical, Inc.
6349 Paseo Del Lago, #102
'Carlsbad, California 92009
Presented at
SEVENTH ANNUAL WASTE TESTING AND
QUALITY ASSURANCE SYMPOSIUM
July 8-12, 1991
Washington, DC
1-17
-------
1.0 INTRODUCTION
Since the first of EPA's wastewater analysis methods were
put forth in 1976, the environmentally sensitive chemical,
methylene chloride, has been the solvent specified for
extracting semivolatile organic compounds (BNA's) from
aqueous media. Present methods for determining BNA's in
aqueous samples use about 500 mL of methylene chloride
(MeCl2) for each 1 L sample exclusive of that required for
GPC. Of this, only one mL is retained for analysis; the
rest is either lost to the atmosphere during handling or
disposed of by a waste removal firm.
Because MeCl2 is on most of EPA's lists of undesirable
chemicals, reducing the amount required by EPA's own
analytical methods seems highly desirable. The first
attempt to do this was noted in Method 525, a method for
determining BNA's in drinking water where analyte removal
from the sample is accomplished by adsorption in a "SEP"
cartridge or disk.
Although SEP technology seems unlikely to be directly
applicable to complex samples such as those associated with
industrial discharges or those from test wells, one aspect
of Method 525 is important. This is the fact that the GC/MS
calibration curve is pushed downward from the typical
10 nG/uL low point to 0.1 nG/uL. The significance of this
is that it should be possible to use a smaller initial
sample size, on the order of 100 mL, yet retain the
existing 1 mL final volume and still be well within the
calibration range of the GC/MS equipment. This should make
it possible to reduce the amount of MeCl2 required by about
a factor of ten.
Further, by calibrating to the existing method's high
point, the effective dynamic range is increased by an order
of magnitude. This should result in fewer sample extracts
requiring dilution and reanalysis. Clearly this would save
on analytical costs. Perhaps less clear is the savings
which should result from being able to predict analytical
effort more accurately due to reducing the uncertainty in
the time and effort associated with reruns.
There are several minor benefits associated with using
100 mL water samples. These include reduced sampling and
shipping costs, smaller, i.e., less expensive glassware and
general improvements associated with improved space
utilization. Such improvements, while significant in a
conventional laboratory, are much more important in a
mobile/field laboratory environment.
1-18
-------
The purpose of the activity reported here was to evaluate
the possibility of using smaller initial sample size,
100 mL versus 1000 mL, for the analysis of semivolatile
organic compounds in aqueous waste samples. For the purpose
of these experiments, a 100 mL continuous liquid-liquid
extractor was designed and fabricated. It was then used to
prepare seven spiked clean water samples for method
detection limit assessment and five spiked field samples
(provided by ICF Technology Inc., Las Vegas) for analyte
recovery assessment. The quantities of solvent used in
preparing the samples were recorded for comparison to those
used with the standard 1000 mL extractor design.
2.0 EXPERIMENTAL
CONTINUOUS LIQUID-LIQUID EXTRACTOR DESIGNS
A set of 100 mL continuous liquid-liquid extractors (CLLE)
were fabricated according to the design shown in Figure
2.1. To use them, 25 mL of methylene chloride (MeCl2) is
added to the CLLE and Another 25 mL, to a 50 mL round
bottom flask (RB) which is attached and used as the
collector. A 100 mL aliquot of water sample is then added
to the CLLE. Spikes are added to the water at this point.
The water is then acidified with 2 mL of six N sulfuric
acid, the condenser is placed on top of the extractor and
the solvent in the RB is heated to boiling for 18 to 24
hours.
Figure 2.1 - 100 mL Continuous Liquid-Liquid Extractor.
11-19
-------
GC/MS CONDITIONS
The instrument used to collect the data in this study was a
VG Trio-1. This is a current generation instrument which
provides better sensitivity than older instruments.
Acquisition parameters were equivalent to those specified
in the CLP 10/89 Low Concentration Water Method for
Semivolatiles. The instrument was calibrated for each
target analyte using a 5-point linear (regression)
calibration curve. Concentrations of the calibration
solutions ranged from one to 100 ng/uL; 1 uL injections
were used in all cases.
A method detection limit (MDL) study was carried out using
seven clean water samples spiked to 30 ug/L with each
target analyte with CLLE design A and 10 ug/L with CLLE
design B. MDLs were calculated from data acquired on both
instruments. Percent recoveries were also calculated from
the calculated concentration values for the five spiked
field samples.
3.0 RESULTS AND DISCUSSION
SOLVENT USAGE
Solvent usage, identical for both CLLE designs, is
summarized in Table 3.1. The quantities of methylene
chloride (MeCl2) were recorded for both the CLLE and RB
charge volumes and the glassware washing volumes. The
washing volumes were included because they are of
significant magnitude and because the washing cycle is a
true part of the analysis. For comparison purposes, the
solvent volumes use with 1000 mL CLLEs is also included in
Table 3.1. Also included in the table is the volume of
solvent used to clean up the K-D apparatus.
TABLE 3.1 - METHYLENE CHLORIDE USAGE
Usaqe 100 mL CLLE 1000 mL CLLE
CLLE Charge
RB Charge
CLLE Cleanup
RB Cleanup
K-D Cleanup
25 mL
25
75
30
100
Total 255 mL
250
250
225
150
100
975 mL
As can be seen, the charge volumes used win the CLLEs and
the RB collector are directly proportional to the volumes
1-20
-------
of sample extracted. Thus, by going from 1000 mL to 100 mL,
there is a savings of 90 percent in MeCl2 usage.
The wash volumes for the CLLE and RB however, are not
proportional to sample volume. This is because the washing
process must clean the surface area of the glass and the
ratio between the two devices in terms of surface area is
on the order of 0.3-to-l. This correlates well with the
MeCl2 usage of 0.28-to-l.
It is important to note that both CLLE sizes result in an
extract which must be concentrated via Kuderna-Danish (K-
D) . The wash volume for the K-D is remains unchanged with
sample size because the K-D apparatus, in particular the
Snyder column, is the same size in both cases. In principle
it should be possible to decrease the size of the Snyder
column but in practice this may be difficult due to
fabrication difficulties.
The overall reduction in MeCl2 achieved by going to 100 mL
initial sample size and 100 mL CLLEs is 82.2 percent
excluding the K-D. It drops to a savings of 73.8 percent
when the K-D is included.
CONTINUOUS LIQUID-LIQUID EXTRACTOR DESIGN
Two significant behavior characteristics were noted during
the extraction process. First there was a tendency for the
RB to go dry. This results in target analytes being
volatilized and driven up into the solvent vapor return arm
where they condense. When dry RBs were noted, the extracts
were discarded and an additional sample aliquot prepared,
this time with less heat applied to the RB. The second
problem resulted when insufficient heat was applied to the
RB to maintain solvent condensation in the condenser. When
this happened, solvent condensed in the solvent return arm
of the CLLE and the sample was not extracted. This second
problem was harder to monitor than the first because it
was dependant on laboratory temperature. This tended to
change due to the day/night settings on the thermostat
which result in the lab becoming rather cool on cold winter
nights. When the lab temperature drops, the volume of hot
solvent needed to keep solvent condensation taking place in
the condenser and not in the transfer arm becomes larger.
Both of the above problems could be overcome by using a 100
mL RB for the collector and charging it with 50, rather
than 25 roL of MeCl2. This however, would result in a
significant increase in solvent usage.
11-21
-------
INSTRUMENT CALIBRATION
The VG Trio-1 was readily calibrated across the one to
100 ng/uL range. Most of the polar compounds were detected
in the 1 ng/uL injection. Exceptions were Benzoic acid,
Hexachlorocyclopentadiene, 2,4-Dinitrophenol, 4,6-Dinitro-
2-methylphenol and Pentachlorophenol.
METHOD DETECTION LIMITS AND SPIKE RECOVERIES
Method detection limits (MDLs) were calculated for each
analyte based on the seven spiked clean water samples. The
MDLs were not as low as had been anticipated when
calculated from data acquired from all seven runs. They
ranged from about 20 to 30 ug/L for most analytes. Problems
were most significant with the highly polar analytes and
the reactive analytes. These included Benzoic acid,
Hexachlorocyclopentadiene, 2,4-Dinitrophenol and 4-
Chloroanaline.
TABLE 3.2 - METHOD DETECTION LIMITS (ug/L)
7 Sample Data 4 Sample Data
Phenol
bis ( -2-Chloroethyl ) Ether
2-Chlorophenol
Benzyl Alcohol
2-Methylphenol
bis (2-Chloroisopropyl ) Ether
4 -Methylphenol
N-Nitroso-Di-n-Propylamine
Hexachloroethane
Nitrobenzene
Isophorone
2-Nitrophenol
2 , 4-Dimethylphenol
Benzoic Acid
bis (2-Chloroethoxy) Methane
2 , 4-Dichlorophenol
1,2, 4-Trichlorobenzene
Naphthalene
4-Chloroaniline
Hexachlorobutadiene
4-Chloro-3 -Methylphenol
2-Methylnaphthalene
Hexachlorocyclopentadiene
2,4, 6-Trichlorophenol
2,4, 5-Trichlorophenol
2-Chloronaphthalene
2-Nitroaniline
39.1
29.6
29.9
31.5
29.3
23.0
27.4
16.3
29.3
26.3
23.5
21.8
10.8
nd
35.4
23.1
26.7
26.4
nd
27.8
24.8
23.9
nd
18.6
20.0
22.9
17.6
2.8
2.2
3.0
3.6
4.8
7.2
5.6
17.4
4.6
2.6
1.4
2.0
8.7
nd
3.7
5.4
6.6
2.9
nd
10.0
3.4
3.5
nd
8.7
2.4
2.1
2.7
1-22
-------
Dimethyl Phthalate 29.3 5.6
Acenaphthylene 29.6 3.3
2,6-Dinitrotoluene 16.9 4.8
3-Nitroaniline 16.1 10.8
Acenaphthene 30.0 2.6
2,4-Dinitrophenol nd nd
4-Nitrophenol 16.8 9.4
Dibenzofuran 22.8 4.6
2,4-Dinitrotoluene 15.2 6.0
Diethylphthalate 28.3 6.4
4-Chlorophenyl-phenylether 29.0 14.3
Fluorene 29.6 4.3
4-Nitroaniline 18.7 17.8
4,6-Dinitro-2-Methylphenol 8.5 11.4
N-Nitrosodiphenylamine 32.8 25.4
4-Bromophenyl-phenylether 27.3 5.0
Hexachlorobenzene 15.0 6.6
Pentachlorophenol 8.9 5.6
Phenanthrene 27.8 4.5
Anthracene 25.1 4.7
Di-n-Butylphthalate 32.9 29.0
Fluoranthene 31.6 13.8
Pyrene 33.0 12.0
Butylbenzylphthalate 29.8 18.1
3,3'-Dichlorobenzidine 14.4 14.3
Benzo(a)anthracene 34.0 10.8
Chrysene 32.1 7.5
bis(2-Ethylhexyl)phthalate 37.2 31.4
Di-n-Octylphthalate 36.9 31.3
Benzo(b)fluoranthene 37.6 31.8
Benzo(k)fluoranthene 46.0 36.3
Benzo(a)pyrene 24.0 10.5
Indeno(l,2,3-cd)pyrene 28.8 15.1
Dibenz(a,h)anthracene 29.2 18.2
Benzo(g,h,i)perylene 25.1 8.1
1,3-Dichlorobenzene 25.9 5.9
1,4-Dichlorobenzene 26.7 4.1
1,2-Dichlorobenzene 27 .1 4. 0
avg 26.0 9.4
The higher than anticipated MDLs are believed to result
from the operational problems with the extractors (see
above). The reason for this conclusion is that the
recoveries for the target analytes in three of the seven
samples was quite low (less than 40 percent). Further, the
recoveries were particularly low for the more polar
compounds. This suggests that there may have been problems
with incomplete extraction caused by the low night time lab
temperature.
11-23
-------
If it is assumed that the three low recovery extracts are
the result of a circumventible problem and MDLs are
recalculated based on four, rather than seven samples, the
MDL values are much closer to those anticipated (4 Sample
data in Table 3.2) with an average of 9.4 ug/L.
Results from the five field samples spiked in duplicate
were used to calculate mean percent recoveries (Table 3.3).
The percent recoveries averaged between 65 and 80 percent
which seems reasonable.
TABLE 3.3 - PERCENT RECOVERIES
Analvte Ava %R
Phenol 47.8
bis(-2-Chloroethyl)Ether 71.2
2-Chlorophenol 54.0
Benzyl Alcohol 83.5
2-Methylphenol 52.5
bis(2-Chloroisopropyl)Ether 73.1
4-Methylphenol 51.1
N-Nitroso-Di-n-Propylamine 73.3
Hexachloroethane 72.0
Nitrobenzene 69.8
Isophorone 81.3
2-Nitrophenol 55.2
2,4-Dimethylphenol 45.8
Benzoic Acid 54.6
bis(2-Chloroethoxy)Methane 63.5
2,4-Dichlorophenol 56.1
1,2,4-Trichlorobenzene 78.0
Naphthalene 73.7
4-Chloroaniline nd
Hexachlorobutadiene 73.9
4-Chloro-3-Methylphenol 62.7
2-Methylnaphthalene 74.1
Hexachlorocyclopentadiene nd
2,4,6-Trichlorophenol 60.1
2,4,5-Trichlorophenol 59.8
2-Chloronaphthalene 76.1
2-Nitroaniline 92.4
Dimethyl Phthalate 83.6
Acenaphthy1ene 77.7
2,6-Dinitrotoluene 85.7
3-Nitroaniline 51.7
Acenaphthene 73.6
2,4-Dinitrophenol 79.7
4-Nitrophenol 70.3
Dibenzofuran 79.7
2,4-Dinitrotoluene 86.4
11-24
-------
Diethylphthalate
4-Chlorophenyl-phenylether
Fluorene
4-Nitroaniline
4,6-Dinitro-2-Methylphenol
N-Nitrosodiphenylamine
4-Bromophenyl-phenylether
Hexachlorobenzene
Pentachlorophenol
Phenanthrene
Anthracene
Di-n-Butylphthalate
Fluoranthene
Pyrene
Butylbenzylphthalate
3,3'-Dichlorobenz idine
Benzo(a)anthracene
Chrysene
bis(2-Ethylhexyl)phthalate
Di-n-Octylphthalate
Benzo(b)fluoranthene
Benzo(k)fluoranthene
Benzo(a)pyrene
Indeno(1,2,3-cd)pyrene
Dibenz(a,h)anthracene
Benzo(g,h,i)perylene
1,3-Dichlorobenzene
1,4-Dichlorobenzene
1,2-Dichlorobenzene
Nitrobenzene-d5
2-Fluorobiphenyl
Terphenyl-dl4
Phenol-d5
2-Fluorophenol
2,4,6-Tribromophenol
2-Chlorophenol-d4
1,2-Dichlorobenzene-d4
Avg
4.0 CONCLUSIONS
The experiments described are promising in terms of
reducing pollution associated with environmental sample
preparation. The technique of using a smaller initial
sample size in conjunction with calibrating the GC/MS to a
lower concentration provided a reduction of approximately
75 percent in the amount of Methylene chloride required to
perform an extraction. The quality of the analytical
results was approximately equivalent to that achieved with
11-25
-------
the traditional 1 L sample and extractor but a new design
is recommended to increase the ruggedness of the ruggedness
of the experimental design.
1-26
-------
47 EXTRACTION OF PHENOLIC COMPOUNDS FROM WATER SAMPLES
USING STYRENE-DIVINYLBENZENE SPE DISKS
Craig G. Markell. Research Specialist, New Products Department,
Donald F. Hagen, Corporate Scientist, Corporate Research Analytical,
3M, 3M Center Bldg. 201-1S-26, St. Paul, Minnesota 55144
ABSTRACT
Phenolic compounds, especially the more polar ones, can be difficult
to extract from water samples using solid phase extraction with CIS
functional silica as the paniculate. The cause of low recoveries is
almost certainly unfavorable partitioning between the CIS and
water, resulting in rather low breakthrough volumes and recoveries.
Our research has shown that the use of pH adjustment and heavy
salting, along with low sample volumes, can help the situation by
altering the partitioning, but another solution is a different solid
phase particulate.
Styrene-divinlybenzene particles were incorporated into 47 mm
solid phase extraction disks and used to extract a variety of phenolic
compounds from water samples. To preserve the high flow rates
that make solid phase extraction disks so attractive, small particles
(3-10 um) were used to preserve the fast kinetics seen with the
usual 8 um CIS silica.
This presentation will briefly discuss the basics of extracting phenols
from water using SPE disks, followed by the details and results of our
research. The preliminary conclusion is that the resin disks do have
some advantages over CIS disks for the extraction of phenols, and
perhaps other polar compounds, from water samples.
INTRODUCTION
One of the more significant trends in environmental sample
preparation is the replacement of liquid/liquid extraction (LLE) with
liquid/solid extraction (LSE), also called solid phase extraction (SPE),
for concentrating semi- and non-volatiles from aqueous samples.
Although LSE works very well for extracting most analytes of
environmental significance, low recoveries are expected for the more
11-27
-------
polar, water-soluble analytes, such as certain phenols. This is
expected, since LSE and LLE use very similar mechanisms and low
recoveries of water-soluble compounds are well known in LLE.
Partition ratios between the organic phase and the aqueous phase
govern the percentage of analyte extracted in both LLE and LSE,
where it is convenient to think of the organic portion of the particle
as being analogous to the solvent in LLE. For hydrophobic, water-
insoluble compounds, such as PAH's, PCB's, and many other
pollutants, the partition ratio is overwhelmingly in favor of the
organic phase, resulting in good recoveries from large volumes of
water. On the other hand, polar and water-soluble compounds have
less favorable partition ratios, resulting in relatively low recoveries.
Often, the addition of additives to the aqueous phase is effective in
changing the partition ratio, thus increasing low recoveries. This
practice is well known in LLE in the form of sodium chloride
addition, or "salting out." Another familiar example is pH adjustment
to convert ionic analytes to the corresponding neutral species.
Salting out or buffering can be equally effective in LSE and is one
approach to increasing LSE recoveries of difficult compounds.
Other methods of increasing low recoveries in LLE are to use larger
volumes of extracting solvents, different extracting solvents, or
extractions of the same water sample with several portions of
organic solvent. Approximate LSE analogs of these techniques are
respectively a higher mass of sorbent, a sorbent with more
selectivity for the analytes, and multiple sorbent beds. An
alternative to a higher mass of sorbent is a smaller volume of
sample, which also increases the sorbent/sample ratio.
This paper/presentation will explore the use of experimental solid
phase extraction disks which are similar to the Empore disks used
in Method 525, but containing polystyrene/divinylbenzene (SDVB) in
place of the CIS silica. The premise of this work is that the SDVB
disks offer both a higher mass of sorbent (in terms of organic
content) and perhaps more selectivity for aromatic compounds than
CIS. These features are expected to result in a disk which will offer
significantly higher recoveries for polar compounds, such as
H-28
-------
phenolics, from water samples. The analytes used for this work are a
series of phenols ranging from relatively hydrophilic to hydrophobic.
EXPERIMENTAL
The experimental portion of this work consisted of two parts:
scouting to determine the efficiency of several sorbents at extracting
phenolics from water, and more thorough triplicate extraction studies
of the most efficient sorbents identified by scouting. The scouting
was done by spiking several phenolic compounds into 100 ml of
reagent water and passing the water through experimental 47 mm x
0.5 mm SPE disks containing a variety of sorbents, using a standard
47 mm filtration apparatus. The use of the disks has been well
documented elsewhere and won't be detailed here (1,2). The disks
were then eluted using acetonitrile or acetonitrile followed by ethyl
acetate, depending on how tightly the analytes were sorbed to the
disk. The final determination was done with HPLC, using a reverse
phase system with UV detection. The compounds were: phenol, o-
cresol, 2-nitrophenol, 4,6-dinitro-o-cresol, 2,4-dichlorophenol, 2,4,5-
trichlorophenol, and 2,4,6-trichlorophenol at concentrations in the
water ranging from 0.5 ppm to 20 ppm, depending on the extinction
coefficient. The effect of salt addition and pH adjustment was also
briefly studied during this phase.
Recovery data were determined by spiking the 11 phenols shown in
Table II into 100, 200, 300, and 500 ml of water and passing the
water through 47 mm disks, as in the scouting studies. Samples
were processed using a vacuum of about 25 inches Hg, generated
with an aspirator. Sample flow times ranged from 0.5 minutes for
100 ml through the SDVB disk to 7 minutes for 500 ml through the
CIS disk. The 500 ml sample took 2.5 minutes to pass through the
SDVB disk, which would correspond to 5 min/L. Each spike was done
in triplicate. The approximate concentration of each phenol in the
water was 200 ug/L. Since the 11 phenols weren't well resolved by
the HPLC, the phenols were tested in two mixtures, one with five and
one with six of the compounds. Elution was done with 2 x 2 ml
aliquots of tetrahydrofuran (THF), followed by 2 x 2 ml aliquots of
methanol. These aliquots were then combined and made to 10 ml
with methanol for HPLC analysis. Again, CIS reverse phase HPLC
was used with 270 nm detection and a water:methanol gradient.
1-29
-------
Each of the mobile phases contained 0.1 percent acetic acid to
suppress ionization of the phenols and improve peak shape. Three
types of disks were used - CIS and cyclohexyl (CH) bonded silica, and
SDVB.
In both phases of the study, calibration was single point, the
standard being at the concentration expected from a 100 percent
recovery. Except for the standard 47 mm CIS and CH disks, which
are commercially available (Varian Sample Prep Systems, Harbor
City, CA), the Empore disks were experimental, each being
prepared at 3M from the specific sorbent particles mentioned.
Except for the SDVB particles used in the scouting phase, which were
in the 50-100 um size range, all of the particles were 5-15 um. All
disks were 47 mm x 0.5 mm with particle loadings of 75-90 percent
by weight.
RESULTS AND DISCUSSION
The intent of the scouting work was to quickly test several sorbents
for their ability to extract the probe phenolics from 100 'ml of water,
then use the more promising phases for further study. Standard CIS
disks were used as a control, since these disks are beginning to find
wide use in environmental laboratories. CH bonded silica was also
incorporated into scouting and the subsequent recovery studies,
since CH has gained a reputation of being effective for phenol
extractions. Besides the CIS bonded silica and CH bonded silica, two
proprietary bonded silicas were tried, plus a cyano bonded silica and
the SDVB. The phases showing the best recoveries were CIS, CH, and
SDVB. Although CH is often mentioned as an effective phase for the
SPE of phenolics from water, the results failed to show a clear
advantage over CIS. These results are shown in Table I. The
compounds that presented problems with the extraction were the
more polar, water soluble compounds, while the hydrophobic phenols
were easily extracted from 100 ml samples by most of the sorbents
tried.
As an extension of this work, 25% NaCl was added to the water and
the pH was lowered to 2 with HC1. These modifications of the
sample, done before extraction, were successful in raising the
recoveries of several analytes, also shown in Table I. Sample
11-30
-------
modification steps, which added little time or cost to the analysis,
make it possible to use standard CIS disks for the quantitative
extraction of many phenolics of environmental interest. The only
test compound showing a low recovery was phenol, which has a
water solubility of almost 10 g/100 ml in water.
In the scouting work, the SDVB resin was clearly the most effective
for quantitative extractions of the test compounds from 100 ml of
water, even without modification of the sample. Because of the
strong interaction between some of the analytes and the resin,
acetonitrile elution alone wasn't strong enough, as evidenced by
generally low recoveries (not shown) of even the hyrophobic
phenols. To overcome this problem, the usual acetonitrile elution
was followed by 2 x 1 ml ethyl acetate elutions, which were added to
the acetonitrile. There are undoubtedly a number of alternative
elution solvents which would have been equally effective.
Once the scouting work had identified SDVB as an effective sorbent
for phenolic compound extractions, a more rigorous recovery study
was undertaken to confirm the scouting results and progressively
increase sample volumes to define the limits of this technique. CIS
and CH disks were again included for comparison. The SDVB disks
used for these results contained sorbent particles approximately 10
um in size. For this study, the analyte list was modified to contain
the traditional priority pollutant phenols, at approximately 200 ug/L
each. The reagent water used was unmodified in terms of pH or salt
content.
Recovery results are shown in Tables II, III, and IV, at several
sample volumes, for CIS, CH, and SDVB disks, respectively.
Generally, the results contain no surprises. In order of effectiveness
for phenol extractions from water, SDVBปC18>CH, which also
parallels the organic content of each sorbent particle. As expected,
increasing the volume of the samples decreased the recoveries of
marginally recovered compounds. There are a few anomalous
results, e.g. the 2,4-dinitrophenol results on the SDVB disk at 300 ml.
This may be a reflection of compounds with pKa's near the pH of the
matrix, where a slight change in pH would result in a substantial
change in the percentage of the ionic form of those compounds.
11-31
-------
Dropping the pH of the samples to 2 would overcome this effect and
may increase some of the low recoveries.
While resin sorbents have occasionally been reported in the
literature for phenolic extractions, including one of the pioneering
publications in SPE, heroic efforts are often needed to clean up these
resins before use (3). Although our work was conducted well above
method detection limits, we saw no evidence of interferences from
the SDVB, which would be expected to contain UV chromophores.
The only cleanup step used for these disks was the initial wash step
with a few ml of the eluting solvents (1). Given the small particle
diameter and short distances needed for contaminants to diffuse into
the wash solvent, the initial wash step plus the methanol
conditioning step may be sufficient to remove any contaminants. An
independent researcher, using these disks prior to LC/MS, also
noticed no interferences (4).
Conclusions
This work demonstrates the utility of experimental SDVB, SPE disks
as a technique for isolating phenolics from water. Even at 500 ml,
quantitative recoveries were seen for all but a few phenols with
extraction times corresponding to about 5 min/L. With pH
adjustment and salting out, the low recoveries may have been
improved. Using 10 um SDVB particles in the disks, no interference
problems were encountered, in contrast to literature reports of
extensive soxhlet extractions needed for the much larger SDVB
particles used in previous research.
1-32
-------
References
[1] Hagen, D. R, Markell, C. G., Schmitt, G. A., and Blevins, D. D.,
Analytica Chimica Acta. 236, 157-164, (1990).
[2] Markell, C. G., Hagen, D. R, and Bunnelle, V. A., LC-GC. 9, 332-
337, (1990).
[3] Junk, G. A. et al., Journal of Chromatography. 99, 745-762,
(1974).
[4] Behymer, T. D., personal communication (1991).
1-33
-------
TABLE I. Scouting Results for Selected Phenols
Sorbent
Phenol
o-Cresol
2-Nitrophenol
2-Methyl-4,6-Dinitrophenol
2,4-Dichlorophenol
2,4,5-Trichlorophenol
2,4,6-Trichlorophenol
CIS
4
20
35
14
106
108
110
CH
5
9
23
19
102
109
1 11
SDVB*
90
120
108
96
107
95
96
C18**
23
94
90
94
92
92
92
CH**
14
99
69
94
88
87
89
* eluted with acetonitrile followed by ethyl acetate
u* pH = 2, 25% NaCl
-------
CO
01
TABLE II. Results Using CIS Disks - % Recovery (RSD, n=3)
Volume (ml)
100 200 3ฃQ 500
Phenol 12.8 (9.5) 7.8 (4.1) 4.5 (6.7) 2.8 (7.3)
2-Nitrophenol 63.9 (3.2) 41.2 (3.2) 25.0 (5.9) 15.4 (5.2)
4-Nitrophenol 29.9 (6.1) 19.1 (6.2) 11.3 (3.8) 6.8 (5.8)
2-Chlorophenol 50.6 (3.9) 30.3 (4.3) 18.2 (5.0) 10.7 (5.2)
2,4-Dinitrophenol 5.9 (17.3) 4.2 (15.2) 8.8 (1.6) 1.6 (5.3)
2,4-Dichlorophenol 67.4 (5.1) 82.9 (10.9) 63.8 (1.1) 47.5 (4.8)
2,4-Dimethylphenol 99.2 (3.0) 89.5 (4.7) 61.8 (7.2) 37.1 (6.7)
4-Chloro-3-Methylphenol 93.0 (7.6). 104.9 (3.9) 89.7 (7.4) 60.5 (7.9)
2,4,6-Trichlorophenol 92.7 (2.5) 111.3 (1.7) 99.7 (2.7) 97.0 (1.4)
2-Methyl-4,6-Dinitrophenol 31.0 (6.6) 31.4 (3.0) 18.6 (4.0) 11.7 (4.3)
Pentachlorophenol 95.1 (2.0) 98.8 (3.8) 102.9 (1.5) 98.2 (1.8)
-------
TABLE III. Results Using CH Disks - % Recovery (RSD, n=3)
Volume (ml)
100 200 300 500
Phenol 12.2 (8.3) 6.9 (14.9) 2.9 (16.0) 2.0 (16.6)
2-Nitrophenol 41.1 (2.5) 20.2 (6.1) 15.0 (10.5) 8.9 (8.2)
4-Nitrophenol 28.8 (1.4) 16.5 (5.7) 7.9 (7.6) 5.2 (6.1)
2-Chlorophenol 41.6 (3.2) 21.4 (4.6) 15.2 (8.2) 8.9 (6.9)
2,4-Dinitrophenol 2.7 (15.5) 1.9 (8.7) 3.6 (6.5) 0.8 (19.7)
2,4-Dichlorophenol 64.9 (12.9) 59.2 (10.1) 53.4 (6.9) 34.4 (12.6)
2,4-Dimethylphenol 91.6 (0.8) 70.9 (14.4) 33.6 (10.0) 21.6 (18.4)
4-Chloro-3-Methylphenol 89.9 (6.0) 96.5 (9.3) 55.9 (9.3) 35.7 (20.5)
2,4,6-Trichlorophenol 92.4 (4.9) 109.5 (3.1) 79.4 (11.2) 86.7 (9.3)
2-Methyl-4,6-Dinitrophenol 19.8 (18.8) 14.2 (6.6) 9.8 (9.1) 6.3 (11.0)
Pentachlorophenol 99.9 (2.4) 93.5 (2.2) 105.0 (2.0) 98.7 (1.1)
-------
CO
-vl
TABLE IV. Results Using SDVB Disks - % Recovery (RSD, n=3)
Volume (ml)
100 200 300 500
Phenol 50.6 (12.4) 26.8 (10.0) 16.7 (20,9) 9.2 (13.6)
2-Nitrophenol 90.8 (3.6) 101.6 (0.8) 95.9 (2.0) 93.0 (2.7)
4-Nitrophenol 88.2 (10.3) 84.3 (6.1) 53.7 (13.4) 33.1 (12.9)
2-Chlorophenol 89.7 (3.6) 107.7 (0.9) 91.8 (3.4) 71.6 (5.9)
2,4-Dinitrophenol 43.7 (28.2) 54.1 (15.8) 105.1 (12.3) 22.2 (11.6)
2,4-Dichlorophenol 91.3 (6.3) 102.4 (1.2) 94.1 (3.7) 91.8 (0.6)
2,4-Dimethylphenol 98.5 (4.7) 108.6 (3.1) 93.6 (8.4) 95.9 (1.3)
4-Chloro-3-Methylphenol 91.3 (12.3) 111.4 (3.3) 97.3 (14.0) 100.3 (2.9)
2,4,6-Trichlorophenol 90.9 (5.5) 115.7 (3.0) 100.5 (16.3) 97.8 (1.6)
2-Methyl-4,6-Dinitrophenol 64.0 (7.6) 96.9 (4.9) 100.0 (14.4) 85.1 (1.5)
Pentachlorophenol 95.3 (4.6) 100.2 (3.2) 105.0 (0.9) 98.2 (0.1)
-------
AQ Comparison of Alternative Methods for Analysis of Volatile Organic Contaminants
by
Thomas C. Voice, Associate Professor of Environmental Engineering,
Michigan State University, East Lansing, Ml 48824, and
James F. Ryan, Environmental Marketing Manager, The Perkin-Elmer Corporation,
Norwalk, CT 06859
The U.S. Environmental Protection Agency (EPA) has published methods for the quantitative analysis of
volatile organic contaminants (VOCs) in a variety of environmental sample matrices such as ground water,
industrial effluents, drinking water, sludge, soil, and so forth. These methods are largely based upon a
"purge-and-trap" methodology in which the volatile constituents are purged from the sample, collected on
an adsorbent trap directly connected to a gas chromatograph, and then thermally desorbed onto the GC
column for separation and quantitative analysis. These methods vary somewhat in the type of GC column
and detectors specified, but the purge-and-trap technique for collecting the organic constituents and in-
troducing them to the GC is essentially the same in all methods. While the purge-and-trap technique of-
fers the advantage of sub parts-per-million sensitivities for many compounds with specific GC detectors, it
suffers from several significant disadvantages, including lack of a universally applicable trap adsorbent
material, high sample-to-sample carryover, introduction of large quantities of water to the GC and detec-
tors, poor compatibility with capillary columns, and limitations to automating the overall technique.
This study describes an investigation of two alternatives to the common implementation of the EPA purge-
and-trap procedures: (1) automated static headspace analysis and (2) what might be termed "off-line"
purge-and-trap. Static headspace analysis involves equilibrating a sample with a fixed gas volume in a
closed vessel and subsequently introducing an aliquot of this gas directly into the GC. The entire process
can be automated using equipment available from a variety of manufacturers. Off-line purge-and-trap
involves purging samples using separate adsorbent traps for each sample independent of the GC system.
The traps are then thermally desorbed into the GC using automated equipment. Analyses of a series of
VOCs have been performed using these two alternative techniques, along with the traditional purge-and-
trap approach. Samples analyzed include both water and soil matrices.
In the figures presented below, we show that off-line purge-and-trap methodology compares very
favorably with the chromatographic and reproducibility data generated by an on-line methods. However,
the carryover is reduced by a factor of 10 with the off-line method because of the use of multiple traps.
On-line purge and trap
Off-line purge and trap
5 ml of sample containing 10 ug/L of
benzene, trichloroethane, toluene,
tetrachloroethylene, ethylenebenzene,
p-xylene, and o-xylene
0.53 mm x 50 m DB-624 column, PID detector
4 min @ 50ฐC, 8ฐ/min to 190 ฐ, 4 min hold
250 mg Tenax trap
4mindesorb@ 180ฐ
I-38
-------
Comparison of Sample Reproducibiltty and Sample Carryover.
%RSD Reproducibility % Carryover
Compound On-line Off-line On-line Off-line
benzene 2.42 2.24 0.10 0.01
trichloroethylene 5.29 1.41 0.16 0.01
toluene 1.10 2.61 0.22 0.04
tetrachloroethylene 4.59 1.37 0.28 0.02
ethylenebenzene 2.49 0.94 0.34 0.03
p-xylene 3.39 1.72 0.36 0.03
o-xylene 1,80 3.40 0.43 0.04
Data has also been gathered on automated headspace analysis of these compounds in water and in soil.
I-39
-------
49 EVALUATION OF SAMPLE PREPARATION METHODS FOR SOLID MATRICES
Viorica Lopez-Avila. J. Milanes, N. Dodhiwala, and J. Benedicto, Mid-Pacific Environmental
Laboratory, Mountain View, California 94043, and W. F. Beckert, U.S. Environmental
Protection Agency, EMSL-LV, Las Vegas, Nevada 89109.
ABSTRACT
Four sample preparation methods: Soxhlet extraction (Method 3540), Soxtec extraction
(Method 3541), sonication extraction (Method 3550), and supercritical fluid extraction (SFE)
with carbon dioxide (Method 3560) have been evaluated. Thirty target compounds
representing organochlorine pesticides, nitroaromatic compounds, haloethers, and
chlorinated hydrocarbons were spiked on wet and dry clay, topsoil, sand, and sand mixed
with organic compost and were extracted by Soxhlet, Soxtec, and sonication techniques using
hexane-acetone (1:1) and methylene chloride-acetone (1:1) and by SFE with carbon dioxide.
Data are also presented for 43 base/neutral/acidic compounds spiked on sand or clay and
extracted by SFE with carbon dioxide and by Soxtec extraction with hexane-acetone (1:1),
and for three standard reference materials extracted by Soxtec and SFE with carbon dioxide.
INTRODUCTION
There are currently two extraction methods listed in SW-846 (1) for the extraction of solid
matrices: Method 3540 (Soxhlet extraction) and Method 3550 (sonication extraction).
Method 3540 is generally applicable, and a large number of samples can be extracted side
by side with limited manpower requirements. However, Soxhlet extractions usually take
between 8 and 26 hours, require relatively large amounts of solvents, and involve extract
cleanup and concentration. Sonication extractions require much shorter extraction times,
but they are labor-intensive, use large amounts of solvent, and require extract cleanup and
concentration.
Two new techniques that have become available recently are Soxtec extraction (Method
3541) and supercritical fluid extraction (SFE) (Method 3560). Soxtec extraction is a
modified Soxhlet extraction: the thimble with the sample is first immersed in hot solvent,
then, after a boiling period of usually up to 1 hour, is raised physically and extracted a la
Soxhlet for another hour. The very limited results reported so far by others indicate that
Soxtec extraction is at least as exhaustive as Soxhlet extraction (2), but the extraction time
is reduced to about 2 hours, less solvent is needed, and the solvent is evaporated and
NOTICE: Although the research described in this paper has been supported by the U.S.
Environmental Protection Agency, it has not been subjected to Agency review and therefore
does not necessarily reflect the views of the Agency, and no official endorsement should be
inferred. Mention of trade names and commercial products does not constitute
endorsement or recommendation for use.
11-40
-------
condensed without requiring extract transfer. SFE uses a supercritical fluid as extraction
solvent in a special extraction system that is operated at pressures and temperatures higher
than the critical pressure and critical temperature of the particular fluid. The most
commonly used fluid is carbon dioxide; others that are being used, or have been
investigated, include nitrous oxide, sulfur hexafluoride, Freon-13, ammonia, xenon, and
several hydrocarbons. Carbon dioxide is so popular because of its low critical temperature
(31.3ฐC) and pressure (72.9 atm) and because it is non-toxic, non-flammable, relatively non-
reactive and inexpensive, and its use does not result in a waste disposal problem. It is a
rather non-polar solvent, similar to hexane or benzene, but both solvent strength and
selectivity can be improved by the addition of small amounts of modifiers such as acetone,
methanol, or toluene.
The objective of this study was to evaluate the applicability of Soxtec extraction to samples
of interest to the EPA and to generate performance data for these four extraction methods
for solids. We focused on 30 analytes covering the following groups of compounds
environmentally significant to EPA: organochlorine pesticides, chlorinated hydrocarbons,
nitroaromatics, and haloethers. To a limited extent, we also generated data for 43
base/neutral/acidic compounds currently on the Hazardous Substances List. The matrices
evaluated included sand, clay, topsoil, sand mixed with organic compost, and standard
reference materials certified for a limited number of organic compounds (mostly polynuclear
aromatic hydrocarbons).
EXPERIMENTAL
Apparatus
Soxhlet extractor - 40 mm ID with 500-mL round bottom flask, condenser and
heating mantle
Sonication system - Horn-type sonicator equipped with titanium tip (Heat Systems
Ultrasonics Inc., Farmingdale, New York, Model W-375)
Soxtec HT-6 extraction system with controlled heated oil bath (Tecator, Inc.,
Herndon, Virginia)
Kuderna-Danish apparatus with 10-mL concentrator tube, 500-mL evaporation flask,
three-ball macro Snyder column
Supercritical fluid extractor - Suprex Model SE-50 including a 4-port and a 12-port
valve configured with electronic actuators for automated operation. The system was
set up either with two or four extraction vessels for parallel extractions. The 3-mL
extraction vessels (1 cm ID x 4 cm length) were obtained from Suprex Corporation
(Pittsburgh, Pennsylvania), the 2-mL extraction vessels (0.9 cm ID x 3 cm length)
from Alltech Associates (Deerfield, Illinois). Supercritical pressures were
maintained inside the extraction vessels by using 60 cm of uncoated fused-silica
tubing (50 urn ID x 375 urn OD) from J&W Scientific (Folsom, California) as
restrictor. Collection of the extracted material was performed by inserting the
11-41
-------
outlet restrictor into a 15-mm x 60-mm glass vial (Supelco Inc., Bellefonte,
Pennsylvania) containing 5 mL hexane.
Gas chromatograph ซ A Varian 6000 equipped with two constant-current/pulsed-
frequency electron capture detectors and two megabore fused-silica open-tubular
columns (30-m x 0.53-mm ID x 0.83-/^n film thickness DB-5 column and 30-m x
0.53-mm ID x 1.0-f/m film thickness DB-1701 column), connected to a press-fit Y-
shaped glass splitter (J&W Scientific Inc., Folsom, California) was used to analyze
for the 30 target analytes. The columns were temperature-programmed from 100ฐ C
(2-min hold) to 275ฐ C (6-min hold) at 5ฐC/min; injector temperature 250ฐ C;
detector temperature 320CC; helium carrier gas 6 mL/min; nitrogen makeup gas
20 mL/min.
Gas chromatograph/mass spectrometer A Finnigan 4510B (Finnigan MAT, San
Jose, California) interfaced with a data system for data acquisition and processing
and equipped with a 30-m x 0.32-mm ID DB-5 fused-silica open-tubular column (1-
|im film thickness) was used for all PAH and base/neutral/acidic compound
analyses. The column was temperature-programmed from 40ฐ C (4-min hold) to
300ฐ C (6-min hold) at 8ฐC/min; injector temperature 270ฐ C; interface temperature
270ฐ C.
Materials
Standards Analytical reference standards of the organochlorine pesticides,
chlorinated hydrocarbons, nitroaromatics, haloethers, PAHs and base/neutral/acidic
compounds were obtained from the U. S. Environmental Protection Agency,
Pesticides and Industrial Chemicals Repository (Research Triangle Park, North
Carolina), Aldrich Chemical (Milwaukee, Wisconsin), Ultrascientific Inc. (Hope,
Rhode Island), and Chem Service (West Chester, Pennsylvania). All compounds,
except the PAHs and the base/neutral/acidic compounds, were obtained as neat
materials. Their purities were stated to be greater than 98 percent. Stock solutions
of each test compound were prepared in pesticide-grade hexane at 1 mg/L.
Working calibration standards were prepared by serial dilution of a composite stock
solution prepared from the individual stock solutions. The PAHs and the
base/neutral/acidic compounds were obtained as composite mixtures in methylene
chloride or methylene chloride/toluene.
SFC-grade carbon dioxide (Scott Specialty Gases, Plumsteadville, Pennsylvania)
Hexane, acetone, methylene chloride ~ nanograde or pesticide-grade
Sample matrices: sand, clay, topsoil, sand mixed with 10 percent organic compost,
marine sediments HS-3 and HS-4 (National Research Council of Canada, Halifax,
Nova Scotia, Canada), PAH-contaminated soil SRS 103-100 (Fisher Scientific,
Pittsburgh, Pennsylvania). The sand and the standard reference materials were dry.
The clay, topsoil, and the sand/compost matrices contained 10.6, 2.6, and
4.2 percent moisture.
1-42
-------
Procedures
Spiked samples (10 g each) of sand, sand with 10 percent organic compost, clay, and topsoil
were extracted with hexane-acetone (1:1) or methylene chloride-acetone (1:1) following the
procedures given in Methods 3540 and 3550. Spiking of the samples (that were extracted
by SFE, Soxhlet, and Soxtec) with the 30 target compounds or the base/neutral/acidic
compounds was performed as follows: the sample was weighed out in an aluminum cup and
a concentrated stock solution (100 to 1000 jiL) containing the target compounds in hexane
or methylene chloride and methylene chloride/toluene was added to the sample with a
syringe while making sure that the solution did not contact the aluminum cup. Mixing was
performed with the tip of a disposable pipette. After the solvent had evaporated completely
(approximately 15 min), the spiked sample was transferred to the extraction vessel. Spiking
of samples that were extracted by sonication was performed directly into the amber bottle
used for extraction.
Soxtec extractions were performed with 10-g samples and 50 mL solvent using an immersion
time of 45 or 60 min and an extraction time of 45 or 60 min as indicated in the tables.
SFEs were performed as specified in the Results Section. All SFEs were carried out using
the Suprex SE-50 system.
RESULTS AND DISCUSSION
Table 1 presents the average recoveries of the 30 target compounds spiked on sand with
10 percent organic compost and on the clay matrix and extracted by sonication and Soxhlet
extraction with hexane-acetone (1:1). The results from the Soxtec extraction are presented
in Table 2 and the SFE data in Table 3.
The following conclusions can be drawn from these data:
The repeatability of the sonication extraction with hexane-acetone (1:1) is much
better than that of Soxhlet extraction. The percent RSDs for the 30 target
compounds for sonication ranged from 2.3 to 3.9 percent (except for one value at
14.7 percent) for the sand/compost matrix and 0.2 to 6.5 percent for the clay matrix.
The percent RSDs for the recoveries from the Soxhlet extraction ranged form 3.9
to 86.9 percent for the clay matrix, with most of the values above 20 percent.
The repeatability of the Soxtec technique is significantly better than that of the
Soxhlet technique. Only the more volatile compounds such as nitrobenzene,
benzotrichloride, 4-chloro-2-nitrotoluene, and the dichloronitrobenzenes exhibited
RSD values above 10 percent when the extraction was performed with either
hexane-acetone or methylene chloride-acetone. The percent RSDs for the other
compounds were below 10 percent. The average recoveries using the Soxtec
technique were significantly higher than those obtained by Soxhlet or sonication, and
similar or slightly higher than the SFE recoveries, for both the hexane-acetone and
the methylene chloride-acetone solvent combinations.
1-43
-------
SFE recoveries were comparable to those obtained by sonication (except for the clay
matrix where SFE recoveries were significantly higher than the recoveries via
sonication) and Soxhlet techniques, but the RSDs for the SFE values were quite
high. The data reported in Table 3 were obtained with our four-vessel setup;
therefore,the RSD values for the SFE data are actually those from the combined
results from four extractions carried out simultaneously. Work is in progress in our
laboratory to investigate the vessel-to-vessel variability.
The data for the base/neutral/acidic compounds are presented in Table 4 and 5 for the SFE
and Soxtec extraction, respectively. The recoveries by SFE and Soxtec extraction are
comparable (except for the very low SFE recoveries for compounds 4, 5, 7, 12, probably
because of their volatilities and for benzoic acid and 4-nitrophenol, probably because of their
low solubilities in supercritical carbon dioxide) and the percent RSDs follow the same
pattern as discussed above for the group of 30 compounds.
In the case of the three standard reference materials, we noticed significant differences in
recoveries obtained by Soxtec and by SFE. For the SRS 103-100 standard reference soil
(Table 6), the SFE naphthalene and acenaphthylene recoveries were only about 50 to 60
percent of those measured in the Soxtec extracts. This could be explained by the high
volatilities of the two compounds. However, the recoveries of the higher-molecular-weight
PAHs benzofluoranthenes and benzo(a)pyrene were 53 and 32 percent by SFE versus 118
and 80 percent by Soxtec. Additional extractions were performed with supercritical carbon
dioxide modified with 10 percent hexane, 1 percent toluene, or 15 percent propylene
carbonate to improve the extractabilities of the higher-molecular-weight PAHs. Only
propylene carbonate showed increased extractabilities for the compounds cited above. For
the HS-3 and HS-4 (Table 7), SFE recoveries were approximately around 20 percent when
the extraction was performed with carbon dioxide. Addition of modifiers increased the
recoveries somewhat. Furthermore, presence of elemental sulfur in these marine sediments
created restrictor plugging problems on four commercial extractors evaluated by us as part
of another study.
In conclusion, sonication and Soxtec extraction of environmental samples with hexane-
acetone (1:1) give comparable results in terms of method precision and accuracy and are
fast. However, they both require large amounts of solvents, and the extracts need to be
subjected to gel permeation chromatography or some type of column chromatography (e.g.,
alumina, silica) especially if an electron capture detector will be used for analysis. SFE, on
the other hand appears to be much faster and more selective. However, the technique is
matrix-dependent, and although we have shown that many compounds of interest to EPA
can be extracted from spiked sand, more developmental work is required before SFE can
be used routinely with environmental matrices.
REFERENCES
1. Test Methods for Evaluating Solid Waste (1986), 3rd Ed., SW-846, U.S. Environmental
Protection Agency, Washington, DC.
1-44
-------
2. J.H. Stewart, Jr., C.K. Bayne, R.L. Holmes, W.F. Rogers and M.P. Maskarinek.
"Evaluation of a Rapid Quantitative Extraction System for Determining the
Concentration of PCBs in Soils." Proceedings of U.S. Environmental Protection Agency
Symposium on Waste Testing and Quality Assurance, July 11-15,1988, Washington, DC.
11-45
-------
TABLE 1. AVERAGE PERCENT RECOVERIES AND PERCENT RSDs FOR 30 TARGET COMPOUNDS EXTRACTED FROM
SPIKED SAND/COMPOST AND CLAY SAMPLES BY SONICATION AND SOXHLET EXTRACTION WITH HEXANE-
ACETONE (1:1)
Sonication8
Soxhletb
Sand/compost with
Compound
no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Compound name
13-Dichlorobenzene
1,2-Dichlorobenzene
Nitrobenzene
Benzal chloride
Benzotrichloride
4-Chloro-2-nitrotoluene
Hexachlorocyclopentadiene
2,4-Dichloronitrobenzene
3,4-Dichloronitrobenzene
Pentachlorobenzene
23>4,5-Tetrachloronitrobenzene
Benefin
alpha-BHC
Hexachlorobenzene
delta-BHC
Heptachlor
Aldrin
Isopropalin
Heptachlor epoxide
trans-Chlordane
Endosulfan I
Dieldrin
2,5-Dichlorophenyl-4'-nitrophenyl ether
Endrin
Endosulfan II
2,4,6-Trichlorophenyl-4'-nitrophenyl ether
P|P:-DDT
23,6-Trichlorophenyl-4'-nitrophenyl ether
23,4-Trichlorophenyl-4'-nitrophenyl ether
Mirex
20 percent
Average
recovery
0
0
0
63.3
0
933
0
923
0
82.4
84.8
99.6
90.0
883
88.6
90.9
84.6
95.0
90.5
89.9
91.6
91.8
86.2
95.3
86.7
47.3
84.2
82.7
79.5
84.2
moisture
Percent
RSD
_
_
14.7
ซ
3.5
_
2.9
2.7
3.4
2.6
23
2.9
3.1
2.8
2.6
3.7
3.7
3.0
3.4
3.7
3.0
3.6
3.7
3.1
3.0
3.1
3.4
3.9
Clay with
20 percent
Average
recovery
0
2
0
0
0
34.1
0
37.0
34.0
30.1
35.5
35.4
45.0
34.4
47.6
40.7
42.1
38.0
46.1
44.7
453
48.9
44.7
44.9
47.4
23.5
44.7
47.1
44.1
513
moisture
Percent
RSD
_
_
_
_
6.5
2.0
5.2
7.1
2.5
5.8
3.4
5.5
0.2
3.7
43
4.8
1.5
1.4
1.0
1.0
3.2
1.1
03
3.1
1.8
3.1
4.6
3.0
Sand/compost with
20 percent
Rep. 1
0
0
0
30.7
29.6
46.0
0
40.4
34.4
55.0
32.0
493
67.5
58.8
78.1
65.4
72.1
61.1
77.7
75.4
73.6
79.4
51.8
86.0
74.6
NSC
69.4
35.6
46.2
74.7
moisture
Rep. 2
0
0
0
30.7
29.2
443
_
43.8
40.2
52.5
42.8
60.7
78.8
69.7
84.8
76.7
79.4
79.4
83.1
82.4
81.0
82.0
75.5
88.5
80.9
NS
82.2
71.1
67.6
79.8
Clay with
20 percent
Average
recovery
0
0
0
0
9.8
34.0
0
32.2
24.8
53.2
27.1
47.8
57.4
55.4
65.0
59.6
69.8
64.2
72.0
75.6
76.4
74.4
65.9
81.0
78.5
NS
73.6
64.4
62.5
75.5
moisture
Percent
RSD
_
_
_
86.9
44.4
_
57.8
44.0
16.6
38.4
11.5
47.5
24.5
27.1
34.1
8.8
20.8
20.8
12.5
5.5
20.0
26.9
3.9
6.7
NS
38.5
34.4
29.2
15.0
a Number of determinations was three. Spiking level was 500 ng/g, except compounds 23, 28, and 29 at 1500 ng/g, compound 26 at 3000 ng/g, compound 3 at 2000
ng/g, and compounds 1 and 2 at 5000 ng/g.
" Number of determinations was three except for sand/compost matrix where only two determinations were performed. Spiking level was the same as for the
sonication experiments. Extraction time was 16 hours.
c NS - not spiked.
-------
TABLE 2. AVERAGE PERCENT RECOVERIES AND PERCENT RSDs FOR THE 30
TARGET COMPOUNDS FROM SPIKED CLAY SAMPLES BY SOXTEC
EXTRACTION WITH HEXANE-ACETONE (1:1) AND METHYLENE
CHLORIDE-ACETONE (!:!)
Hexane-acetone
Methylene chloride-acetone
Compound
no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Compound
name
13-Dichlorobenzene
1,2-Dichlorobenzene
Nitrobenzene
Benzal chloride
Benzotrichloride
4-Chloro-2-nitrotoluene
Hexachlorocyclopentadiene
2,4-Dichloronitrobenzene
3,4-Dichloronitrobenzene
Pentachlorobenzene
23,4,5-Tetrachloronitrobenzene
Benefin
alpha-BHC
Hexachlorobenzene
delta-BHC
Heptachlor
Aldrin
Isopropalin
Heptachlor epoxide
trans-Chlordane
Endosulfan I
Dieldrin
2,5-Dichlorophenyl-4'-
nitrophenylether
Endrin
Endosulfan II
2,4,6-Trichlorophenyl-4'-
nitrophenylether
p,p'-DDT
23,6-Trichlorophenyl-4'-
nitrophenylether
23,4-Trichlorophenyl-4'-
nitrophenylether
Mirex
Average
recovery
0
0
77.1
383
33.4
92.8
46.0
115
783
48.6
122
82.0
94.9
81.7
104
87.1
78.2
97.5
92.4
85.8
90.5
68.8
99.7
112
903
127
61.4
97.2
91.6
84.0
Percent
RSD
_
18
7.8
17
17
21
8.0
83
12
4.6
3.7
5.5
7.1
9.7
5.4
5.7
6,9
0.6
2.2
2.0
2.6
2.0
4.4
10
5.0
6.5
2.0
13
5.1
Average
recovery
0
b
0
0
32.5
41.6
0
39.9
543
58.7
89.8
84.8
91.8
85.6
103
89.4
70.7
95.2
91.0
95.8
92.8
73.4
106
119
89.5
70.7
41.1
96.9
943
106
Percent
RSD
_
_
_
41
27
_
18
16
8.9
23
3.4
6.3
1.8
5.7
3.0
33
8.8
4.2
4.2
43
8.1
53
4.6
6.1
8.8
16
53
53
7.4
a The operating conditions for Soxtec apparatus were as follows: immersion time - 60 min; extraction time - 60 min;
the sample size was 10 g clay, the spiking level was 50 ng/g, except compounds 23, 28, and 29 at 150 ng/g,
compound 26 at 300 ng/g, compound 3 at 200 ng/g, and compounds 1 and 2 at 500 ng/g. The number of
determinations was four. The moisture content of the matrix was not altered.
"Not able to determine because of interference.
11-47
-------
TABLE 3. AVERAGE PERCENT RECOVERIES AND PERCENT RSDs FOR 30 TARGET COMPOUNDS EXTRACTED
FROM VARIOUS SPIKED MATRICES WITH SUPERCRITICAL CARBON DIOXIDE8
Sand with 10 percent
Sand
Compound
no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Compound
1,3-Dichlorobenzene
1,2-Dichlorobenzene
Nitrobenzene
Benzal chloride
Benzotrichloride
4-Chloro-2-nitrotoluene
Hexachlorocyclopentadicne
2,4-Dichloronitrobenzene
3,4-Dichloronitrobenzene
Pentachlorobenzene
2,3,45-TetrachIoronitrobenzene
Benefit!
alpha-BHC
Hexachlorobenzene
delta-BHC
Heptachlor
Aldrin
Isopropalin
Heptachlor epoxide
trans-Chlordane
Endosulfan I
Dieldrin
25-Dichlorophenyl-4'-nitrophenyl ether
Endrin
Endosulfan II
2,4,6-Trichlorophenyl-4'-nitrophenyl ether
p,p'-DDT
2,3,6-Trichiorophenyl-4'-nitrophenyl ether
2,3,4-Trichlorophenyl-4'-nitrophenyl ether
Mirex
Average
recovery
0
0
0
0
0
56.9
16.9
57.8
68.7
60.9
75.3
70.4
72.5
60.6
68.7
82.9
76.9
112
79.6
71.0
76.7
82.9
66.9
68.5
62.8
81.6
71.1
71.3
56.7
81.1
Percent
USD
_
_
25.9
11.8
16.1
30.1
16.8
15.1
9.5
13.6
10.2
2.8
125
19.2
11.8
11.6
4.3
32.2
29.5
2.5
27.8
26.2
9.9
15.6
3.5
22.6
30.3
Clay
Average
recovery
0
0
0
0
0
57.8
15.1
62.3
62.9
50.9
65.8
65.9
66.1
56.7
73.8
63.4
62.0
70.8
70.7
71.1
68.9
114
74.3
76.4
76.7
76.6
86.9
72.9
68.7
67.3
Percent
RSD
_
-
-
28.9
60.1
27.9
25.6
21.8
26.3
27.8
31.3
28.1
25.8
28.6
37.8
26.0
27.1
28.2
24.3
21.1
25.2
24.7
26.5
23.3
19.7
225
27.3
27.6
Topsoii
Average
recovery
0
0
53.6
30.1
0
683
47.4
68.9
69.9
68.3
74.2
76.0
75.3
735
81.3
74.9
75.3
76.9
79.4
80.4
79.2
845
793
79.7
76.9
78.9
82.6
79.1
68.9
79.0
Percent
RSD
_
32.9
36.3
-
24.4
33.1
24.2
23.4
20.1
19.1
18.7
18.7
18.2
14.7
17.6
19.3
16.8
18.7
17.3
15.7
15.8
16.0
15.8
13.8
14.9
15.6
18.2
18.6
16.4
organic compost
Average
recovery
93.6
795
76.6
81.6
80.3
86.4
81.8
73.4
76.0
84.0
73.9
73.9
75.3
74.1
81.7
87.7
775
81.1
81.9
80.4
78.3
82.3
80.6
83.9
82.8
88.3
84.7
79.1
75.7
78.2
Percent
RSD
29.2
24.3
22.7
24.6
27.1
245
23.6
22.9
18.9
24.0
205
21.9
22.7
21.9
21.7
20.9
195
17.9
19.6
21.3
16.2
19.6
19.7
18.8
22.8
15.4
12.7
17.9
19.1
20.2
aThe number of samples extracted in parallel for each matrix was four. The experiments were performed with supercritical carbon dioxide at 300 atm/70ฐC/60 min dynamic. The
sample size was 2 g. The spiking level was 25 ng/g, except compounds 23, 28, and 29 at 75 ng/g, compound 26 at 150 ng/g, compound 3 at 100 ng/g, and compounds 1 and 2 at
250 ng/g. The moisture content of the matrix was not altered.
-------
TABLE 4. AVERAGE PERCENT RECOVERIES AND PERCENT RSDs FOR
BASE/NEUTRAL/ACIDIC COMPOUNDS EXTRACTED FROM SPIKED
SAND WITH SUPERCRITICAL CARBON DIOXIDE3
Compound
no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Compound name
Phenol
Bis(2-chloroethyi)ether
2-Chlorophenol
1 ,3-Dichlorobenzene
1,4-Dichlorobenzene
Benzyl alcohol
1,2-Dichlorobenzene
2-Methylphenol
Bis(2-chloroisopropyl)ether
4-Methylphenol
N-nitroso-di-n-propylamine
Hexachloroethane
Nitrobenzene
Isophorone
2-Nitrophenol
2,4-Dimethylphenol
Benzoic acid
Bis(2-chloroethoxy)methane
2,4-Dichlorophenol
1 ,2,4-Trichlorobenzene
Hexachlorobutadiene
4-Chloro-3-methylphenol
2-Methylnaphthalene
Hexachlorocylopentadiene
2,4,6-Trichlorophenol
2,4,5-Trichlorophenol
2-Chloronaphthalene
Dimethylphthalate
2,4-Dinitrophenol
4-Nitrophenol
Dibenzofuran
2,4-Dinitrotoluene
2,6-Dinitrotoluene
Diethyl phthalate
4-Qilorophenyl-phenylether
4,6-Dinitro-2-methylphenol
4-BromophenyI-phenyiether
Hexachlorobenzene
Pentachlorophenol
Di-n-butylphthalate
Butylbenzylphthalate
Bis(2-ethylhexyl)phthalate
Di-n-octylph thalate
Terphenyl-dj4
Spike level
(ng/g)
300
150
300
150
150
150
150
300
150
300
150
150
150
150
300
300
300
150
300
150
150
300
150
150
300
300
150
150
300
300
150
150
150
150
150
300
150
150
300
150
150
150
150
20.0b
Average
recovery
50.9
23.6
25.9
4.2
4.7
54.4
8.2
54.4
27.1
64.5
58.9
5.4
41.9
60.4
50.2
65.5
7.3
61.6
63.6
32.6
25.0
71.8
62.2
46.6
71.5
80.2
69.5
59.2
37.2
10.0
78.0
71.0
78.6
66.7
79.9
53.9
773
78.2
65.2
73.0
545
71.9
58.0
92.1
Percent
USD
26.3
75.0
64.6
160
156
135
119
25.1
69.1
155
17.9
151
38.7
12.1
31.8
16.4
24.6
23.1
16.6
51.7
66.4
10.1
19.4
33.8
9.4
7.3
14.4
15.5
19.7
27.4
6.9
7.3
7.3
135
10.6
16.7
11.0
7.7
12.4
12.0
19.8
14.0
16.1
2.2
a The number of samples extracted in parallel was four. The experiments were performed at 150 atm/50ฐC/10 min static
followed by 200 atm/60ฐC/10 min dynamic and 250 atm/70ฐC/10 min dynamic. The sample size was 3 g dry sand.
b Spiked at 20 ng//iL in the collection vial.
11-49
-------
TABLE 5. AVERAGE PERCENT RECOVERIES AND PERCENT RSDs FOR
BASE/NEUTRAL/ACIDIC COMPOUNDS EXTRACTED FROM SPIKED CLAY
BY SOXTEC EXTRACTION WITH HEXANE-ACETONE (1:1)
Coaccntration (ng/jiL)
CoMpound
a.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Compound Mine
Phenol
Bis(2-chloroethyl)ether
2-Chlorophenol
13-Dichlorobenzene
1,4-Dichlorobenzene
Benzyl alcohol
1,2-Dichlorobenzenc
2-Mcthylphenoi
Bis(2-ch!oroisopropy1)ether
4-Methylphenol
N-nitroso-di-n-propylamine
Hexachloroethane
Nitrobenzene
Isophorone
2-Nitrophcnol
2,4-Dimethylphenol
Benzoic acid
Bis(2-chloroethoxy)mcthane
2,4-Dichlorophenol
1,2,4-Trichlorobenzene
Hexachlorobutadiene
4-Chloro-3-mcthylphenol
2-Methylnaphthalene
Hexachlorocylopentadiene
2,4,6-Trichlorophenol
2,45-Trichlorophenol
2-ChloronaphthaIene
Dimethyl phthalate
2,4-Dinitrophcnol
4-Nitrophcnol
Dibenzofuran
2,4-Dinitrotolucne
2,6-Dinitrotol uene
Diethyl phthalate
4-Chiorophenyl-phenylether
4,6-Dinitro-2-methylphenol
4-Bromophenyi-phenylether
Hexachlorobenzene
Pentachlorophenol
Di-n-butyl phthalate
Butyibenzyl phthalate
Bis(2-ethylhexyl)phthalate
Di-n-octyl phthalate
Clayl
14.4
65
12.3
ND
ND
16.9
ND
5.3
3.8
7.2
12.6
ND
7.8
16.7
10.0
14.7
12.5
13.0
16.5
4.4
ND
20.0
12.7
45
20.8
7.9
17.2
22.2
28.1
20.8
23.9
25.1
20.8
22.9
19.7
20.1
18.7
21.4
19.7
33.2
20.4
23.6
23.8
Clayl
15.1
8.2
13.4
ND
ND
17.9
ND
5.6
4.6
7.4
13.1
ND
8.5
17.6
11.1
16.0
12.9
17.5
17.5
S3
ND
20.2
14.7
6.4
22.5
83
19.4
23.6
29.7
20.3
26.3
26.7
21.5
23.4
20.9
19.4
193
22.7
19.2
23.0
20.6
24.0
26.2
Clay 3
13.5
8.2
12.7
ND
ND
155
ND
4.9
5.1
65
11.6
ND
9.1
16.2
11.3
14.4
11.1
13.0
16.0
6.6
ND
18.4
14.9
65
19.9
7.9
185
21.3
24.9
24.9
23.7
24.0
19.2
21.1
19.9
17.6
18.2
212
175
14.3
18.7
21.9
24.8
Average
reeoveiy
(percent)
47.8
25.4
42.7
0
0
55.9
0
17.6
15.0
23.4
41.4
0
28.2
56.1
36.0
50.1
40.6
44.1
55.6
18.1
0
65.1
47.0
193
70.2
26.8
61.2
74.6
91.9
62.9
82.1
84.2
683
74.9
67.2
63.4
62.4
72.6
62.7
78.3
663
77.2
83.1
Percent
USD
5.6
13
43
-
7.2
-
6.6
14.6
6.7
6.2
7.7
4.2
65
5.7
7.7
3.0
4.6
31
-
5.1
8.6
19
6.3
2.9
6.0
5.2
8.9
16
5.9
5.4
5.8
5.4
3.2
6.8
3.0
3.7
6.1
40
5.2
4.8
4.8
Soxtec samples included additional 21 compounds not listed here. The operating conditions for the Soxtec apparatus were as
follows: immersion time - 45 min; extraction time 45 min; the sample size was 10 g clay, the spiking level was 6 ft-gjg- The
moisture content of the matrix was not altered.
11-50
-------
TABLE 6. PERCENT RECOVERIES OF COMPOUNDS EXTRACTED FROM THE
SRS 103-100 STANDARD REFERENCE MATERIAL BY SOXTEC
EXTRACTION (HEXANE-ACETONE 1:1) AND BY SFE WITH
SUPERCRITICAL CARBON DIOXIDE
Compound name
Naphthalene
2-Methylnaphthalene
Acenaphthylene
Acenaphthene
Dibenzofuran
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Benzo(k + b)fluoranthene
Benzo(a)pyrene
Pentachlorophenol
Certified
value
(rag/kg)
32.4 ฑ 8.2
62.1 ฑ 11.5
19.1 ฑ 4.4
632 ฑ 105
307 ฑ 49
492 ฑ 78
1618 ฑ 348
422 ฑ 49
1280 ฑ 220
1033 ฑ 289
252 ฑ 38
297 ฑ 26
152 ฑ 22
97.2 ฑ 17.1
965 ฑ 374
Soxtec3
percent recovery
127
127
110
108
123
92.7
81.3
131
81.3
69.1
95.2
91.6
118
80.2
111
SFEb
percent recovery
63.8
82.6
64.6
98.2
92.9
80.4
124
78.4
92.3
78.2
67.6
68.4
53.3
32.2
141
a Single determination. The operating conditions for the Soxtec apparatus were as follows:
immersion time - 45 min; extraction time - 45 min; the sample size was 10 g.
" The values given represent the average recoveries for three replicate samples extracted
sequentially. The sample size was 2.5 g. The extraction was performed with carbon
dioxide at 300 atm and 70ฐ C for 60 min; 10 percent moisture was added to each sample
prior to extraction.
1-51
-------
01
ro
TABLE 7. PERCENT RECOVERIES OF COMPOUNDS EXTRACTED FROM THE HS-3 AND HS-4 MARINE SEDIMENTS
BY SOXTEC EXTRACTION (HEXANE-ACETONE 1:1) AND BY SFE WITH SUPERCRITICAL CARBON DIOXIDE
Compound name
Naphthalene
Acenaphthylene
Acenaphthene
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Benzo(a)pyrene
Benzo(b)fluoranthene
Benzo(k)fluroanthene
Benzo(ghi)perylene
Dibenzo(ah)anthracene
Indeno( l,2,3-cd)pyrene
Certified I
value
(rag/kg)
9.0
0.3
4.5
13.6
85
13.4
60
39
14.6
14.1
7.4
7.7
2.8
5.0
1.3
5.4
ฑ 0.7
ฑ 0.1
ฑ 1.5
ฑ 3.1
ฑ20
ฑ 0.5
ฑ 9
ฑ 9
ฑ 2.0
ฑ 2.0
ฑ 3.6
ฑ 1.2
ฑ 2.0
ฑ 2.0
ฑ 0.5
ฑ 1.3
IS-3 percent recovery
Soxtec8
47.8
167
129
53.7
44.9
75.4
51.3
43.1
56.2
57.4
48.6
71.4
175
56.0
92.3
51.9
SFEb
11.1
18.9
15.1
26.1
12.3
28.7
27.7
28.1
30.5
9.5
22.1
85.7
Certified
value
(mg/kg)
0.15
0.15
0.15
0.15
0.68
0.14 ฑ0.07
1.25 ฑ0.10
0.94 ฑ0.12
0.53 ฑ0.05
0.65 ฑ0.08
0.65 ฑ0.08
0.70 ฑ0.15
0.36 ฑ0.05
0.58 ฑ0.22
0.12 ฑ0.05
0.51 ฑ0.15
HS-4 percent recovery
Soxteca
~
85.3
129
88.0
95.7
71.7
76.9
58.5
71.4
133
~
SFEb
26.5
ซ
21.6
18.1
-
~
~
a Single determinations. The operating conditions for the Soxtec apparatus were as follows: immersion time - 45 min; extraction
time - 45 min; the sample size was 10 g.
b The HS-3 sample (2 g) was extracted at 350 atm and 60ฐC for 20 min (single determination). The HS-4 sample (four 1.5-g
samples extracted in parallel, the extracts were then combined) was extracted at 350 atm and 70ฐ C for 30 min.
-------
CH ANALYSIS FOR SELECTED APPENDIX IX COMPOUNDS IN ENVIRONMENTAL MATRICES BY
HIGH PERFORMANCE LIQUID CHROMATOGRAPHY/PARTICLE BEAM MASS SPECTROMETRY
Jeffery L. Cornell, Senior Scientist, Jeffrey C. Lowry, Director of
Organics, Marshall D. Tilbury, Scientist, Enseco-Rocky Mountain Analytical
Laboratory, 4955 Yarrow Street, Arvada, Colorado 80002
ABSTRACT
A method is described for the analysis of selected Appendix IX analytes in
environmental matrices, using high performance liquid chromatography-mass
spectrometry (LC/MS) employing a particle beam interface. The method is
targeted at compounds that are not easily extracted and/or analyzed by the
current methodologies. The five selected compounds are chromatographed on
an octadecylsilane (C-18) reverse phase column, using methanol and ammonium
acetate in water. A commercial particle beam interface is used in
conjunction with a quadrupole mass spectrometer for detection and
quantitation of the analytes. One internal standard and two surrogate
standards have been included. Multipoint calibration curves indicate that
concentration versus response data fit a second order polynomial model.
This second order equation can then be used for quantitation of subsequent
check standards and samples.
Sample preparation methods are described to extract these compounds from
water and soil samples. For controlled matrices, the average recovery of
the analytes from water samples is 73% and 84% for soils.
INTRODUCTION
In 1987, the Environmental Protection Agency (EPA) promulgated regulations
which required owners and operators of hazardous waste treatment, storage
and disposal facilities to analyze their ground water for a list of 232
constituents listed in Appendix IX of 40 CFR, Part 264 (52 Federal Register
25942). The Appendix IX list consists of metal, anions, and a wide variety
of organic compounds, including nitrosamines, phenols, polynuclear aromatic
hydrocarbons, volatile organics, pesticides, herbicides and chlorinated
dioxins.
Since promulgation of this regulation, commercial laboratories have
struggled to develop analytical protocols to address this extensive list.
In particular, a number of polar compounds which are not amenable to
conventional methodologies have presented challenges to the laboratory
community. A review of the public docket to the Appendix IX rule making,
funded research by EPA, presentations at analytical methods caucuses and
SW-846, indicate that these exotic compounds still pose a challenge for
routine analysis (1). Although several laboratories have evaluated
conventional HPLC as an analytical tool, this approach has not been widely
accepted, due to detection limit and identification reliability concerns,
especially in contaminated matrices (2).
1-53
-------
Enseco's approach for Appendix IX analyses has been described elsewhere
(3). The current approach provides reliable data for all Appendix IX
compounds except for five polar compounds. These compounds are:
p-phenylenediamine, dimethoate, 4-nitroquinoline-n-oxide, famphur and
hexachlorophene. Although it is possible to incorporate dimethoate and
famphur into method 8140, method 8270 was the recommended method for these
compounds. Our recovery studies using method 8270 have consistently
indicated that these compounds are not measurable using method 8270 with
conventional sample preparation techniques (4). Because of this, we have
developed the following LC/MS technique, combined with modified 8270-1 ike
sample preparation methods to provide reliable data for these compounds in
environmental matrices.
EXPERIMENTAL
Reagents and Chemicals
All analytes were obtained at 98% purity or higher from Aldrich, Cambridge
Isotope Laboratories and MSD Isotopes. Standards at working concentrations
were verified using USEPA certified check standards, where available.
Labeled compounds were verified against their native counterparts. All
reagents used were HPLC grade or equivalent. All standards stocks and
working concentrations are made in acetonitrile (p-phenylenediamine
degrades in methanol). The compounds of interest are listed below and
their structures shown in Figure 1.
CAS Number Target Compound
106-50-3 p-Phenylenediamine
60-51-5 Dimethoate
56-57-5 4-Nitroquinoline-n-oxide
52-85-7 Famphur
70-30-4 Hexachlorophene
Internal Standard: Caffeine-13-C3
Surrogate Standards: p-Phenylenediamine-d4
Malathion-dlO
The deuterated phenylenediamine (PDA) was chosen as a surrogate as the
native PDA has been the most difficult compound to extract, and can be
quite reactive. Labeled malathion was included due to its similarity to
dimethoate and famphur.
Instrumental Conditions
The HPLC instrumentation consisted of a Hewlett Packard 1090L liquid
chromatograph with a ternary pumping system and a filter photometric
detector. The UV detector was useful for off-line method development and
as a diagnostic tool. The LC was equipped with a variable volume injector
(2 uL was the nominal injection volume) and an autosampler. The column
used was an Ultracarb ODS(30) 2x250mm manufactured by Phenomenex. The
mobile phase consisted of water (modified with 0.01M ammonium acetate) and
methanol used in the following gradient at a flow rate of 0.20 mL/min.
11-54
-------
Figure 1
CRS Number = 166563
CAS Number = 56575
CRS Number = 68515
CH3
I
0
CH3-0 P S CH2CNH-CH3
ii ii
S 0
(taut oner)
CRS Number = 78384
Cl OH Ci
CAS Number = 52857
CH, 0
NS=0
0
CH3-0 P0-CH3
OH Cl
11-55
-------
Water* Methanol Minutes
50% 50% 0
0% 100% 9
* with 0.01M arnmomium acetate
The particle beam interface used was a Hewlett Packard model 59980A.
Typical helium pressures were from 35 to 45 psi, desolvation chamber
temperatures from 45-55 degrees and nebulizer position was determined
experimentally based on flow injections of caffeine. The interface was
connected (via the standard transfer line) to a Hewlett Packard model 5988A
mass spectrometer operated in the electron impact ionization mode at 70 eV
and 300 uA emission current. The ion source was operated at 250 to 300
degrees and the scan range 62 to 450 amu at sufficient speed to allow for
at least 10 scans per chromatographic peak. The electron multiplier was
operated at between 2100 and 2300 volts.
As a starting point an autotune routine using perfluorotributyl amine
(PFTBA) can be used for mass spectrometer tuning. It was often useful to
then maximize the tune on m/z 219 to provide good mid-mass sensitivity.
The instrument was tuned using the following guidelines for PFTBA.
m/z abundance
69 100%
131 25-75%
219 25-75%
502 >0.5%
These abundances will allow reasonable correlation with NIST or other El MS
libraries. Mass peak width and axis calibrations are performed as needed.
The particle beam interface is optimized using manufacturer guidelines.
This performance is verified on a daily basis following tuning but prior to
the injection of calibration standard(s). Several flow injections (column
bypassed) of 20 ng (2uL injections of a 10 ng/uL solution) of caffeine are
performed at 50:50 methanolrwater with 0.01M ammonium acetate. Data is
acquired in the SIM mode monitoring m/z 194. The peak areas are
integrated and evaluated for sensitivity and precision. It was typical to
expect approximately 500,000 area counts with a precision of approximately
5%, injection to injection. This step establishes that the system is
functioning properly before doing any chromatography.
Calibration and Quantisation
There has been a great deal of discussion regarding quantisation in the
area of particle beam LC/MS (5,6,7). We have found that the relationship
between concentration and response is not linear in the traditional sense
(e.g. GC/MS and the use of average response factors in environmental
analyses). Instead, this relationship is best described using a second
order polynomial expression. Although this has not historically been the
approach for environmental analysis, data has recently been shown that the
11-56
-------
accuracy of an existing method (based on linear calibration) can be
improved by utilizing a second order calibration (8). Once established,
quantitation can be performed using the second order equation, rather than
an average response factor.
The initial or multipoint calibration consists of a minimum of five points
covering one order of magnitude for each analyte and surrogate. Plots of
concentration versus response (using extracted ion areas) are generated
following analysis and data processing of the points. Given a reasonable
fit (r=0.95 or better), the data system is updated with the second degree
equations and the points requantitated against the curve. The percent
difference between actual and theoretical concentration is calculated to
determine the quality of the calibration curve for each analyte.
The continuing calibrations consist of a midpoint level standard of all
analytes and surrogates, to be performed after the nebulizer performance
verification, but before sample analysis. Again, concentration values are
calculated by the data system using the second order equations determined
in the initial calibration. The percent difference between these
concentrations and the theoretical values are calculated for each compound
to determine if samples can be analyzed. Other continuing calibrations (at
differing concentration levels) will be analyzed every five samples and
percent differences checked as before.
Background Subtraction
Due to the constant presence of background spectra, characteristic of the
interface, a background subtraction procedure was used to make low level
spectra identification easier. First, a copy of the original data file is
made and archived so that an unaltered version will always be available.
Next, spectra for ten scans (+/- 5 scans at 1 minute into the run) is
averaged and the resulting spectra is subtracted from each scan in the data
file. (This spectra is also archived on the data system). One minute was
chosen because this is before the void volume of the column has eluted as
was found to be representative of the background ions present. A report is
generated showing the total ion chromatogram (TIC) before subtraction, the
spectra used for subtraction and the TIC after subtraction.
Sample Preparation
The preparation of water samples consisted of extraction with methylene
chloride by continuous liquid-liquid extractor and Kuderna-Danish (K-D)
evaporative concentration. For the controlled matrix experiments, 1 liter
of de-ionized, carbon filtered water was spiked with the target compounds
and surrogates. The pH was measured, and buffered at pH 7 with a potassium
dihydrogen phosphate/sodium hydroxide buffer. This is added because pH
control is critical to the extraction of phenylenediamine. Additionally,
sodium chloride (35 g) is added to the water to facilitate the extraction
of dimethoate. Methylene chloride was then added and the extraction run
for 18 hours, as in method 3520, SW-846. Following this, the extract was
concentrated to approximately 5 ml in a K-D. The concentration was
continued to about ImL under a stream of nitrogen and then exchanged to
acetonitrile. The extract was then evaporated to a final volume of 0.5 ml.
11-57
-------
The sample preparation method for soils uses a 30 gram extraction using 1:1
methanoltmethylene chloride by sonication (as in method 3550), and
concentration by Kuderna-Danish (K-D). For the controlled matrix
experiments, ottowa sand was spiked with the target compounds and
surrogates. A 100ml aliquot of 1:1 methanol:methylene chloride was added
and the samples sonicated for 3 minutes at an output setting of 10 and a
duty cycle of 50%. The extract was decanted and two more 100ml sonications
were performed. The methanol/methylene chloride extract was then filtered
and concentrated to approximately 5 ml in a K-D. The concentration was
continued under a stream of nitrogen and simultaneously exchanged to
acetonitrile. The extract was then evaporated to a final volume of 0.5ml_.
RESULTS AND DISCUSSION
Several authors have discussed that mobile phase modifiers such as ammonium
acetate can enhance the MS response of some compounds when using the
particle beam interface (9,10). Although we have also confirmed this
effect, ammonium acetate was also used for chromatographic reasons. It was
found that without this modifier, p-phenylenediamine and hexachlorophene
showed very poor peak shape. Two possible explanations are that the
ammonium acetate is acting as either an ion pairing agent, or simply
deactivating silanol sites on the stationary phase. New columns must be
conditioned for several hours with the gradient described before acceptable
chromatography can be achieved. Once this is done, however, all compounds
show good peak shape and separation, as shown in the total ion chromatogram
in Figure 2.
The numbers labeling the peaks in figures 2 and 3, as well as the numbers
labeling spectra in figure 4 all correspond to the following definitions:
# Compound
1 - p-Phenylenediamine and p-phenylenediamine-d4
2 - Dimethoate
3 - 4-Nitroquinoline-n-oxide
4 - Famphur
5 - Malathion-dlO
6 - Hexachlorophene
It is interesting to note that in the MS total ion chromatogram,
hexachlorophene exhibits significantly more peak tailing than in the UV
chromatogram shown in Figure 3. This is likely due to memory effects in
the ion source itself. This effect becomes more pronounced at lower source
temperatures (200-250 degrees) and diminishes as source temperature
increases (250-300 degrees). Since a 300 degree source temperature had no
negative affect on other compounds response, much of the work was done at
this temperature.
All of the spectra obtained for these analytes show good correlation with
NIST spectra with one exception. 4-Nitroquinoline-n-oxide shows a base m/z
of 190 (the molecular ion) in the reference spectra, but this was not
obtained experimentally. The base mass we obtained was m/z 144 and a small
(10% relative abundance) peak at m/z 190. The difference of 46 is
1-58
-------
Figure 2
File >PB89462.8-456.0 ami. 2UL-RP9 STD.5/9-180 fSB'SB MปOH/H20+.8ln HH
TIC
1608001
148880-
120008-
100800-
80000-
60000-
48888-
28880-
\
188
-90
ซe
70
60
50
r40
r30
-20
-18
1.8 2.8 3.0 4.0 S.0 6.8 7.8 8.8 9.818.011.012.013.014.015.0
UV = 254 nm
-100
99
ซe
70
-68
-50
40
-30
20
10
0
010 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.010.011.012.013.014.016.0
Figure 3
11-59
-------
Figure 4
File >PB894 2UL-RP9 STD. Scan 69
Bpk Rb 22888. ENH 3.45 Bin.
25888} I08 I?2 r
28888:
16888-
leeoo-
6888-
a.
^^-
j
1
88
(
7 .1
{ 1 9l
-188
80
60
48
-28
jป
88 188 128
Fil* >PB894 2UL-RP9 STD. Scan 149
Bpk Rb 8378. ENH 7.26 Bin.
-j 87 125 r
8888-
600B-
4888-
2008-
_
gj
!
1 i
V
u
J
143 ,97
/ { 229
' f
LLk. 1.
-188
-68
-68
-40
-28
_o
188 150 208 258
Fil* >PB894 2UL-RP9 STD. Scan 115
Bpk Rb 15486.
16888-
-
1288*
-
8888-
_
4888-]
ENH 5.64 Bin.
197 r
111
68
/
[JJ
ie'0
*
168
\
/
223
iee
ee
ซe
48
28
s
158 280 250
File >PB894 2UL-RP9 STD. Scan 178
Bpk Rb 12335.
12000-
8800-
-
4000-
-
B.
ENH 8.65 Bin.
144
*
117
88
s
1
jj \_ Jy^j
4
X
1 88
, /
-188
-60
-68
-48
-28
ja
128 168 288
File >PB894 2UL-BP9 STD. Scan 213
Bpk Rb 11328. ENH 18.32 Bin.
218 .
12888-
8888-
4888-
a_
^
5
93 125
\ f
1
1 C7 9 A 2
1 PB894 2UL-RP9 STD. Scan 245
Bpk Rb 8787.
8808-
-
6008-
4080-
2888-
:
6
72
/
L
'se'
ENH 11.84 Bin.
132 -
t
188
r
,
JU
158 183
iee
ee
68
48
28
a
128 168 288
Fil* >PB894 2UL-RP9 STD. 5/9-188 |58/50 H*OHxH20+.81H NH40RC-> Scan 261
Bpk Rb 11482. ENH
12888-
8888-
4888-
1 198
7
89 '/
\ / 132
1 1 { lซ
; L
JLAL^JL^L^.
299
231 2f5 \
dLjL JL .Jl
88 128 168 288 248 288
12.61 Bin.
327 /5 3?e 486
^Ik. . ,< \
328 368 ' 480
-iee
88
68
-48
20
8
11-60
-------
accounted for by the loss of the nitro group, caused by decomposition in
the ion source. The relative abundance of m/z 190 increases( with a
decrease in m/z 144) at lower source temperatures (225 or less). Given the
reduction in overall response of all compounds at lower temperatures (and
the increased memory effect), source temperatures of 250 to 300 degrees
were best for this analysis. Spectra obtained from a midpoint standard are
shown in Figure 4.
Calibration and Quantisation
Several calibration curves have been run for the analytes and surrogates
using from five to ten points, covering nearly one order of magnitude. The
concentration versus response plots indicate that the relationships are
best described by second order polynomial equations. An example of a 5
point calibration for 4-nitroquinoline-n-oxide is shown in Figure 5. The
correlation of these curves is usually 0.99 or better. Once the data
system is updated with this information, the curve can be requantitated
against itself and percent differences calculated. The data for four
different calibrations is shown in Table 1.
Figure 5
Cซlib File I CREBP9IIQT Coปp
Cซlib Datvi 910611 88|42
Coapi 4-Nitroquinolinป-l-oxidป
328-
Cane.
ug/BL
248-
zee-
168-
128-
88-
48-
fivปrปgซ RF
lปt Dปgrปป
2nd Dปgrป*
8.8
.2
1.4
1.6
Response Ratio
Compound I 4 Callb File: CA5AP9::QT
Conpound: 4-N1troqu1nol1ne-l-ox1de
Istd: !3-C3-Caffe1ne
File: >PB895 >PB896 >PB894 >PB897 >PB898
Cone: 20.00 50.00 100.00 150.00 200.00
Rf: .22431 .44609 .54931 .58918 .70075
Average of 5 Rfs: .50193 (35.86 X Rsd) Rx: .0000000 Ry: .0000000
1st Degree Equation: y - .2089807 + 1.339279(x)
'gd^'g.tfi! ;"361066935 + 1.872203(x) +
2nd Degree Corr Coef: .9996580
In the above equations:
Cone Std Area Std
y - X
Cone Istd Area Istd
Istd Cone for all calibration points Is: 100.00
11-61
-------
Table 1
INITIAL CALIBRATION:
Percent Difference Data
Curve "A"
Curve "B1
Compound
p-Phenylenediamine
Dimethoate
4-Nitroquinoline-n-oxide
Famphur
Hexachlorophene
Ave. %Diff.
19
5.9
16
4.7
22
S.D. Ave. %Diff
12
6.8
4.8
4.3
8.5
11
5.2
4.1
4.1
3.8
S.D.
5.3
3.4
3.4
3.4
3.2
Curve "C"
Curve "D"
Compound
p-Phenylenediamine-d4
p-Pheny1enedi ami ne
Diemthoate
4-Nitroquinoline-n-oxide
Famphur
Malathion-dlO
Hexachlorophene
Ave. %Diff.
n/a
2.7
2.7
0.72
1.8
n/a
1.0
S.D. Ave. %Diff.
n/a
2.2
2.5
0.93
1.8
n/a
0.89
3.1
6.1
8.6
1.3
1.9
4.8
1.5
S.D.
2.6
3.8
12
1.5
1.9
5.8
1.1
* For curves "A" and "B", the concentration range covered is as follows:
p-phenylenediamine; dimethoate and 4-nitroquinoline-n-oxide ranged from 40
to 180 ug/mL in 20 ug/mL increments; famphur covered 20 to 90 ug/mL in 10
ug/mL increments; and hexachlorophene ranged from 200 to 900 ug/mL in 100
ug/mL steps.
I For curves "C" and "D", the concentration range covered in five points is
as follows; p-phenylenediamine, p-phenylenediamine-d4 ("D" only) and
4-nitroquinoline-n-oxide ranged from 25 to 200 ug/mL; dimethoate and
malathion-dlO ("D" only) ranged from 50 to 400 ug/mL; famphur ranged from
12.5 to 100 ug/mL; and hexachlorophene ranged from 125 to 1000 ug/mL.
11-62
-------
Based on the data in Table 1, the points fit the second order calibrations
very well. However, one has to look at how well the continuing
calibrations over time, compare with the curves to know if this response
remains predictable. Table 2 summarizes the percent differences obtained
for 16 standards run over a two week period after the analysis of curve
"A". Concentrations were calculated by the data system using the second
degree equations and then percent differences were calculated comparing
these concentrations to true values. The data shows that all percent
differences were less than 30% until the 13th day after the initial
calibration, indicating good stability of the initial calibration. Based
on this data, one would have likely decided to establish a new initial
calibration on the 13th day.
Table 2
CONTINUING CALIBRATION:
Percent Difference Data
2-M 2-L
Day and Standard Level
2-H 3-M 3-M 6-M 6-L
6-H 6-M
PDA
DMT
NQO
FMR
HXN
0.2
22
3.6
12
12
0.6
1.6
26
4.6
3.2
6.3
8.6
6.4
0.4
1.7
5.8
14
13
16
8.4
3.6
18
9.5
20
6.5
11
12
15
7.0
20
6.4
5.0
7.1
3.0
10
12
4.6
19
3.8
1.6
14
15
9.9
11
16
7-M 8-M 9-M 9-M 13-M 13-M 14-M avq.%d1ff S.D.
PDA
DMT
NQO
FMR
HXN
13
1.4
5.7
2.8
3.9
13
4.1
13
11
5.2
11
19
23
11
25
23
4.4
19
3.3
2.3
28
0.8
37
9.9
13
19
25
39
5.7
45
0.9
4.4
28
3.9
9.0
10
10
17
7.8
11
8.0
7.9
11
5.4
11
Notes:
The initial multipoint calibration (curve "A") was run on Day 1.
L = Low standard, usually half the concentration of the midpoint,
M = Midpoint standard, the middle of the calibration range.
H = High standard, usually 1.5-2 time the level of the midpoint.
11-63
-------
Sample Extraction Results
Seven 1L replicates of carbon filtered, deionized water were spiked at the
levels shown below. The average percent recovery and standard deviation
are also shown.
ug/L
Compound Cone.Spiked Avg.% Rec. S.D.
p-Phenylenediamine-d4 50 59 16
p-Phenylenediamine 50 59 15
Dimethoate 100 86 5.8
4-Nitroquinoline-n-oxide 50 74 6.2
Famphur 25 81 6.8
Malathion-dlO 100 87 9.3
Hexachlorophene 250 63 7.9
Similarly, seven controlled soils (Ottowa Sand) were spiked at the levels
shown below and the results expressed as average percent recovery and
corresponding standard deviations.
ug/kg
Compound Cone.Spiked Avg.% Rec. S.D.
p-Phenylenediamine-d4 1670 64 23
p-Phenylenediamine 1670 65 21
Dimethoate 3330 90 9.0
4-Nitroquinoline-n-oxide 1670 77 12
Famphur 833 97 12
Malathion-dlO 3330 97 14
Hexachlorophene 8330 96 10
The levels at which the waters and soils were spiked, falls at the mid
point in terms of instrument calibrations and final extract concentration.
Because of this, the actual method detection limits will likely be 2 to 4
times lower than the spiking level shown above.
SUMMARY
A method using particle beam LC/MS has been developed for the analysis of
some intractable Appendix IX compounds. Together, with previously
established methods, it will be possible to measure the entire Appendix IX
list. Calibration and subsequent quantisation is performed by taking
advantage of the second order behavior that appears to be characteristic
of the particle beam interface for these compounds. The data demonstrates
that this approach provides a reliable method of initial calibration.
Furthermore, the analytical stability of the curves over several days has
been demonstrated.
1-64
-------
It is noteworthy that quantisation by UV detection was not practical given
the poor UV response of two compounds and one surrogate. Additionally,
the labeled PDA could not have been used as a recovery surrogate with UV
as the quantitation method. Lastly, when "real" environmental samples
with various contaminants are analyzed, the interferences associated with
UV detection could make quantitation difficult.
ACKNOWLEDGEMENTS
Special thanks go to Todd Burgesser, Robert E. Moul, Bill Zdinak, Susan
Davis, and members of the Enseco-RMAL organic sample preparation group.
Thanks also to Jerry Parr for his valuable input.
Additionally, Hewlett Packard is to be acknowledged for their outstanding
level of technical assistance and support.
REFERENCES
1.) "Summary of Analytical Methods for Appendix IX",RCRA Docket number
F-87-AX9F-FFFFF.
2.) Porter, J.W., "Technical Guidance on Ground Water Analysis for
Appendix VIII", USEPA, February 1986.
3.) Parr, J.L., et.al., "Establishing an Analytical Protocol for the
Measurement of EPA's Appendix IX List", in Ground Water Quality and
Analysis of Hazardous Waste Sites, Marcel Dekker Inc., In Press.
4.) Capability of EPA Methods 624 and 625 to Measure Appendix IX
Compounds, American Petroleum Institute, Publication Number 4454,
January 1987.
5.) Budde, W.L., Bellar, T.A., Behymer, T.D., Analytical Chemistry, 1990,
62, 1686-1690.
6.) Winkler, P., et.al., "The Effect of Sample Concentration on PBI
Response", Finnigan Technical Report.
7.) Brown, M.A., et.al., Analytical Chemistry 1991, 63, 819-823.
8.) Parr, J.L., et.al., "Improving The Pesticide Method: A Laboratory
Perspective", USEPA Analytical Methods Caucus, March, 1991
9.) Bellar, T.A., Behymer, T.D., Budde, W.L., "Investigation of Enhanced
Ion Abundances from a Carrier Process in High Performance Liquid
Chromatography Particle Beam Mass Spectrometry", J. Am. Soc. Mass
Spectrometry, 1990,1,92-98.
10.) Perry, L., "Effect of Mobile Phase Additives on Liearity in Particle
Beam LC/MS", Pittsburgh Conference, March 1991.
11-65
-------
THE IMPLEMENTATION OF HPLC/POST-COLUMN TECHNIQUES
FOR RUGGED CARBAMATE AND GLYPHOSATE ANALYSIS
Michael V. Pickering, Ph. D. Pickering Laboratories, 1951 Colony
Street, Mountain View, California 94043; Michael W. Dong, Ph. D.,
The Perkin-Elmer Corporation, 761 Main Avenue, Norwalk,
Connecticut 06859-0250
ABSTRACT
This paper provides an overview of the fundamental concepts of
post-column derivatization techniques used in EPA Methods 531.1
(carbamates) and Method 547 (glyphosate). Problem areas for
their practical implementation are described. Specific solutions
leading to a more reliable analysis are discussed.
INTRODUCTION
Carbamates are broad spectrum pesticides which exhibit strong
cholinergic effects on insects. Their low soil persistence and
phytotoxicity, make them a favorite for food crop applications.
The recent discovery of aldicarb (Temik ) in the ground waters
of agricultural regions has prompted the U.S. Environmental
Agency (U. S. EPA) and other agencies to regulate pesticide use
and require routine monitoring of drinking water and raw source
water. The recommended HPLC analytical method (EPA method 531.1
for drinking water and method 8318 for solid wastes) is based on
a 2-stage post-column reaction followed by fluorescence
detection. Carbamates are hydrolyzed at elevated temperatures by
sodium hydroxide to provide methylamine, which subsequently
reacts with o-phthalaldehyde (OPA) and 2-mercaptoethnaol (MCE) at
a high pH to produce a highly fluorescent isoindole. This
technique has excellent sensitivity and selectivity to allow
direct injection of drinking water samples without sample
enrichment or cleanup.
TM
Glyphosate (N-(phosphono-methyl)-glycine) or Roundup ) is a
nonselective herbicide commonly used in post-harvest application.
Maximum residue tolerance limits for glyphosate and its
metabolite aminomethylphosponic acid (AMPA) in various food crops
vary widely from 0.1 to 15 mg/kg. Glyphosate is a trivalent
negative anion under neutral pHs (pK. =2.3) , though it can be
analyzed by cation exchange chromatography under acidic pHs.
Analysis of glyphosate according to EPA method 547 utilizes the
same HPLC post-column equipment used in carbamate analysis.
1-66
-------
Hypochlorite is used as the first post-column reagent to oxidize
glyphosate into glycine which is subsequently reacted with OPA to
form a fluorophore.
IMPROVED HPLC/POST-COLUMN TECHNIQUE
The practical implementation of several improvements to enhance
method performance and ruggedness is discussed. For carbamate
analysis, which utilizes a 0.05 N sodium hydroxide hydrolysis
reagent, the prevention of backflow of this reagent into the
silica-based analytical column is critical (1). For glyphosate
analysis, the replacement of the calcium hypochlorite oxidant
with sodium hypochlorite eliminates reactor blockage problems
(due to the formation of calcium phosphate from the reaction of
calcium ions with phosphate ions of the mobile phase). Also, the
regeneration of the cation exchange column after each analysis
with 5mM potassium hydroxide is necessary to maintain retention
time reproducibility.
For both carbamate and glyphosate assays, the incorporation of
several post-column pressure monitoring points and pressure
relief valves in the system significantly enhances system
reliability by aiding problem diagnostics and preventing rupture
of the heated fluorocarbon reaction coil. The use of guard
columns is mandatory to prolong analytical column lifetime.
Additional sample cleanup (i.e., filtration and solid-phase
extraction) are required for some -water samples and vegetable
extracts. The substitution of volatile 2-mercaptoethanol in the
ortho-phthalaldehyde (OPA) reagent with the nonvolatile N,N-
dimethyl-2-mercaptoethylamine hydrochloride (ThioFluor ) reduces
odor problems in the laboratory. The use of borate salts which
contain high levels of insoluble matter, should be avoided in
preparing the OPA buffer. Boric acid, available in very pure
form, should be used instead and adjusted to pH 10 with sodium
hydroxide. The teflon tubing in the OPA reagent line should be
replaced by Saran tubing which prevents oxygen permeation (causes
OPA degradation). The proper operating sequence for system
start-up and shutdown is also important to prevent possible
reagent precipitation or system damage (2).
REFERENCES
1. M. V. Pickering,"Assembling an HPLC post-column system:
practical considerations," LC.GC, 6, 1988, 994-997.
2. M. W. Dong, F.L. Vandemark, W. M . Reuter and M. V. Pickering,
"Analysis of carbamate pesticides by LC," Amer. Environ. Lab.,
2(3), 1990, 14-27,
11-67
-------
52
^"~ DETERMINATION OF LOW-LEVEL EXPLOSIVE RESIDUES IN WATER BY HPLC:
SOLID-PHASE EXTRACTION VS. SALTING-OUT SOLVENT EXTRACTION
Michael G. Winslow. Manager, Organic Analytical Division, Bradley A. Weichert,
Manager, GC/HPLC Department, and Robert D. Baker, Senior Associate Scientist,
GC/HPLC Department, Analytical Services, Environmental Science & Engineering, Inc.,
P.O. Box 1703, Gainesville, Florida 32602.
ABSTRACT
The December 1990 SW846 draft protocol for the determination of low concentrations
(1-50 u,g/L) of nitroaromatic and nitramine compounds in ground and surface water by
high pressure liquid chromatography (HPLC) (Method 8330) proposes the use of a
salting-out solvent extraction technique using sodium chloride (NaCI) and acetonitrile
(ACN), followed by a Kuderna-Danish extract concentration. This sample preparation
procedure was developed and validated by the U.S. Army Cold Regions Research and
Engineering Laboratory (CRREL) for 8 selected analytes - RDX, 1,3,5-trinitrobenzene
(1,3,5-TNB), 1,3-dinitrobenzene (1,3-DNB), 2,4,6-trinitrotoluene (2,4,6-TNT),
4-amino-2,6-dinitrotoluene (4-Am-2,6-DNT), 2-amino-4,6-dinitrotoluene (2-
Am4,6-DNT), 2,6-dinitrotoluene (2,6-DNT), and 2,4-dinitrotoluene (2,4-DNT).
The adoption of this procedure in Draft Method 8330, which includes six additional
analytes - HMX, nitrobenzene (NB), tetryl, and the 2,3,4-isomers of nitrotoluene (2
-NT, 3-NT, 4-NT)- should be assumed to be applicable only to the eight analytes
validated by CRREL, because its applicability to the six additional analytes is not
supported with experimental data. A discussion of this salting-out solvent procedure
and the results of laboratory analyses applying it to the determination of all fourteen
target analytes are presented.
An alternative sample preparation procedure for the determination in water of low
concentrations of all fourteen nitroaromatic and nitramine compounds listed in Draft
Method 8330 is proposed. This procedure, which has been routinely used to determine
explosive residues in water samples for the Army, uses Porapak R solid sorbent for the
extraction of explosive residues from water samples. A discussion of this solid-phase
procedure and the results of laboratory analyses are presented.
Broad scope applicabilty, ease of use, and cost effectiveness are three factors which
should be considered when adopting an analytical method or procedure for inclusion in
SW-846. In proposing the salting-out extraction procedure in Draft Method 8330 for
the determination of low concentration explosive residues in water samples, these
factors seem to have been neglected. For comparison, the solid-phase extraction
procedure, which does successfully test itself against these three factors, is presented.
INTRODUCTION
Nitroaromatic and nitramine compounds are some the most widely used munitions
components. They have been and continue to be produced in large quantities and are
therefore, along with certain of their degradation products and production impurites,
subject to environmental regulation. The primary concern has been the contamination
II-68
-------
of ground and surface waters near ballistic test ranges and munitions processing and
storage facilities. In recent years, the EPA has issued health advisories on several of
these compounds in drinking water (2,6-DNT 2,4-DNT, TNT, RDX). The result has
been the need for an analytical method that can routinely acheive detection limits of 1
and less for the majority of the compounds of interest.
Gas chromatographic (GC) methods have been used effectively to detect nitrated
munitions components with excellent sensitivity and selectivity, especially when
employing the electon capture detector. However, these methods have been applicable to
only a limited number of target analytes, for various reasons: partial or complete
degradation of thermally labile species; loss due to volatilization of some species during
extract concentration; the difficulty in selecting a single organic extraction solvent.
The use of HPLC with UV detection has become the preferred method for the analysis of
wide range of munitions compounds. In order to achieve detection limits less than 1
|ig/L in water samples, sample concentration prior to HPLC analysis is required.
SW846 Draft Method 8330 has proposed a salting-out solvent extraction procedure
using NaCI and ACN. It is the intent of this paper to suggest that a solid sorbent
extraction procedure using the hydrophilic resin Porapak R warrants strong
consideration as the preferred extraction procedure for incorporation into Method
8330. Tests employing both procedures are described below, and the results are
presented for comparison and discussion.
EXPERIMENTAL
Analytical standards were prepared from Standard Analytical Reference Materials
(SARMs) obtained from the U.S. Army Toxic and Hazardous Materials Agency
(USATHAMA), except for the three nitrotoluenes and the surrogate 3,4-DNT which were
obtained from Aid rich Chemical Co., and the two amino-dinitrotoluenes which were
obtained from the Naval Surface Weapons Center (NSWC). For each of the two
extraction procedures, ten 500-mL samples were prepared at ten concentration levels
on four consecutive days - a total of 40 samples per procedure were analyzed. The
laboratory samples were prepared in ASTM Type II/HPLC grade water. Target analytes
were spiked at the levels indicated in Table 1. Concentration level x represents the
target limit of detection. In addition, each of the laboratory samples was spiked with a
surrogate compound, 3,4-dinitrotoluene, at approximately 5 u.g/L. Calibration
standards were prepared in 30% ACN/70% H20 so that the UV responses of each of the
target analytes would bracket the predicted responses of the target analytes in the final
extracts of the spiked laboratory samples.
Salting-Out Solvent Extraction Procedure
A 400-mL aliquot of water sample was placed into a 500-mL separatory funnel and
shaken vigorously with 130 g of NaCI until the NaCI was completely dissolved. 100 ml
of acetonitrile (ACN) was added to the separatory funnel and the contents were shaken
for 5 min. The phases were then allowed to separate for 30 min. The lower water layer
was then drained off and discarded. The upper layer (~23 ml) was collected in a 25-mL
Kuderna-Danish (K-D) receiver. The separatory funnel was rinsed with 5 mL of ACN
and the rinsate was added to the extract in the receiver. (If the ACN extract is turbid, it
I-69
-------
should be transferred to a 40-mL centrifuge tube with teflon-lined screw cap and
centrifuged at 4000 rpm for 5 min. The ACN layer is then removed with a pasteur
pipette to the 25 mL K-D receiver.) The receiver was then fitted with a a micro (40
mL) K-D flask and modified two-ball micro snyder column. The ACN extract was
reduced to less than 1.0 mL and brought to the 1-mL mark with ACN. The extract was
then diluted to a final volume of 4 mL with ASTM Type II/HPLC water and filtered
through a 0.45 uM Teflon filter. The first 0.5 mL was discarded. The remaining filtrate
was then ready for HPLC analysis. The salting-out solvent extracture procedure is
summarized in Table 2.
Solid-Phase Extraction Procedure
An empty 6-mL Baker Disposable Extraction Column with a 20-p.M frit at the bottom
was packed with 0.5 g of cleaned 80-100 mesh Porapak R. Another frit was placed at
the top of the sorbent bed to assist packing and help prevent channeling. The column was
first conditioned with 15 mL of ACN followed by 30 mL of ASTM Type II/HPLC water. A
500-mL aliquot of water sample was passed through the column at 10 mL/min. utilizing
a Visiprep Solid-Phase Extraction Vacuum Manifold (Supelco).
Figure 1 plots the results of an experiment to determine the optimum sample flow rate
through the extraction system. Five representative target analytes were spiked at 40-
50 ug/L into 500 mL of ASTM Type II/HPLC water. Duplicate samples were extracted at
five different flow rates ranging from 2 to 50 mL/min.
This system allows 12 samples to be processed simultaneously in about 50 min. The
sorbent column was then eluted with 3 mL of ACN at <. 3 mL/min. into a graduated
centrifuge tube. The ACN eluent was concentrated to 2 mL under a gentle stream of
nitrogen. The eluent was diluted to a final volume of 6 mL with ASTM Type II/HPLC
water prior to HPLC analysis. Table 2 summarizes the solid-phase extraction
procedure.
HPLC Analysis
A Shimadzu model LC-6A high-pressure liquid chromatograph (HPLC) equipped with a
Shimadzu SPD-6A autosampler and a Kratos 757 variable wavelength ultraviolet
absorbance (UV) detector set at 250 nanometers was used for analysis of the ACN/H20
extracts. The samples were eluted from a 25 cm x 4.6 mm I.D. Phenomenex ODS (5-u.M
particle size) reverse-phase column. Analyses were performed isocratically with a 55
% methanol/45 % H2O (V/V) mobile phase at a 0.8 mL/min. flow rate. Analytical runs
lasted 30 min., the last target compound eluting at about 25 min. Refer to Figure 3, 4,
and 5 for calibration standard and spike sample chromatograms. The injection volume
was 500 uL. The instrument operating conditions are summarized in Table 2. Data was
collected and quantitated using a Nelson 2700 Turbochrom data system.
Calculations
The percent recovery for each target analyte in the spiked water samples was calculated
by comparison of the calculated concentrations to the target concentrations. The
calculated concentrations were obtained from the initial calibration quadratic regression
II-70
-------
equations. Calibration standards were analyzed at a minimum of five concentration
levels with responses that bracketed the responses of the samples. The lower limits of
detection for each procedure were determined by the Certified Reporting Limit (CRL)
test used by USATHAMA (1990) rather than by the Method Detection Limit (MDL) test
outlined by EPA (Federal Register 1984). For an excellent comparison of the two tests,
see Reference 2.
RESULTS AND DISCUSSION
The accuracy and precision data for both sample preparation procedures are presented
in Table 3. Percent recovery outliers were eliminated from the calculation of the mean
percent recovery of each target analyte. Figure 2 graphs a comparison of the mean
percent recoveries for each procedure. Table 4 lists the CRLs for each procedure.
For both procedures, the mean percent recoveries exceeded 70% for all analytes.
However, the overall accuracy of the solid-phase procedure was significantly greater
than that of the salting-out solvent procedure. The average percent recovery of the solid-
phase procedure (94.3) exceeded that of the salting-out solvent procedure (84.6) by
nearly 10%. The overall precision of both procedures were very similar, although the
average standard deviation of the solid-phase procedure (10.3) was nearly one percent
point lower than that of the salting-out solvent procedure (11.2). As a result of these
differences in accuracy and precision, for all analytes except 2,4,6-TNT, the calculated
CRLs were significantly lower for the solid-phase procedure.
From a comparison of the test results, it is clear that the solid-phase procedure, when
applied to laboratory water samples spiked with the 14 nitroaromatic and nitramine
munitions compounds listed in SW846 Draft Method 8330, performs better than the
salting-out solvent procedure proposed in the method. But this is not surprising,
considering the complexity and labor intensive nature of the salting-out procedure. It
required 16 hours using the salting-out procedure to prepare a batch of 20 samples for
HPLC analysis. It required only 6 hours using the solid-phase procedure. Further, it
was quite surprising that the test results for the salting-out procedure were as good as
they were. This would not be predicted for a procedure that applies considerable
amounts of heat to a group of target analytes containing species that are either very
thermally labile, such as tetryl, or quite volatile, such as the nitrobenzenes.
In light of the above, it is strongly recommended that the solid-phase sample
preparation procedure using Porapak R be given consideration for adoption in SW846
Method 8330 as the sample preparation procedure for determination of low
concentrations of nitroaromatic and nitramine compounds in water.
SUMMARY
For the determination of low concentrations of fourteen nitroaromatic and nitramine
compounds in water, SW846 Draft Method 8330 (Revision 1, December 1990)
proposes a salting-out solvent extraction procedure using sodium chloride and
acetonitrile. An alternative procedure, which employs the solid sorbent Porapak R for
the extraction process, is presented. Both extraction procedures are tested on spiked
water samples. The sample extracts are analyzed by HPLC with UV detection. The
1-71
-------
analytical results from both extraction procedures are presented for comparison. The
advantages for inclusion of the solid-phase extraction procedure into SW846 Method
8330 are discussed.
ACKNOWLEDGEMENTS
We thank P. Dumas, D. Dabney, and S. McMillen for their laboratory assistance.
REFERENCES
1. Environmental Science & Engineering, Inc., "Development of Standard Analytical
Methods for the Analysis of Explosives in Water by High Pressure Liquid
Chromatography Precertification/Certification Report and Method Writeup", March
1991, Contract No. DAAA15-90-D005, U.S. Army Toxic and Hazardous Materials
Agency, Aberdeen Proving Ground, Maryland.
2. C.L Grant, A.D. Hewitt, and T.F. Jenkins, "Experimental Comparison of EPA and
USATHAMA Detection and Quantitation Capability Estimators". American Laboratory.
15-33, February, 1991.
3. B. Lesnik, "The HPLC Methods Development Program: An Overview", Environmental
Lab. 18-41, April/May, 1990.
4. M.P. Maskarinec, D.L. Manning, R.W. Harvey, W.H. Griest and B.A. Tomkins,
"Determination of Munitions Components in Water by Resin Adsorption and High-
Performance liquid Chromatography-Electrochemical Detection," Journal of
Chromatography. 302, 51-63, 1984.
5. P.H. Miyares and T.F. Jenkins, "Salting-Out Solvent Extraction Method for
Determining Low Levels of Nitroaromatics and Nitramines in Water", Special Report
90-30, U.S. Army Corps of Engineers, Cold Region Research & Engineering
Laboratory, 1990.
6. SW846 Draft Method 8330, "Nitroaromatics and Nitramines by High Pressure
Liquid Chromatography (HPLC)", Revision 1, December 1990, U.S Environmental
Protection Agency, Office of Solid Waste and Emergency Response, Washington, D.C.
7. USATHAMA, U.S. Army Toxic and Hazardous Materials Agency, Quality Assurance
Program, January 1990.
II-72
-------
Table 1 TARGET ANALYTE CONCENTRATIONS fag/L)
J
H0C
~fv '
ซK '- ,
1,3^TNฃ
1f^f>N8
Teiryj
"; .
iป *;",\~ ^
ฃ,4>TNT
% f . .
t : :
4ปAJR*&&8NT
sj
ฃปAffi *2*443i NT
^DNT .> .
: ^
2J4HDHT ^
t^T , \ '
4-Mf
3-NT
OX
0
0
0
0
0
0
0
0
0
0
0
0
0
0
JX ,
0.302
0.292
0.148
0.138
1.08
0.645
0.28
0.052
0.055
0.061
0.053
0.307
0.301
0.292
X. ;
0.604
0.584
0.296
0.275
2.15
1.29
0.56
0.104
0.11
0.122
0.106
0.613
0.602
0.584
2X
1.21
1.17
0.592
0.55
4.3
2.58
1.12
0.208
0.22
0.244
0.212
1.23
1.2
1.17
6X
3.02
2.92
1.48
1.38
10.8
6.45
2.8
0.52
0.55
0.61
0.53
3.07
3.01
2.92
10K
6.04
5.84
2.96
2.75
21.5
12.9
5.6
1.04
1.1
1.22
1.06
6.13
6.02
5.84
aox
12.1
11.7
5.92
5.5
43
25.8
11.2
2.08
2.2
2.44
2.12
12.3
12
11.7
$OX i
30.2
29.2
14.8
13.8
108
64.5
28
5.2
5.5
6.1
5.3
30.7
30.1
29.2
t&OX
60.4
58.4
29.6
27.5
215
129
56
10.4
11
12.2
10.6
61.3
60.2
58.4
200X
121
117
59.2
55
430
258
112
20.8
22
24.4
21.2
123
120
117
II-73
-------
Table 2 SUMMARY OF EXTRACTION PROCEDURES AND CHROMATOGRAPHIC
CONDITIONS
SALTING-OUT SOLVENT EXTRACTION
400 ml water sample in a 500 mL separately funnel.
Add and dissolve 130 g of NaCI.
Add 100 mL ACN and shake for 5 min.
Let phases separate for 30 min.
Discard lower water layer; recover ACN; (-23 mL) in a 40 mL vial.
Rinse separatory funnel with 5 mL ACN; recover in the 40 mL vial.
If ACN extract is turbid, centrifuge at 4000 rpm for 5 min.
Remove ACN layer with a pasteur pipette to a K-D evaporator.
Reduce to < 1.0 mL and bring to 1.0 mL with ACN.
Dilute with 3.0 mL reagent water to 4.0 mL final volume.
Filter through 0.45 u.M Teflon filter; discard first 0.5 mL.
Analyze by RP-HPLC/UV.
SOLID-PHASE EXTRACTION
6 mL Disposable Extraction Column packed with 0.5 g of cleaned 80-100mesh
Porapak R.
Precondition column first with 15 mL ACN and then with 30 mL ASTM Type
II/HPLC water.
Measure 500 mL of water sample and pass through column at 10ml_/min.
Elute column with 3 mL ACN at < 3 mL/min. into a graduated centrifuge tube.
Concentrate eluent to 2 mL under a gentle stream of nitrogen.
Dilute to 6 mL final volume with ASTM Type II/HPLC water.
Analyze by RP-HPLC/UV.
CHROMATOGRAPHIC CONDITIONS
Column: Phenomenex ODS reverse phase HPLC column, 25-cm x 4.6-mm, 5-
Mobile Phase: Isocratic, 55% methanol/45% water (V/V).
Flow Rate: 0.8 mL/min.
Injection Volume: 500 uL.
UV Detector: 250 nm.
I-74
-------
Tables PRECISION AND ACCURACY DATA
Salting-Out
Solvent Extraction
IftK
mx
t^TNB
1,3-DNB
tEraYL . .
MB ....
3,4-DNT (SUR)
2t4h6"TMT
4'Attf-2,e*DNT
2'AMซ4>6ปDNT
2,$*ONT
2,4*ONT
2*NT
4-MT
3-NT
Solid-Phase
Extraction
HVIX
BOX
tf%5"TNB
1,3-DNB
TEmYL
N8
3,4-DNT (SUR)
2^eปTNT
4*AM'2,e'DNT
2*Ay*4,e*D*JT
2,$>DNT
2,4'ONf
2-NT
4-NT
3ปNt
MEW
$RBฃ
91.8
86.7
70.8
83.9
80.1
81.5
aus
89.4
126
78.4
83.6
76.7
76.6
75.1
77.3
STDDEV
5.5
6.2
18.6
8.6
14.5
9.1
12.5
14.1
19.1
11.2
12.8
9.9
7.3
7.1
11.1
RANGE
84.2 - 106
67.2 - 98.6
44.3 - 98.3
64.5 - 103
46.5 - 103
67.8 - 109
70.0 - 121
64.5 - 123
101 - 167
54.9 - 102
58.6 - 117
56.4 - 99.2
63.4 - 103
63.2 - 95.7
62.6 - 96.2
K
36
36
36
36
36
34
21
35
19
35
36
35
34
34
35
97.8
95.5
84.7
97.1
91.1
92.7
102
96.8
123
92.2
90.3
85.2
91.9
84.1
90.8
4.1
6.6
16
6.5
11.6
7
8.3
12.6
18.3
12.4
11
9.1
8
10
12.6
87.9 - 106
81.3 - 109
54.5 - 105
78.8 - 112
73.6 - 120
77.4 - 108
69,4 .- 125
75.4 - 123
96.8 - 164
68.2 - 117
64.4 - 108
66.2 - 101
67.7 - 106
63.6 - 96.3
69.4 - 129
36
36
35
36
36
36
40
35
35
35
35
36
34
36
32
11-75
-------
Table 4 CERTIFIED REPORTING LIMITS (UG/L)
HMX
FOt
1^5-TNB
1>DMB
TEm^L
NITROBENZENE
2t4^7#T
4*AM-2re>I>NT
2-AM-4>e'DNT,
2,6-DMT
2,4*DNT
2ปNT ^ ^
4*NT .
3*NT
$OOD^PHA$E
0.3
0.29
0.45
0.15
2.49
0.65
0.64
1.57
0.16
0.074
0.064
0.41
0.62
1.4
SALTiNO-OUT
BOLVENFF
0.45
0.64
0.75
0.38
4.07
4.4
0.57
3.98
0.86
0.123
0.088
2.11
2.07
2.03
II-76
-------
100
95
90
%Rec 85
80
75
70
Figure 1 PERCENT RECOVERY vs FLOW RATE
4 1 0
mL/min
20
RDX
-*- NB
O TNT
-* 2,4 DNT
50
-------
Figure 2 PERCENT RECOVERY COMPARISON
si
CXI
50
SOLID PHASE
SALTING OUT
60
70
80
90
%Rec
100
110
120
130
-------
FileNane : D:\2700\DATA\HTRC012.raw
Start Tiie : 0.00 lin End Tile : 30.00 lin
Scale Factor: 0 Plot Offset: 0 งV
Date : 5/16/9! 2:58 PM
Low Point : -0.00 nV
Plot Scale: 100 iV
Hioli Point
Page 1 of 1
: 100.00 iV
CNซO
>
E
D
U)
C
o
Q.
V)
0>
ntf/mL
1,3,5-TNB
1,3-DNB
TETRYL
NB
3,4-DNT
2,4,6-TNT
4-Am-2,6-
2-Am-4,6-
2,6-DNT
2,4-DNT
2-NT
4-NT
3-NT
1
X
I
I 1 I 1 1 !
1 5
i
X
o
a:
1 1
I
CD
t
in
n
i i i
UJ
ISJ
1-,!
T-~}^ 2
~TTm7~r
Retention
1 1 1 1 1 1 i
CN i" ^
H- 1 1 P
ZT Z Z Z Z ฃฃ
O 1 23 OO f-:
rOCN? -^ CMCNCNI CM
1 1 1 1 1 1 1 1
15 20
Time [min]
i
ir
z
I
1
I
0
cr
i
' 2J5 ' ' ' '
Figure 3 CHROMATOGRAM OF CALIBRATION STANDARD
1-79
-------
C h
Hb <: g -r ฃ
FileNaie : D:\2700\DATA\SALT021.rav
Start Tiie : 0.00 tin End Tiie ; 30.00 tin
Scale Factor: 0 Plot Offset: 0 ซV
Date : 5/16/91 3:26 PH
lav Point : -0.01 iV
Plot Scale: 50 iV
High Point
Page 1 of 1
: 49.99 iV
45-
>
CD
CO
C
O
a
t/i
(D
QL
r- CM
1
X
2
X
1 1 1 I 1 1
b
I I
m
z.
\
x in-
o K!
ce ซ-
t
m
^
TEJRYL
NITROBENZE -
1 1 1 1 1 1 1 1
Retention
i i
1:
KTCM
iiit
CM <*-"
1 1
O O t t
^ =500
< < ' 1
1 I <ฐ*.
M- CM CMCM
I ' '
Time
1 I
2
[mi
I
n
l
]
2-NITROTOL -
1
i
o
5
a:
5E
\
I
o
s
a:
7Z.
1
to
1 I 1 1 1 1
25
Figure 4 CHROMATOGRAM OF SALTING-OUT EXTRACTION SAMPLE
1-80
-------
FileNaw I Ds\2700\DATA\HTRC039.ran
Start Tite : 0.00 tin End Tiie : 30.00 nin
Scale Factor: 0 Plot Offset: 0 iV
Date : 5/16/91 3:06 PH
Low Point : -0.00 iV
Plot Scale: 50 tV
High Point
Page 1 of 1
: 50.00 iV
45
4O-
35-
30-
QJ
CO
C
O
Q.
V)
0)
o:
2O-
15-
10-
oooo
I I 1 1
lปป
I
co CM
10 r^_
oC) O'
I I
IO- U"> U")CN CM CO CO ID
oq^r ro ojcq CD in *ป;ซ-;
I'" "~ "~"~ ~~ "~ ~~~
I ill I II
O
7
7
a
CT
Compound
HMX
RDX
1,3,5-TNB
1,3-DNB
TETRYL
NB
3,4-DNT
2,4,6-TNT
4-Am-2,6-DNT
2-Am-4,6-DNT
2,6-DNT
2,4-DNT
2-NT
4-NT
3-NT
UQ/L
3.02
2.92
1 .48
1 .38
10.8
6.45
4.94
2.80
0.52
0.55
0.61
0.53
3.07
3.01
2.92
1
X
^
X
1 1 1 1 1 1
ct
1
X
O
tt
1 1
1
m
z
H~
**!
*^
T-
' 1
1 1 1
UJ
INJ
LU , ,
"^^ 111
ง ' Q-1
1 ttf CK
^i fTj i
' t- Z
III!
I I I
ป
cst
1- 1
PZ o
d^.' . Z
o ^ ^
1 <;
"^"^ |
r^CM *
1 1
1 1 1
*
1
OH- i-
"Z. "Z. "Z.
^K ฃ$ C)
< 1 1
1 _ .
CM CM CM
1 1 1
1
O
c5
cr
z
i
CM
1 ' '
1
O
5
cr
t:
z
i
^
1
i
o
5
o:
t=
z
i
o
Retention Time [min]
Figure 5 CHROMATOGRAM OF SOLID-PHASE EXTRACTION SAMPLE
11-81
-------
53 REDUCTION OF AZO DYES TO AROMATIC AMINES FOR ENVIRONMENTAL MONITORING
Robert D. Voyksner and Jeffrey T. Keever, Analytical and Chemical Science,
Research Triangle Institute, P. 0. Box 12194, RTP, NC 27709
Harold S. Freeman and W. N. Hsu, North Carolina State Uuniversity,
Box 7003, Raleigh, NC 27695
Leon D. Betowski, EMSL US-EPA, Las Vegas, NV 89193-3478
Azo dyes are of great environmental concern due to their potential to form
carcinogenic aromatic amines under reducing conditions. As a result, it is
necessary to evaluate both the intact molecule and its potential reductive
cleavage products to adequately assess the potential risk of a dye stuff to the
health of man and the environment. With over 100 million pounds of azo dyes
produced annually, the development of a method that effects the reductive
cleavage products in vitro and permits their characterization would aid in
determining modern complex and structurally unknown dyes and their genotoxicity.
A logical approach would involve the evaluation of procedures for the
reductive cleavage of azo dyes followed by mass spectroscopic (MS) analysis of
their products. To best determine the approach in achieving this goal, the
reduction of representative samples from several of the major azo dye classes
(e.g. disperse and solvent dyes), was accomplished using chemical means. Two
chemical procedures were evaluated for the reduction of azo dyes. The first
reduction agent, SnCU, is especially important in the reductive cleavage of azo
linkages in the presence of other easily reduced groups such as a nitro group.
The second method involved sodium dithionate, Na2S20., which has been used to
effect the reductive cleavage of water soluble azo dyes for decolorizing
purposes and for Salmonella bacterial assays for mutagenicity.
Initial screening of the various reduction products from each procedure was
accomplished using thin layer chromatography (TLC). Confirmation of the
postulated reduction products involved a combination of particle beam and
thermospray LC/MS and GC/MS. Standards of the proposed reduction products and
11-82
-------
aromatic amines of the products tentatively identified, when available, were
employed to confirm identities. It was observed that the chemical reduction
methods resulted in nearly 100% reduction of the azo bond to form the
characteristic amines for the 16 dye standards evaluated. Overall the SnCl2
method was a more powerful reducing agent yielding a greater number of products.
In addition to the reduction of the azo bond, dyes containing acetate groups
exhibited both acid and base catalyzed hydrolysis of the ester groups to form
the respective alcohols. The presence of electron withdrawing halogen groups on
the aromatic ring appear to make the nitro groups more susceptible to reduction.
Also small yields of N-dealkylation products were observed for SnClp reduction
of some dyes.
Current efforts are evaluating the use of chemical reductions for
determining aromatic amine content of wastewater, sludge and sediment
contaminated with azo dyes. Reduction conditions using SnCl2 or Na2S204 needed
to be modified to insure complete reduction of azo dyes in wastewater and
sludge. Variables, including the presence of sediment in a sample, temperature,
reaction time and amount of reductant, influence the yield of the aromatic
amines. The analysis of the reduced azo dye samples provided an estimate of
total dye (amine) content, but identity of specific azo dye could not be
determined. Most mono and diamino reduction products could be analyzed by
GC/MS. More polar reduction products containing three amine groups or multiple
function groups (e.g. SO,, OH, NO,,) were best analyzed by LC/MS.
O ฃ
Although the information described in this article has been funded wholly or
in part by the Environmental Protection Agency under contract 68-02-4544 to
Research Triangle Institute, it does not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
1-83
-------
54 HAZARDOUS WASTE COMPONENT IDENTIFICATION
USING AUTOMATED COMBINED GC/FTIR/MS
Roger J. Leibrand, Scientific Instruments Division,
Hewlett-Packard Co, 1601 California Avenue, Palo Alto, CA
94304
INTRODUCTION
Our modern industrial civilization contains many items which
have the potential to produce hazardous waste in their
production or use. In general, plastics, paints, petroleum,
petrochemicals, leather, textiles, pesticides, and medicines
can generate volatile solvents, reaction residues, oils, and
non-volatile heavy metals, dyes, pigments, salts, acids, and
caustics. To properly dispose of these hazardous wastes is
a complex problem. The first step in this process is
determining what is there, i.e., chemical analysis of the
waste sample. The scheme for identification and
quantitation of certain specific toxic compounds is quite
well delineated in the U.S. Environmental Protection Agency
methodology. However, tentatively identified compounds
(T.I.C.s) and other hazardous components that are non-target
compounds in the E.P.A. methods are not accurately
determined by those methods. These contaminants need to be
identified before proper disposal of hazardous waste can
occur or a waste site can be remediated.
Qualitative analysis of organic pollutants currently relies
heavily on the mass spectrometer, both in its GC/MS and
LC/MS forms. The mass spectrometer produces powerful
structural information based on molecular fragmentation,
often including molecular weight data. It is weak in the
areas of aromatic substitution, isomer differentiation, ring
junctions, alcohol identification, and functional group
classification. Fortunately, the infrared spectrometer is
strong in these areas. The E.P.A. has recognized this and
has developed Method 8410 for the GC/FT-IR analysis of
semivolatile organics (1). Combining GC/MS and GC/FT-IR
provides a higher confidence result than either one alone
(2). Combining the two into one doubly hyphenated system
with powerful automated computing capability provides the
analyst with an efficient system for analyzing hazardous
waste. An example of that is shown in this paper, utilizing
a sample from an actual hazardous waste drum.
HARDWARE
The gas chromatograph, an HP 5890A, was set up using the
common, most routinely used column for environmental
1-84
-------
screening, a 25 meter 5% phenyl methyl silicone (HP-5) under
standard analytical operating parameters. The column
effluent was split at the end of the column at a 10 to 1
ratio with the bulk of the flow going to the HP 5965B IRD
and the lesser amount to the HP 5970B MSD. The details of
this parallel flow configuration are described elsewhere
(3).
SOFTWARE
One of the characteristics of environmental samples in
general and hazardous waste samples in particular is that
they contain many components. The sample used in this paper
contains many dozens of compounds, the specific number
analyzed for is determined only by the instruments'
sensitivities and operational parameters. Figure 1 is the
Total Ion Chromatogram (TIC) from the MSD and the Total
Response Chromatogram (TRC) from the IRD. Easily well over
100 components are detectable in each Chromatogram, but for
illustration, the integration threshold was set to allow
only about 50 to be analyzed. Those peaks are shown in the
integrated chromatograms of Figure 2.
In order to aid the chemist in characterizing
multicomponent, complex chromatograms, the standard software
of the HP 5965B includes the Macro Program 'Qualrpt'. This
automated software routine performs a qualitative analysis,
i.e., library searchs on previously integrated
chromatograms. This Qualrpt macro works on the integrated
Total Response Chromatogram (TRC) from the IRD and/or the
Total Ion Chromatogram (TIC) from the MSD. The Qualrpt
output for individual IRD and MSD data consists of a
tabulation of the GC peaks found (times , areas, etc.), a
Chromatogram, and library search results. When the data
from the IRD and MSD are combined in Data Editor, the
Qualrpt macro produces a tabulation of the peaks found, a
combined chroraatogram, and combined list of library search
results for each peak. The combined search list is merged
by common CAS Registry numbers into three categories. Class
1 contains those entries which are on both lists. When
comparing IRD and MSD library searches, these entries, (or
more often, this entry) have a high probability of correctly
identifying unknowns. Class 2 contains those entries which
are only on one hit list because they exist in only one of
the two libraries searched. Class 3 consists of those
entries which are in both libraries but nonetheless showed
up in only one of the two hit lists. Isomers, because of
their nearly identical spectra often appear only on the MSD
hit list while homologous series because of their spectral
similarities only appear on the IRD hit list. If the IR
spectrum of a specific unknown compound is not in the
library, typically the near misses are of the same chemical
1-85
-------
class. This is a very powerful feature of infrared
spectroscopy which is favorably exploited in the IRD Qualrpt
software.
Once the appropriate libraries are selected, in this case
the 49000 entry NIST/NBS library of mass spectra and the
3000 entry EPA Vapor Phase Infrared library, Qualrpt can be
initiated. Several examples have been selected from the
Qualrpt combined search results for discussion.
RESULTS
The sample was from a waste drum presumed to contain paint
and perhaps other hazardous solvent materials. 50 peaks
were integrated in the TRC and the TIC. The chromatograms
are not totally identical. There are some differences, air
and water peaks in the TRC, and a few more small components
in the TIC. There were 41 common peaks found and 23 Class 1
hits. Some peaks were found in the TIC only and some in the
TRC only. Differing detector response factors account for
this fact. Carbon dioxide (from air), water (from the
sample and/or air), and the solvent methylene chloride are
early components found by the IRD. In addition, acetone and
methanol were found and elute before the solvent raethylene
chloride.
A typical example of a Class 1 hit is shown in Figure 3.
The PBM mass spectrometry search and the IRD search both
indicate that the compound is ethyl acetate. Note that the
highest quality hits are in Class 1. All of the lesser
quality IRD matches are acetates. Infrared spectroscopy is
very good at functional group/compound class
differentiation. IR and MS confirm each other in this case
for a very high confidence result.
Figure 4 shows an example where the top quality PBM hit is
not confirmed by the IRD. In this case the top IRD hit is
ethyl benzene and the top MSD hit is meta xylene. As can be
seen from the Class 3 listing, all three xylenes are in the
EPA IR library but their spectra are so different they don't
appear on the hit list while the spectrally similar ethyl,
propyl, and butyl benzenes do. Clearly the infrared
information confirms that ethyl benzene is the correct
assignment.
Figure 5 is an example where the top IR hit is not confirmed
by PBM. Here the top quality IR hit is butyl benzene and
the top MSD hit is propyl benzene. The likely molecular ion
at m/z 120 is very important information pointing toward the
assignment of propyl benzene. Note that all the listed IR
hits are alkyl benzenes.
11-86
-------
Since the NIST/NBS MS library is more than sixteen times
larger the than EPA IR library it is obvious many MS hits
cannot be confirmed by IR. Most of the time in this case of
Class 2 hits, however, the chemical class and isomeric
configuration is confirmed. Figure 6 is an example of this
situation where the spectra of the ethyl methyl benzenes are
not in the EPA Library but the closest IR hits are mostly
all 1,4-disubstituted benzenes. This indicates that the
best qualitative assignment is l-ethyl-4-methyl benzene, not
the 1,2- or 1,3- isomers.
Table 1 is a condensation of the Qualrpt combined search
results. It can be seen there are 56 total peaks reported,
41 common ones, and 23 Class 1 hits. The compound
assignments when no Class 1 hit was found is the most likely
based on further examination of the spectra. In some cases
the molecular ion was of some use. Clearly use of the
largest MS and IR libraries available would improve the
searching and is the subject of future work.
CONCLUSION
The combined technique of GC/FT-IR/MS using the HP 5890A, HP
5965B, and HP 5970B with the Qualrpt automated spectral
output and combined library searching has been shown to be
very useful in the rapid high confidence qualitative
analysis of hazardous waste components.
11-87
-------
REFERENCES
1. Method 8410, U.S.E.P.A., Rev. 1, 1990
2. Gurka, D. F. and Pyle, S.M., QUALITATIVE AND
QUANTITATIVE ANALYSIS BY CAPILLARY COLUMN
CHROMATOGRAPHY/LIGHTPIPE FOURIER TRANSFORM SPECTROMETRY,
Environ. Sci. Technol. 1988, 22, 963-967
3. Leibrand, R. J. and Duncan, W. P., INVESTIGATION OF
THE CHROMATOGRAPHIC OPTIMIZATION OF COMBINED GC/FT-IR/MS,
Int. Lab., 1989, 46-52
CONDITIONS
Gas Chromatograph Column: 25 m x 0.32 mm id HP-5 (5%
phenylmethyl silicone), 0.52 micrometer film Carrier Gas:
Helium at 10 psi, 2.0 mL/min Oven: 40 C (1.0 min) to 240 C
at 4 C/min Injection Port: 250 C Sample Injection: 2
microliters split 10:1
IRD Parameters Light Pipe: 250 C Transfer Lines: 260 C
Sweep Gas: Nitrogen, 15 psi inlet, 5 psi outlet Scan
Parameters: 8 cm-1 resolution, 2 co-adds, 3 scans/second
stored Detector: Wide band (550 to 4000 cm-1) MCT
MSD Parameters Mass Range: 10 to 310 daltons Scan
Parameters: 2 A/D samples, 1.4 scans/second stored Transfer
Line: 280 C
11-88
-------
CD
CO
TRC of DHTHiMRSTCIRZ.D
ukuwu
T:C of DRTRiWHSTEMSZ.D
JLJL
JL
Figure 1. Total Response Chromatogram (TRC) from IRD and Total Ion Chromatogram (TIC) from MSD of
hazardous waste sample
-------
ป TIC of DF1TR; WRSTEMS2. D
TRC of DfiTfl:WRSTEIR2.D
CD
O
ป
n
I
M IB
20 52
26 23
Figure 2. Total Response Chromatogram (TRC) from IRD and Total Ion Chromatogram (TIC) from MSD of
hazardous waste sample showing integration of peaks used in this brief
-------
Peak 7 of 56.
(3 10:313 -
7.0E>5:
o
o 5.0L>5-
ซ 4.0E+5'
1 3.0ฃ>5-
1 2-0H ,-
\.BfH-5- <
(704:709 -
3 12-
i IB-
o 8-
8 B-
.0 4-
L.
ฐ 2~
K 0-1-+
4000
Both MS and IR found
...) flvg 3.571:4.088 ml n . from DfiTR : WBSTEMSE
43
ซ5
1 . .' / t2t 27t
, 1 / H2 \ ป' 175 187 H7 \
Hl^|li / \ / /// \
50 100 150 200 250
Mass/Charge
...) RSf 3.976:4.004 min. from DRTR: WRSTEIR2
- 5
at II m 1 1 "~
en 1 H
A A JIL. .
3500 3000 2500 2000 1500 1000
Trequency (cm-1)
COMPARISON OF RESULTS FROM
PBM Search of Library file: DATA:NBS43K.L
AVQ 3.971:4.008 nin. fron DATA:UASTEnSZ.O
AND
IR Search of Library file: DATA:EPA_REVA.L
ASP 3.376:4.004 nin. fron DATA:UASTEIR2.D
Class 1 (on both lists)
.D
. D
PBM IR
CK Nunber Dual Oual MUt Fornula Nane
1. OO014I-78-6
64 379 88 C4H802 Acetic acid, ethyl eater
Class 2 (in only one library)
PBM IR
Cfl^ Nunber Oual Oual NUt Fornula Nane
Z. OOOS95-46-0 25 I3Z CSH804 Propanedloic acid, dimthyl-
3. OO0078-98-8 12 72 C3H402 Propanal . 2-oxo-
4. 002203-36-3 9EO ISO C6HI1C10Z ACETIC ACID. 3- CHLOROBUT YL ESTER
5. 006363-44-6 953 188 C9HI604 1 ,5-PENTANEDIOL . DIACETATE
Class 3 (in both libraries, but on only one list)
PBM IR
CAS Nunber Oual Oual MUt Fornula Nane
6. 000628-63-7 952 130 C7HI402 ACETIC ACID. PENTYL ESTER
7. O0443S-53-4 951 146 C7HI403 ACETIC ACID. 3-METHOXYBUTYL
ESTE
Figure 3. Qualrpt output of peak 7, Class 1 result indicating ethyl acetate
11-91
-------
Peak Z3 of S6.
Both HS and IR found
(BG4:874 - ...) Rvg 10.821:10.945 min. from DflTfi: WRSTEMS2 . D
a
0
C
n
T3
c
3
jo
a:
c
g
cr
E
"
a
o
c
a
JO
t.
o
cc
3.
2.
2.
1 .
1.
5.
0.
0E+6:
5E+6:
0E+6-
5E*S-
0E+S-
91
SI
V \. K
\ ,NJ J .1 .
IBC
it? IBI 214 nt 245 as ai
J / Vs ',"/ / / / / \
50 100 150 800 250 300
Mass/Charge
1925:1938 - ... RSP 10.868:10.942 m!n. from DflTfl: WHSTEIR2 . D
B.0:
B.0:
4.0-
2.0:
01
2
CO 1
01 f
01 H "" - * m /
raoi m u JL a ru g I
^jJ^L^J L
4000 3500 3000 2500 2000 1500 1000
Frequency (cm 1)
COnPARlSON OF RESULTS FROM
PBH Search of Library file: DflTfl:NBS49K.L
Avg 10.821:10.945 ซln. fron DrtTfl:UftSTEMSZ.D
AND
IR Search of Library file: DflTfl:EPa_REVfl.L
RSP la.BEB: 18.942 din. froH DflTfl:UftSTEIR2.D
Class I
-------
Peak 34 of 56.
- Both US and IR found
(
C
3
_Q
cc
1IGB:1176 - ... flvg 14.579:14.677 m 1 n . from DRTR: WRSTEMS2 . D
1.0E+6-
8.0E+5-
6.0E>5-
4.0E+5-
2.0E+5-
cs
39 \ 7ซ
(5 / 1 i
' i. li. iJ jl . i ,
91
121
/
ISI 22C 27<
IซI / (77 212 \ /
. .1 . . ' / / / \ /
50 100 150 203 250
Mass/Charge
(2586:2600 - ... RSP 14.600:14.679 mm. from DRTR: WRSTE IR2 . D
oz
E
ffi
O
C
a
c
O
-O
a:
3. 0-
2.5-
2.0-
1 .5"
1 .0-
0.5-
0.0-i
4
$
Jli
s
n 11 <"
i IR , 2
/I "ft" - M
^~" -^ \ A J VA. _ f^. ^ J Ly
4000 3500 3000 2500 2000 1500 1000
Frequency (cm 1)
COMPARISON OF RESULTS FROM
PBM Search of Library file: DATA:NBS49K.L
Avg 14.578:14.677 din. fron DATA:UASTEMSZ.D
AND
IR Search of Library file: DATA:EPA_REVA.L
ASP 14.600:14.679 nin. fron OATA:UASTEIRZ.O
Class
(on both lists)
CAS Nunber
PBM
Oual
IR
Oual
MUt Formula
Nane
I. 800103-65-1 80 941 IZ8 C9H1Z
Benzene, propyl-
CF)S Nunber
PBtl
Qual
Class Z (in only one library)
IR
Oual nut Formula Nane
Z.
3.
4.
5.
6.
7.
8.
8B4I5Z-89-4 64 150 C9H14NZ 1,Z-Ethanedianine. N-(phenylneth
084545-85-I 43 I9Z C7H7C1ZP Phosphonous dichloride. (phenyln
017G34-51-4 33 1Z0 C9H1Z I,3.5-Cycloheptatriene. 7-ethyl-
000620-05-3 32 218 C7H7I Benzene, (iodonethyl)-
084464-74-8 23 184 C8H803S I.Z-Ethanediol. phenyl-. cyclic
00O148-Z8-3 1O --- 240 C16HZ0NZ 1.Z-Ethanedianine. N.N'-bis(phen
000588-67-0 8 --- 164 C1IH160 Benzene, (butoxynethyl)-
Class 3 (in both libraries, but on only one list)
PBM IR
CAS Nunber Oual Oual MUt Formula
Nane
9. 000104-63-Z 64 - 151 C9HI3NO Ethanol. Z-[ (phenylnethyl >aninol
10. 008IZZ-78-1 37 IZ0 C8H80 Benzeneacetaldehyde
II. 000184-51-8 - 953 134 C10HI4 BENZENE. BUTYL
12. 800538-68-1 949 148 CI1H16 BENZENE. PENTYL
13. 881877-16-3 - S4Z I6Z CIZH18 HEXANE. I-PHENYL
14. 8BZ189-60-B 933 190 C14HZZ OCTANE. 1-PHENYL
Figure 5. Qualrpt output of peak 34, Class 1 result indicating propyl
benzene
1-93
-------
Peak 36 of 56.
Both MS and IR found
<
a
u
n
T.
c
D
I
1205:1208 - ... Rvg 1 5 . 05 8 : 1 5 . BSE mm. from DRTfi: WflST EH92 . D
a.5E>6-
2.0E + 6:
l.SEi-G-
1.0E + 6:
5.0E+S-
ป C 77 SI
1 T t , \, i ' ,
15
(II
us n?
i .. 7' T \/
B.BE-t-B"1-1 ' ' i ' ' i i ""
50 100 150 200
Mass/Charge
(2666:2674 - ... RSP 15.051:15.096 mm. from DflTR: WF1STEIR3 . D
Zi
rr
E
U
C
n
X!
t
O
(A
jD
rr
5.0:
4.0:
3.0:
2.0;
1 .0-
I
mซ
ป F -
Is r -
I 1 | -v 0- ft
1 1 ^ ^ II <** nu M
I <* S 1 S is /
/I 2" K- \\
J v^ y..^--' v-^ vx; s^v
4000 3500 3000 3500 2000 1500 1000
Frequency ( cm- 1 )
COMPARISON OF RESULTS FROM
PBn Search of Library file: DflTfl:NBS43K.L
Avg IS.eSB: 15.836 nin. fron DflTfl:UflSTErtSZ.0
AND
IR Search of Library file: WTA:EPn_REVA.L
ASP IS.eSI:IS.83E nin. fron DflTfl:UftSTEIRZ.0
Class 1 (on both lists)
NO COMMON COMPOUNDS FOUND IN SEPARATE REPORTS
1 ibrory)
pen IR
CAS Nunber Dual Oual HUI Fornula
None
z. eeeies-67-8
3. 0ซ3l4!-ซZ-4
4. 895814-85-7
S. 061142-17-4
6. BZ9634-3S-6
7. eซ7Zt4-BI-1
8. (M1B38-B3-9
9. eei484-se-e
la. 0MI3Z-77-B
9,
83
S6
53
58
43
37
zs
t;
937
,3(i
,7<
Z67
1S,
,zซ
27/i
I4H
C9H1Z Benzene, l-ethyl-4-nethyl-
C9HIZ Benzene. I,3.S-trinethyl-
C9H12 1.3-Cyclopentadtene. S-(l-nethyl
CISHI6 Benzene, I.1'--
C9HIZ 2.3-Heptadien-S-yne. 2.4-dinethy
CUHItBrO Ethanone. 2-brono-1 ,2-diphenyl-
CIIH16 BENZENE. 1 .Z-DItlETHYL-4-ISOPROPY
Class 3 (in both libraries, but on only one list)
PBH IR
CAS Number Oual Oual HUI
Fornula
n. ceaeze-u-4
12. eeeen-14-3
13. eซeSZB-73-B
is. mee95-63-6
16. aeeese-es-z
17. eooies-es-s
IB. eeee93-87-E
13. WI632-I6-2
ze.
91
91
47
2S
950
949
935
934
,7ป
,;ซ
,7^
,70
,;ป
|Zซl
134
13ซ
,,/
I4B
C9HIZ Benz ne. 1-ethyl-3-nethyl-
C9HI2 Benz ne. l-eปhyl-2-nethyl-
C9HIZ Benz ne. I.2,3-trinethyl-
C9HI2 Benz ne. (l-nethylethyl)-
C9HI2 Benz ne. 1.Z.4-trinethyl-
C8H80 Ethanone, 1-phenyl-
CiaHU BENZENE. P-OIETHYL
CI8HI4 BENZENE. l-ISOPROPYL-4-METHYL
C8HI6 1-HEXENE. Z-ETHYL
CIIHI6 TOLUENE. P-TERT-BUTYL
Figure 6. Qualrpt output of peak 36, Class 2 and 3 results indicating
l-ethyl-4-methyI benzene
1-94
-------
Peak
#
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
. 53
54
55
56
Retention
Time, TIC
_
2.364
2.964
3.033
3.150
3.690
3.983
4.271
4.799
5.174
-
5.368
5.604
5.665
5.943
6.517
-
7.299
7.583
7.663
-
8.264
10.913
11.362
11.879
12.184
-
12.495
12.996
-
13.423
14.053
14.360
14.624
14.999
15.075
15.313
15.514
15.689
15.899
16.339
17.105
17.449
17.882
18.090
18.332
18.602
18.710
18.865
18.992
19.772
19.838
20.112
21.231
24.526
27.531
PBM
Quality
_
-
72
-
-
59
64
49
45
47
-
64
47
95
91
95
-
53
91
83
-
64
64
91
43
91
-
59
74
-
87
90
90
80
91
91
91
72
70
45
91
96
80
59
86
81
91
38
90
94
93
94
83
94
43
91
Retention
Time, TRC
2.293
1372
2.971
3.038
3.157
3.693
3.989
4.278
4.801
5.179
5.229
5.372
5.608
5.670
5.948
6.523
6.828
7.303
7.588
7.667
7.887
8.268
10.919
11.369
11.887
12.192
12.378
12.502
13.004
13.130
13.431
14.061
14.368
14.633
15.006
15.081
15.319
15.521
15.701
15.905
16.350
17.113
17.456
-
18.097
18.339
18.606
-
-
-
19.790
-
20.119
21.239
-
27.540
IRQ Class
Quality 1 Hit
_
843
978 +
940
915
921 +
979 t
984 +
930
965
923
958
917
939
984 +
971
938
920
981 +
977
906
922
966 +
953 +
952
976 +
948
972 +
969 +
945
967 +
970 +
951
941 +
972 +
950
901 +
950
926 +
950
967 +
982
949 +
-
951
947 +
939 +
-
-
-
920
-
929 +
983
-
989 +
Identification,
Most likely if not Class 1
Carbon dioxide (air)
Water
2-Propanol
Acetone
Methylene chloride
2-Butanone
Acetic acid, ethyl ester
1-Propanol, 2-methyl
2-lsopropoxy ethanol
Branched paraffin
2,3-Dimethyl pentane
Branched paraffin
Olefin or cycloparaffin
1,2-Dimethyl cyclopentane
Heptane
Methyl cyclohexane
Ethyl cyclopentane
1-Methoxy butane
Toluene
iso Butyl acetate
2-Methyl Heptane
2,2-Diethoxy propane
Ethyl Benzene
Meta xylene
alcohol
Ortho xylene
1,4-Dimethyl cyclohexane
2-Butoxy ethanol
Isobutyl butyrate
1,2-Dimethyl cyclohexane
Cumene
Propyl cyclohexane
3-Methyl nonane
Propyl benzene
1-Ethyl-3-methyl benzene
1-Ethyl-4-methyl benzene
1,2,4-Trimethyl benzene
4-Methyl nonane
1-Ethyl-2-methyl benzene
3-Methyl nonane
1,2,4-Trimethyl benzene
Decane
1,2,3-Trimethyl benzene
1-Ethenyl-2-methyl benzene
4-Methyl decane
Isobutyl cyclohexane
1,3-Diethyl benzene
Diethyl benzene
U,8-P-Menthatriene
1-Ethyl-2,4-dimethylbenzene
2-Ethyl-1,4-dimethylbenzene
1 -Methyl-3-isopropylbenzene
1,2,4,5-tetramethylbenzene
Undecane
1,3-Dioxolane-2-methanol
Phthalic anhydride
Table 1. Summary of compounds found in hazardous waste sample
11-95
-------
Environmental Applications of Multispectral Analysis
by
John M. McGuire
Environmental Research Laboratory
U.S. Environmental Protection Agency
Beginning in the early '70s, extensive application of gas
chromatography/mass spectrometry (GC/MS) for identification of organics in
water led to its now being the accepted method for positive identification of
target analytes. GC/MS with automated spectra matching against a reference
collection of mass spectra is known to be excellent for specific
substantiation of target compounds, but its current success rate for tentative
identification of unknowns is poor. In particular, it fails to detect and/or
identify compounds whose mass spectra are not in the spectral libraries.
In order to improve the identifications of nontarget compounds, we have
applied other existing techniques to environmental sample extracts. This
Multispectral Analysis approach uses high resolution mass spectrometry (HRMS)
to determine elemental compositions of ions, Fourier transform infrared (FTIR)
spectroscopy to recognize sub-molecular structures, and chemical ionization
(Cl) mass spectrometry to establish molecular weights of the unknowns. The
spectral information is then melded together to postulate the structures of
the unknown compounds. Results on application of this technique to
unidentified compounds in environmental samples have been excellent. Upon re-
examination of samples from a survey conducted by the EPA Office of Water, we
identified two series of aldehydes as well as a variety of organophosphates
whose spectra were not included in the reference collection of mass spectra.
In the course of the work, the approach was also applied to correct a
misidentification made by routine spectra matching.
1-96
-------
56
SAMPLE PREPARATION USING SUPERCRITICAL FLUID EXTRACTION METHODOLOGY
Werner F. Beckert, U.S. Environmental Protection Agency, EMSL-LV, Las
Vegas, Nevada 89109, and Viorica Lopez-Avila, Mid-Pacific Environmental
Laboratory, Mountain View, California 94043.
ABSTRACT
Although extraction of analytical samples with supercritical fluids
(SFs) has received much attention during the last 10 or 20 years,
applications of supercritical fluid extraction (SFE) techniques to the
extraction of compounds regulated by the Environmental Protection
Agency (EPA) from matrices of concern to the EPA have been rather
limited. In late 1988, we started a project to develop SFE methods for
samples of interest to the EPA. Based on our results, which are summa-
rized in this paper, we developed a draft protocol for SFE of environ-
mental samples that has undergone a limited multi-laboratory evalua-
tion. Furthermore, an EPA work group for SFE development has been
formed, with participants from EPA, other Government agencies, industry
(especially the SFE equipment manufacturers) and academia, and evalua-
tion of commercially available SFE instrumentation is continuing. The
results to date demonstrate that SFE is a viable alternative to conven-
tional methods for the extraction of organic pollutants from solid
samples. However, our results also demonstrate that the many factors
affecting SFE efficiency make it difficult to optimize the method, and
that more developmental work has to be done before SFE becomes an easy-
to-use, off-the-shelf method.
INTRODUCTION
The Environmental Protection Agency (EPA) is interested in new and
improved analytical methods which are faster, better and cheaper than
present methods, and which, at the same time, are safe and environment-
friendly (by minimizing the generation of waste). Such methods, when
not developed specifically for the analysis of environmental samples,
must be adapted to EPA needs with respect to matrices, analytes or
analyte groups, sample sizes, data quality objectives (precision and
accuracy requirements), etc. The methods should be generic, as far as
analytes and matrices are concerned, and they should not be restricted
to any particular brand of instrumentation or equipment.
NOTICE: Although the research described in this paper has been
supported by the U.S. Environmental Protection Agency, it has not been
subjected to Agency review and therefore does not necessarily reflect
the views of the Agency, and no official endorsement should be
inferred. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
11-97
-------
Sample extraction techniques should, to the extent possible, yield
quantitative recoveries of the target analytes from the matrices, be
selective so that extraction of interferants is minimized, not generate
large volumes of waste solvents, require little sample and extract
handling to minimize analyte losses and contamination, and be fast and
inexpensive. The two methods that are at present included in the SW-
846 methods manual1, Soxhlet extraction (Method 3540) and sonication
extraction (Method 3550), only partially fulfill these extraction
goals. A third extraction method, Soxtec extraction (which is basic-
ally a modification of Soxhlet extraction), comes somewhat closer to
reaching these goals. These three methods have recently been evaluated
for their relative merits2.
For a number of years now, supercritical fluid extraction (SFE) has
been publicized as a new and promising technique for the extraction of
organic compounds from solid matrices. Some of the claimed advantages
of SFE over conventional extraction methods include much shorter
extraction times and close to quantitative recoveries. No toxic and
expensive solvents are required which results in reduced materials and
waste disposal costs, and in reduced environmental pollution. No sol-
vent removal is required, and no glassware cleaning. SFE conditions
can be optimized by varying pressure and temperature and by using
modified supercritical fluids (SFs), and extractions can be performed
at relatively low temperatures, if desired. Overall, the use of SFE
techniques in place of conventional methods could result in substantial
cost and labor savings.
In principle, SFE is similar to other solvent extraction techniques,
except that the solvent is in its supercritical (SC) state. SFs have
some unique properties that put them between liquid and gases. Their
viscosities are much lower than those of liquids and their surface
tension is zero, that means, they can penetrate into the pores of
solids much more easily than liquids. Their densities are close to
those of liquids which means their capacities for carrying dissolved
materials are similar to those of liquids.
The most commonly used SF is C02; others that are being used, or have
been investigated, include nitrous oxide, sulfur hexafluoride, Freon-
13, ammonia, xenon and several hydrocarbons. SF C02 is so popular
because of its low critical temperature (31.3ฐC) and pressure (72.9
atm), and because it is non-toxic, non-flammable, relatively non-
reactive and inexpensive, and its use does not result in a waste dis-
posal problem. It is a rather non-polar solvent, similar to hexane or
benzene, but both solvent strength and selectivity can be improved by
the addition of small amounts of modifiers, such as acetone, methanol,
or toluene.
Application of SFE to the extraction of compounds regulated by the EPA
from solid analytical samples has been limited. Brady et al.3
extracted PCBs, DDT and toxaphene from spiked soil samples with SC C02.
Schantz and Chesler4 extracted urban particulate matter and sediments
11-98
-------
with SC C02 and found that recoveries of PCBs and PAHs were
approximately equivalent to those obtained via Soxhlet extraction.
Smith and coworkers5'6 used SC CO, and SC isobutane to extract various
condensed aromatic and heterocyclic compounds from urban dust and from
XAD-2 Spherocarb. Hawthorne et al.7~10 used SC CO and SC nitrous
oxide to extract PAHs from samples of urban dust, fly ash and river
sediment. Other authors reported extraction of triazine herbicides
with SC C02 from spiked sediment samples11, PCDDs and PCDFs from fly
ash samples with SC nitrous oxide12, and PCDDs from sediment samples13.
Additional applications of SFE techniques to environmental sample
extractions have been reported at recent scientific meetings and
symposia14"17.
SUMMARY OF EXPERIMENTAL RESULTS
We started with our experimental studies in late 1988 with a Suprex
Model SE-50 extraction system using either a single-extraction vessel
arrangement, or a two- or four-vessel arrangement where two or four
extractions were performed simultaneously. In the multi-vessel experi-
ments we were mainly interested in establishing the equivalency of the
results obtained from the parallel extractions. All experimental work
was performed at the Mid-Pacific Environmental Laboratory (formally
Acurex Corp.). The bulk of the results is summarized below (for
details see ref. 18):
Single-Vessel Extractions
0 Seventeen organochlorine pesticides (OCPs) were spiked on sand at
500 and 2,500 ppb and extracted with SC C02 for 30 min at 150 atm
and 50ฐC. The mean recoveries were almost quantitative for most of
the compounds. A combination of static and dynamic extraction, as
well as variations of P and T, gave similar results.
0 Forty-one OCPs were spiked on sand and extracted with SC C02
modified with 10% methanol. The recoveries from the triplicate
samples were 79% or higher for 38 of the 41 compounds.
0 OCPs were spiked at two levels on soil containing 10% moisture and
extracted with SC CO, using a combination of static and dynamic
steps at various p and T settings. The mean recoveries were 80 to
90% (but only about 50% for endrin aldehyde). The moisture (which
can be regarded as a modifier) obviously did not drastically change
the extraction efficiencies under these conditions.
0 Aroclors 1232 and 1260 were spiked at 5000 ppb on Florisil and
extracted for 40 min with SC C02 at various conditions for p and T.
The recoveries were quantitative at relatively low temperatures and
high pressures.
11-99
-------
0 Fourteen phenols were spiked on sand at 3.6 to 18 ppm and extracted
with SC C02. The recoveries ranged from 53 to 129%, except for
2,4-dinitrophenol with only 27% recovery.
0 Twenty-five organophosphorus pesticides (OPPs) were spiked on sand
at 2.5 ug/g and extracted with SC CO or SC C02 modified with 10%
methanol. The recoveries were significantly higher when methanol-
modified CO was used (20 recoveries > 80%, compared to only 8
recoveries > 80% for SC CO alone); however, in both cases,
several OPPs were not recovered at all.
0 Sixteen polynuclear aromatic hydrocarbons (PAHs) were spiked on
coal, coal fly ash, sand and urban dust. Mean recoveries after
extraction with SC CO at 150 atm/50ฐC/60 min were almost quantita-
tive for samples of the coal and coal fly ash but only 57% for the
urban dust samples. Mean recoveries from the spiked sand samples
under two sets of conditions improved with added modifier (200 yL
acetone, added to the sample) from 74 to 81% in one case and from
58 to 89% in the other case.
0 Soil samples (SRS 103-100, Fisher Scientific), certified for 13
PAHs, dibenzofuran and pentachlorophenol, were extracted with SC
C02 at 300 atm/70ฐC/60 min. Ten percent water was added to each
sample prior to extraction. All recoveries were >60%, except for
benzo(b and k)fluoranthene (53%) and benzo(a)pyrene (32%).
0 Sand was spiked with 43 neutral/acidic compounds and extracted with
SC C02, with and without modifier (200 pL acetone) added to the
sample. Some 20 recoveries were lower when the modifier was used,
and only 14 recoveries increased.
Two-Vessel Extractions
0 Sand was spiked with 36 nitroaromatic compounds and extracted with
SC C02 under two sets of experimental conditions. The agreement
between the duplicate extractions performed in parallel was
excellent (generally within 10%). The more volatile nitroaromatics
gave good recoveries at lower (200 atm) but not at higher (300 atm)
pressures.
0 19 haloethers spiked on sand and extracted with SC C02 gave mean
recoveries of 73 to 99% for all but two compounds. The agreement
between the duplicate extractions performed in parallel was within
15% for most of the compounds.
Four-Vessel Extractions
0 Sand was spiked with 19 haloethers and extracted with SC C02 at 250
atm/60ฐC/60 min. Of the 19 compounds, 15 were recovered at >75%,
and the other four all at above 45%. The agreement between the
1-100
-------
parallel extractions, expressed as % RSD, ranged from 1.2 to 35.6%,
with 12 values being below 10%.
0 Of 42 OCPs, spiked on sand and extracted with SC C02, 35 gave
recoveries >50% whereas two (chlorobenzilate and endosulfan
sulfate) were not recovered at all. Twenty-six of the RSDs were
<10%, and the others were between 10 and 23%, except for
l,2-dibromo-3-chloropropane (30.8%).
The only limitations we have experienced with the four-vessel
arrangement was in the duration of the extraction. When working with
2-mL extraction vessels and using 50-ym restrictors, the 250-mL syringe
pump allows a maximum extraction time of 60 min.
DISCUSSION
As can be seen from the above summary of our experimental results, the
extraction efficiencies we achieved with a variety of samples are in
general good to reasonable, especially since in most cases we had not
tried to optimize our extraction conditions. Some of our recoveries
were much lower than those reported by others. However, one has to
realize that at least in some of the cases reported in the literature
the extraction conditions had been optimized in a trial-and-error
approach. In addition, such experiments were often conducted with
homemade equipment, focused mostly on PAHs, and used small sample sizes
(as small as a few milligrams). In order to develop a SFE method that
can successfully be applied to samples of interest to the EPA, we have
to use commercially available equipment, consider a wide variety of
sample matrices and groups of pollutants, and use sample sizes large
enough (1 to 10 g, preferably at least 5 g) for the inevitable
inhomogeneities of most real environmental and hazardous waste samples.
There is a lack of standard reference materials that include the
matrices and pollutants of environmental concern. The materials that
are available are either spiked matrices (soils, etc.), or they are
certified for only a very limited number of compounds, e.g., PAHs. It
is therefore difficult, even currently impossible, to determine
absolute extraction efficiencies for most analytes because in most
cases removal of a spike from a sample matrix is much easier than
removal of "incorporated" or "native" compounds. This, however, is a
problem that hampers the evaluation of all extraction methods, not just
SFE, and one is usually confined to comparing relative extraction
efficiencies.
Temperature and pressure changes affect the density and viscosity of a
SF and therefore its solubilizing ability. However, it is little
understood what happens on the surfaces of the solid matrices during
the extraction process, and what the desorption, solvation and trans-
port mechanisms are, and little is known about how to optimize p and T
for specific matrices and analyte groups. Just raising the pressure
does not seem to be the answer.
11-101
-------
An understanding of the desorption and transport mechanisms of solutes
under SC conditions would provide clues whether the use of static or
dynamic extraction conditions, or a combination of the two, would be
advantageous; whether rapid pressure fluctuations would improve extrac-
tion rates and, maybe, extraction efficiencies and selectivities; and
whether application of ultrasound could enhance extraction efficiency
and rate, as has been suggested6'19.
CURRENT ACTIVITIES
Summarized below are the SFE activities in which EPA is at present
involved.
EPA SFE Methods Development Group
In January 1990, an SFE Methods Development Group was formed. The
overall goal of this group is to assist and advise in the development
of SFE to make it a viable, attractive and affordable alternative to
conventional extraction methods. This effort includes
0 development of a general, standardized SFE method, or set of
methods, for a variety of analytes and matrices,
0 generating performance data for the method(s) through intra- and
interlaboratory evaluation studies, and
0 improving and, to the extent practical, standardizing hardware.
The SFE Methods Development Group members come from EPA (OSWER, ORD,
and Regional laboratory personnel), instrument manufacturers, academia,
and other interested contractor laboratories. Semi-annual meetings
provide a forum for candid discussions of results and problems, of new
approaches and of specific applications.
Protocol Development and Evaluation
Based on our results we developed a draft protocol in the SW-846 format
"Extraction Procedure Using Supercritical Fluids." Our goal was to
write a generic protocol that is applicable to as many different SFE
systems as possible. It is written for solid matrices like soils and
sediments; the target analytes include organochlorine pesticides,
polychlorinated biphenyls, polynuclear aromatic hydrocarbons, phenols,
phthalate esters, and organophosphorus pesticides. The protocol
addresses interferences, apparatus and materials, sample preparation
(including extraction), and quality assurance. An updated protocol
version was recently evaluated by 10 laboratories for its feasi-
bility21. The analytical data generated by the different laboratories
varied substantially, however, it was confirmed that the protocol could
be followed without problems by all operators involved, independent of
1-102
-------
the particular SFE system used, and this has been the main purpose of
the evaluation study.
Instrument Evaluation
Most of the commercially available SFE systems have been evaluated at
the Mid-Pacific Environmental Laboratory under contract to the EPA.
The generous help from the instrument manufacturing companies is
acknowledged and greatly appreciated. The main goal was to assure that
our protocol was compatible with all instruments. In general, all
instruments performed adequately, although, as can be expected, none of
the instruments was without some problems, and feedback was provided to
the manufacturers as to problem areas and perceived weaknesses of their
instruments. However, it must be understood that evaluation of an
instrument does not constitute endorsement by the EPA.
Extraction and Optimization Studies
ฅork is continuing in EPA laboratories and laboratories under contract
to the EPA on different matrices (e.g., fly ash, bottom ash, clay-type
soil, soil high in organics, river and marine sediments, etc.) and on
method optimization. Currently, a method for the extraction of oil/
grease and total petroleum hydrocarbons from soil is being developed.
The current EPA methods specify extraction with Freon , however,
Bicking et al.21 and others have shown that these materials can be
extracted with C02 under SC conditions. Our own results confirmed
this, and a draft protocol for the determination of oil/grease and
total petroleum hydrocarbons has been prepared. Another method of cur-
rent concern is the extraction of phenoxyacid herbicides and other
acidic compounds for soil. Miller et all22 have shown that acidic com-
pounds can be derivatized by adding trimethyl phenyl ammonium hydroxide
in methanol to the material in the extraction vessel, followed by
static and then dynamic SFE. Finally, we are looking at the effect of
ultrasound application during SFE which, as discussed earlier, seems to
increase extraction rate and possibly extraction efficiency.
CONCLUSION
Supercritical fluid extraction is an attractive method for the
extraction of organic contaminants from matrices of concern to the EPA.
The most-used extraction medium, carbon dioxide, is non-toxic and non-
polluting, and creates no waste-disposal problems. Potential
advantages of the method include reduced material and manpower needs,
speed, high efficiencies, selectivity (in combination with modifiers),
and high versatility, especially in combination with advanced analy-
tical techniques. However, more developmental work has to be done
before SFE becomes an easy-to-use, off-the-shelf method.
11-103
-------
References
1. Test Methods for Evaluating Solid Waste (1986), 3rd Ed., SW-846,
U.S. Environmental Protection Agency, Washington, D.C.
2. V. Lopez-Avila, J. Milanes, N. Dodhiwala, J. Benedicto and W.F.
Beckert. "Evaluation of Sample Preparation Methods for Solid
Matrices." Proc. Seventh Annual Waste Testing and Quality
Assurance Symposium, Washington, D.C., July 8-12, 1991.
3. B.O. Brady, C.P.C. Kao, K.M. Dooley, F.C. Knopf and R.P. Gambrell.
"Supercritical Fluid Extraction of Toxic Organics from Soils."
Ind. Eng. Chem. Res. 26: 261-268 (1987).
4. M.M. Schantz and S.N. Chesler. "Supercritical Fluid Extraction
Procedure for the Removal of Trace Organic Species from Solid
Samples." J. Chrom. 363: 397-401 (1986).
5. B.W. Wright and R.D. Smith. "Supercritical Fluid Extraction of
Particulate and Adsorbent Materials: Part II." EPA Report 600/4-
87/040 (1987).
6. B.W. Wright, J.L. Fulton, A. J. Kopriva and R.D. Smith.
"Analytical Supercritical Fluid Extraction Methodologies." In:
Supercritical Fluid Extraction and Chromatography: Techniques and
Applications. B.A. Charpentier and M.R. Sevenants, Eds. ACS
Symposium Series 366: 44-62 (1988).
7. S.B. Hawthorne and D.J. Miller. "Directly Coupled Supercritical
Fluid ExtractionGas Chromatographic Analysis of Polycyclic
Aromatic Hydrocarbons and Polychlorinated Biphenyls from
Environmental Solids." J. Chrom. 403: 63-76 (1987).
8. S.B. Hawthorne, M.S. Krieger and D.J. Miller. "Analysis of Flavor
and Fragrance Compounds Using Supercritical Fluid Extraction
Coupled with Gas Chromatography." Anal. Chem. 60: 472-477 (1988).
9. S.B. Hawthorne and D.J. Miller. "Extraction and Recovery of
Organic Pollutants from Environmental Solids and Tenax-GC Using
Supercritical C02." J. Chrom. Sci. 24: 258-264 (1986).
10. S.B. Hawthorne and D.J. Miller. "Extraction and Recovery of
Polycyclic Aromatic Hydrocarbons from Environmental Solids Using
Supercritical Fluid." Anal. Chem. 59: 1705-1708 (1987).
11. V. Janda, G. Steenbeke and P. Sandra. "Supercritical Fluid
Extraction of s-Triazine Herbicides from Sediment." J. Chrom. 479:
200-205 (1989).
12. N. Alexandrou and J. Pawliszyn. "Supercritical Fluid Extraction
for the Rapid Determination of Polychlorinated Dibenzo-p-dioxins
11-104
-------
and Dibenzofurans in Municipal Incinerator Fly Ash." Anal. Chem.
61: 2770-2776 (1989).
13. F.I. Onuska and K.A. Terry. "Supercritical Fluid Extraction of 2,
3,7,8-Tetrachlorodibenzo-p-dioxin from Sediment Samples." J. High
Resol. Chrom. 12: 357-361 (1989).
14. Sixth Annual Waste Testing and Quality Assurance Symposium,
Washington, D.C., July 16-20, 1990.
15. International Symposium on Supercritical Fluid Chromatography and
Extraction, Park City, Utah, January 14-17, 1991.
16. 201st American Chemical Society National Meeting, Atlanta, Georgia,
April 14-19, 1991.
17. Thirteenth International Symposium on Capillary Chromatography,
Riva del Garda, Italy, May 13-16, 1991.
18. V. Lopez-Avila and N.S. Dodhiwala. "Method for the Supercritical
Fluid Extraction of Soils/Sediments." EPA/600/4-90/026, September
1990.
19. Carl A. Mabee, Dearborn Chemical Company Limited, personal
communication, 1991.
20. T.L. Jones and T.C.H. Chiang. "An Interlaboratory Comparison Study
of Supercritical Fluid Extraction for Environmental Samples."
Proc. Seventh Annual Waste Testing and Quality Assurance Symposium,
Washington, D.C., July 8-12, 1991.
21. M.K.L. Bicking, F.L. DeRoos and T.G. Hayes. "An Experimental
Design Approach to Optimization of Supercritical Fluid Extraction
Conditions." Presented at the International Symposium on
Supercritical Fluid Chromatography and Extraction, Park City, Utah,
January 14-17, 1991.
22. D.J. Miller, S.B. Hawthorne and J.J. Langenfeld. "SFE with
Chemical Derivatization for the Recovery of Polar and Ionic
Analytes." Ibid.
11-105
-------
57 THE RESEARCH STATUS OF SUPERCRITICAL FLUID EXTRACTION FOR
THE ANALYSIS OF PCBs IN INCINERATOR ASH
Dr. Peter A. Pospisil. Manager Methods Development, Matthew A. Kobus, Chemist
Methods Development, Charles R. Hecht, Senior Chemist Methods Development,
Chemical Waste Management, Inc. 150 West 137th Street, Riverdale, IL 60627
ABSTRACT
The extraction time of PCBs from incinerator ash can be reduced to less than one
hour using supercritical fluid extraction with carbon dioxide.
SW-846 Method 8080 is currently the only EPA approved method for the extraction
of PCBs in solid matrices. The sample is first extracted in a Soxhlet apparatus for 16
hours with hexane/acetone. The solvent volume is then reduced using a Kuderna-
Danish apparatus, prior to GC-ECD analysis. The time intensive extraction step is
the current limiting factor in reducing the turnaround time of the analysis.
Supercritical fluids combine the mass transport properties of a gas with the solvation
properties of a liquids. This research applied supercritical fluid technology to a
specific combustion matrix, incinerator ash, both to reduce the turnaround time of
the method and to minimize solvent usage. The conditions required for extraction,
including sample preparation, extraction temperature, extraction time, modifier type
etc. were determined in a systematic manner to maximize the extraction speed and
PCB recovery. The effect of the adsorptive properties of the matrix on the analyte
were also investigated.
The application of supercritical fluid extraction technology for PCB extraction will
enable laboratories to provide same day analytical service, while reducing
laboratory costs.
PURPOSE OF WORK
The purpose of this work was to apply supercritical fluid technology to develop a
method for the extraction of PCBs from incinerator ash. The technical approach
used to define method parameters was the systematic evaluation of each element
while holding all remaining variables constant. In this type of study it is extremely
important to differentiate between Comparative and Absolute extraction. In the
comparative situation there is a fixed goal based on data generated by an accepted
method. In the absolute case there is confirmation via several alternative
techniques that the extraction is indeed complete. The authors have chosen a
comparative study because of matrix considerations and analyte concentrations.
11-106
-------
Page 2
MATRIX OVERVIEW
Incinerator ash is defined as a combustion matrix. It is a by-product of the
incineration process containing a variety of newly formed active organic and
inorganic adsorption sites. The adsorptive strengths of the sites can vary as well as
their distribution throughout the ash particle. Although the ash is primarily a glassy
product containing a small amount of carbon polymer, it does have some pore
structure. An ash particle is illustrated schematically in Figure 1.
The irregularly shaped polyhedron represents the basic ash particle contaminated
with PCBs. The light and dark dots represent weakly and strongly adsorbed PCBs.
The material on the surface is fairly easy to remove, its "removability" depends on
the strength of the adsorptive site compared to that of the extractant. The material
within the pore not only must be desorbed, but must diffuse to the pore mouth prior
to being swept into the flowing CCซ2. This makes the removal of these materials
diffusion limited. Any material occluded within the vitreous ash will never be
removed unless the ash is physically degraded to expose the PCB to the extractant.
SUPERCRITICAL FLUID OVERVIEW
The molecules of a liquid are bonded by electrostatic forces, which are a function of
the molecule's polarity. The heat of vaporization represents the energy required to
break these associative bonds as the liquid becomes a vapor. When a liquid in
equilibrium with its vapor, is sealed in a tube and heated, the pressure of the closed
system rises and the liquid's heat of vaporization, and corresponding intennolecular
associative forces decrease. When the associative forces reach zero, the liquid and
gas phases become one. This temperature and the corresponding pressure, which
are unique for each liquid, are known at the critical constants.
This non-associated supercritical phase (fluid) has unique physio-chemical
properties. Its viscosity and diffusion constant approximate those of a gas, making it
an ideal material to permeate small pores. Its density and solvency approach those
of a liquid, enabling it to dissolve a broad range of organic compounds. The
technique is not thermally driven, thus it is also possible to extract thermally labile
and non-volatile materials.
The solvency of the mobile phase is a function of its density. Increasing the density
generally increases the solubility of larger molecular weight species. Carbon dioxide
is the most frequently used material because of its low critical temperature,
inertness and safety.
11-107
-------
Page 3
INSTRUMENT DESCRIPTION
The supercritical fluid extraction instrumentation was purchased from the Lee
Scientific Co., and consists of the following components.
Pump - A pump to supply pressurized liquid CO? to the extraction cell, that
has the ability to deliver at least 20 mL ofliquid CC*2 at an operating
pressure of up to 400 atmospheres, (6150 PSI).
Oven - An oven capable of maintaining a temperature of 100 degrees C.
Extraction Cell - consisting of a 10 cm, 4.6 mm ID HPLC column, 2 micron
frits, and hardware to seal the end of the column.
Restrictors - Fused silica capillary producing a carbon dioxide flow rate of 1.8
mL/min of liquid, or 900 mL/min gas through the cell, usually about
60 cm long, 50 micron ID.
Liquid Carbon Dioxide - Supercritical fluid grade, in a tank that must contain
a dip tube to deliver liquid product (Scott Specialty Gases).
Figure 2a shows a schematic of the apparatus used for the study. The components
occupy about 6 ftz of bench and floor space, including space for a single gas
cylinder.
EXPERIMENTAL PROCEDURE
Sample Selection
The authors felt that the best technical approach was to select a hazardous waste
incinerator ash sample containing native, as opposed to spiked, PCBs. Typical
incinerator ash PCB concentrations approximated 3 ppm, about 15 times lower than
the regulatory level.
The initial work for this study was performed on this type of ash. Reproducibility
difficulties, arising from sample size and GC detection limits constraints, required
that a sample of higher concentration be obtained. This, due to the nature of the
typical incinerator ash, dictated that the sample be spiked. A 250 gram sample of
incinerator ash was then spiked to a level of 50 ppm with Aroclor 1260, by the
technique of solvent evaporation in a rotary evaporator.
Sample Preparation and Analysis
A gallon of incinerator ash was crushed in a reciprocating jaw crusher. All of the
ash was sifted through a 9.5mm screen to remove non-extractable items, such as
large metallic shards etc. Particles greater than 1.0mm were hand ground in a
mortar and pestle until they passed through the 16 mesh screen. The ash sample
was homogenized by passing it through a riffler three times. The moisture content
was determined to be 3.14% by oven drying at HOC. The PCB was determined to
be 3.4ppm of Aroclor 1260, using SW-846 Method 8080.
11-108
-------
Page 4
Preparation of Spiked Sample
Preparation of 250g of ash, spiked to a concentration of SOppm, was accomplished
by slurrying 30 gram portions of incinerator ash with acetone and spiking the slurry
with 1.5mL of lOOOppm Aroclor 1260 spiking solution. After mixing for a few
minutes, the solvent was driven off slowly in a rotoevaporator over the period of
approximately 45 min. This spiking method was repeated several times to create
enough ash for experimental purposes. The spiked ash was homogenized by
combining all fractions into a single container and riffling three times. The PCB
was determined to be 54.4ppm of Aroclor 1260, using SW-846 Method 8080.
Extraction Procedure and PCB Analysis
The cell was assembled as shown in Figure 2b. The extraction cell was completely
filled with ash to eliminate dead space. The carbon dioxide was turned on slowly
and brought up to pressure within one to two minutes. The collection fluid was five
mL of hexane in a ten mL graduated cylinder. Additional hexane was added to the
graduated cylinder at the completion of the extraction to compensate for
evaporative losses. All PCB analysis were performed using SW-846 Method 8080,
GC-ECD (a capillary column technique), along with a compliment of QA including
calibration standards, spikes and duplicate spikes.
Effect of Time on PCB Extraction
In order to determine the effect of time on the extractability of PCBs from the ash,
weighed ash samples were extracted with unmodified CC>2 for varying time periods
ranging from 5 to 60 minutes. These data are presented in Figure 3 and show that
the extraction curve reaches a plateau in about 45 minutes. The following
extraction conditions were used:
CC-2 flow = 1.8 mL/min
Temperature = 100 ฐC
Pressure = 400 atm
Effect of Temperature on PCB Extraction
In order to determine the effect of temperature, and thus density, on the
extractability of PCBs from the ash, weighed ash samples were extracted with
unmodified CC>2 for varying time periods and at the following temperatures:
Temperature Phase Density
ฐC g/mL
35 Liquid .98
100 SCF .76
200 SCF .50
300 SCF .36
These data are presented in Figure 4 and show that the maximum amount of PCB is
extracted at a temperature of 100ฐC. This indicates that density is more important
than the increase in diffusion coefficient for removing PCBs from the incinerator
matrix. Liquid CC>2 doesn't work as well as the corresponding supercritical fluid
because of the decreased transport properties of carbon dioxide in the liquid state.
11-109
-------
PageS
Effect of Particle Size PCB Extraction
In order to determine the effect of particle size on the extraction of PCBs from ash,
weighed ash samples of two different mesh sizes were extracted with unmodified
CC>2 for varying time periods ranging from 5 to 60 minutes. The data are presented
below and show that more PCBs are extracted from the sample when it is ground to
a mesh size of 60 or more.
Mesh Size Amount Extracted
10 -20 3.2
40 -60 4.2
The following extraction conditions were used.
CC>2 flow = 1.8 mL/min
Temperature = 100 ฐC
Pressure = 400 atm
Time = 40 min
Static-Dynamic Extraction using a Prc-Modificr
Based on information presented at the NIST conference by Mary McNally [1], liquid
methanol was introduced directly into the cell in an attempt to improve the
extraction efficiency. After sealing the cell containing methanol, the cell was
allowed to equilibrate for five minutes and then the CC>2 pressure was brought up to
400 atmospheres and held in the static mode for ten minutes. The run then
proceeded as described earlier. Samples were extracted using methanol, acetone
and toluene with only small improvements being noted.
Dynamic Extraction using Modified CCh
Based on the slight increase of PCB extracted using the different premodifiers, work
was initiated with modifiers directly added to the CC>2. This work and that of Larry
Taylor [2] at VPI, influenced the choice of 5% toluene as the modifier of choice for
CO2-
Extractive Reproducibility
Comparison of Soxhlet and SFE Extraction
Replicate SFE runs using toluene modified CC>2 extractions and Soxhlet extraction
were made using the same sample. The results are compared in Figure 5. These
data show recoveries, PCB by Soxhlet extraction, of 1009fe from the spiked ash with
relative standard deviation (RSD) of only 5%. The SCF extractions show 81% PCB
recovery, with a RSD of 12%. This brings the SCF extraction for PCBs in
incinerator ash into the equivalency range of Soxhlet technology.
11-110
-------
Page 6
Research Overview and Direction
Although the extraction recovery data has increased over the time period of the
research, more work needs to be done, in the areas of improved PCB recoveries,
and analytical precision. The use of toluene as a modifier improves PCB recoveries,
but considering the superior extractive properties of supercritical fluid CO?, one
would expect the recoveries of the two techniques to be more equivalent. Altering
the toluene modifier content or experimenting with different modifiers are two
possible approaches.
Analytical precision data generated in earlier experiment, for both extraction
techniques is the reverse of that presented in Figure 5. This suggests that there is a
additional degree of freedom that has not yet been addressed. Before an SCF
method can be finalized comparable precision data must be generated, to
demonstrate the absence of any additional variables.
The work clearly shows that SCF is a viable technique for the extraction of PCBs
from incinerator ash. These problems may be overcome and the application of SCF
for PCB extraction from incinerator ash will soon become a reality.
CONCLUSIONS
1 - PCBs can be extracted from incinerator ash using supercritical fluid carbon
dioxide modified with toluene.
2 - The extraction time is reduced to 50 minutes from 18 hours and uses only 5
mL of collection solvent.
3 - SFE is clearly a viable equivalent technique for PCB extraction to minimize
solvent use in the laboratory.
4- Additional work needs to focus on improved recoveries and analytical
precision.
REFERENCES
1 - Consortium on Automated Analytical Laboratory Systems, 1st Workshop on
Supercritical Fluid Extraction of Soild Environmental Samples, October 31,
1990, National Institute of Standards and Techonlogy, Gaithersburg, MD
20899
2- L.T.Taylor; A. J. Sequeira presentation at Pittsburgh Conference, 1991,
paper No. 1003
11-111
-------
FIGURE 1
RGB's IN INCINERATOR ASH
IV)
-------
FIGURE 2A
INSTRUMENT SCHEMATIC
A
CO
COM3KLLEK
FIGURE 2B
CELL SCHEMATIC
11-113
-------
FIGURE 3
SFE OF RGB'S
EFFECT OF TIME ON PCB EXTRACTION
3.5
PPM EXTRACTED
2.5
2
1.5
1
0.5
0
10
20 30 40
TIME MINUTES
50
60
70
100 deg C (10-20)
-------
FIGURE
SFE OF RGB'S
EXTRACTION TEMPERATURE SELECTION
Ol
3.5
3
2.5
2
1.5
1
0.5
0
PPM EXTRACTED
X
0 10
30 deg C (liq)
~B~ 200 deg C
20 30 40
TIME MINUTES
-+- 100 deg C
-*- 300 deg C
50
60
70
150 deg C
-------
FIGURE 5
Dynamic Extraction using Modifier
a>
E
a
a
XX
m
o
D.
o
z
O
o
60
50 -
40 -
30 -
20 -
10 -
MEAN
T I I
SPIKE
\ \ \
i i I i r
SOXHLET
MEAN
\ \
I I I I I I I I I !
SFE TOLUENE MODIFIER
-------
58 SUPERCRITICAL FLUID EXTRACTION (SFE) OF TOTAL PETROLEUM
HYDROCARBONS (TPHs) WITH ANALYSIS BY INFRARED
SPECTROSCOPY
Richard P. Lee, Methods Development Chemist, Mark L. Bruce. Director of Research and
Development, and Marvin W. Stephens, Vice-President, Corporate Technical Director
Wadsworth/ALERT Laboratories Inc.
4101 Shuffel Dr. N.W.
North Canton, Ohio 44720
ABSTRACT
Infrared spectroscopy is an attractive analysis procedure for the screening of petroleum
hydrocarbons in solid matrices because of its low cost and rapid sample throughput.
Coupled with off-line supercritical fluid extraction (SFE), this method provides a rapid
monitoring procedure with an order of magnitude reduction in the amount of solvent used
compared to the present Soxhlet method. This method uses supercritical carbon dioxide as
the extraction solvent to remove the target components from a solid sample and deposit
them into a collection vial containing 5 mL of solvent. Freon-113ฎ has been replaced in
this application by Fluorinertฎ FC-77 as the collection solvent. An extraction time of 25
minutes at 400 atmospheres and an oven temperature of 60ฐC provides a rapid, effective
means of extracting petroleum hydrocarbons from sand and Kaolin matrices.
INTRODUCTION
The methodologies for sample preparation have not kept pace with the developments in
sample analysis. The method that Franz Ritter von Soxhlet developed at the turn of the
century has changed very little. It is still the predominant method for the preparation of
solid samples. The need for an alternative sample preparation method is critical in the
analysis of petroleum hydrocarbons.
According to EPA estimates there are three to five million underground storage tanks in the
United States (1). Approximately 100,000 of these tanks are believed to be leaking. In
addition, as many as 300,000 more tanks are predicted to begin leaking in the next five
years (1). At present, semivolatile petroleum hydrocarbons are extracted by Soxhlet,
sonication, or Soxtecฎ using an organic solvent followed by gas chromatographic or
infrared analysis.
Freon-113ฎ is used when the analysis is performed by infrared spectroscopy. It is a
known ozone depleter (2), making it unacceptable as a laboratory solvent. According to
M.P. McCormick of Langley Center's Aerosol Research Branch, polar stratospheric clouds
provide a surface on which chlorofluorocarbons (CFCs) can react to free the chlorine to
react with ozone (2). In accordance to the Montreal Protocol on Substances that Deplete the
Ozone Layer (Montreal Protocol) and the Clean Air Act Amendments of 1990 (CAA),
CFCs will be phased out by the year 2000.
This paper presents a method in which supercritical fluid extraction (SFE) is used in the
preparation of solid samples containing trace concentrations of petroleum hydrocarbons. A
fluorocarbon is used as the collection solvent followed by infrared spectroscopic analysis.
1-117
-------
Fluorocarbons contain no chlorine and should not be confused with CFCs. There has been
no evidence of the cleavage of carbon-fluorine bonds, nor has there been a link between
fluorine and the depletion of stratospheric ozone (3). The benefits of SFE are well
documented in several informative reviews (4,5,6). The use of the fluorocarbon
Fluorinertฎ FC-77 as the collection solvent in this method eliminates the use of
Freon-113ฎ and compliments SFE's ability to provide an effective alternative to the present
TPH extraction methods.
INSTRUMENTAL. EQUIPMENT and SUPPLIES
Supercritical Fluid Extractor
Suprex, SFE/50
5 mL extraction vessel
600 mm fused silica restricter, 32 micron ID
Infrared Spectrograph
Perkin-Elmer, 710 Infrared Spectrophotometer
10 mm, 3 mL quartz cell
Hardware
16 x 60 mm glass vial
16 x 100 mm glass culture tube
500 mL round bottom flask
Modified Neilson-Kryger distillation apparatus
Heating mantel, Glas-Col 115 volts, 270 watts TM106
Temperature controller, Glas-Col 115 volts, 1500 watts, PL-312 Minitrol
Glass beads, 5 mm OD
Glass Pasteur pipets
Filter paper ashless 41, Whatman
Glass wool-silane treated, Supelco
Reagents and Standards
Freon-113ฎ, EM Science
Fluorinertฎ FC-72,3M
Fluorinert.ฎ FC-77, 3M
Isooctane, Mallinckrodt
Xylenes, Mallinckrodt
Hexadecane, EM Science
Kaolin, Baker Analyzed
Diesel fuel, retail Fuel outlet
Sand, washed and dried, Mallinckrodt
CO2, SFC grade with 1500 PSIA Helium headspace with dip tube, Scott Specialty Gases
Hexafluorobenzene, Aldrich*
Octafluorotoluene, Aldrich*
Bromopentafluorobenzene, Aldrich*
* These compounds were only used in the initial collection solvent search.
RESULTS and DISSCUSSION
The development for this method was conducted in two areas the search for a suitable
collection solvent and the optimization of supercritical fluid extraction parameters. The
approaches and representative results are discussed below.
11-118
-------
Collection solvent
The search for an appropriate collection solvent was broken down into four stages. In the
first stage, the catalogs of seven chemical vendors were examined for potential solvents.
This search resulted in sixty-six possible collection solvents.
In the second stage, vendors were contacted for additional product information regarding
the potential collection solvents as to their hydrocarbon solubility, IR spectra, and material
safety data. This information was used to eliminate all but six solvents. Small amounts of
these solvents were obtained for further testing.
The third stage was to confirm conclusions drawn from the information provided by the
vendors. All solvents were tested for hydrocarbon contamination by measuring the C-H
stretch region of the infrared spectrum (2900-3100 cm'1). Even with background
correction, the concentration of hydrocarbon contamination of the possible collection
solvents was prohibitively high for all solvents except the Fluorinertsฎ FC-77 and FC-72.
The 3M Fluorinertsฎ FC-77 and FC-72 provided an acceptable baseline over the spectral
region of interest with background correction.
A 10,000 mg/L diesel fuel standard was prepared in 25 mL Freon-113ฎ for solubility
comparison studies with solutions of FC-77 and FC-72 that were saturated with diesel fuel.
Using Freon-113ฎ as a solubility reference, it was estimated that the solubility limit of
diesel fuel in FC-77 and FC-72 is approximately 5000 mg/L. The low molecular weight
fraction of the diesel fuel appeared to be preferentially more soluble than the higher
molecular weight components when compared to the Freon-113ฎ standard.
The same three standard solutions used in the hydrocarbon solubility studies were used as a
mock collection solvent. This was to test the ability of the FC-77 and FC-72 to retain
hydrocarbons when COi was bubbled through them. These standard solutions were
purged for 10 minutes with a calculated liquid CO2 flow rate of 2.5 ml/min. Freon-113ฎ
retained 74% of the diesel fuel. FC-77 and FC-72 retained 46% and 35% respectively of
the diesel fuel. When FC-77 was purged with a liquid CC>2 flow rate of 0.7 ml/min the
retention rate unproved to 85%.
In stage four, FC-77 was chosen as the collection solvent over FC-72 because of its
slightly higher retention of hydrocarbons. A sand sample spiked with 100 mg/kg diesel
fuel was extracted off-line using 0.5 mL/min supercritical CO2. The absorbance was
measured by comparing the extract to FC-77. When FC-77 was used as the collection
solvent, a negative absorbance was obtained in the 3050 cnr1 region of the infrared
spectrum (the reference IR cell contained a higher concentration of a component than the
sample cell containing the extract). A volatile component of the solvent had apparently
been purged out by the CCซ2 during the extraction. A 3M technical representative indicated
that this volatile component could be contamination resulting from a methanol wash used in
the production process of the Fluorinertฎ solvent.
A 100 mL volume FC-77 was distilled for thirty minutes. The absorbance of the distilled
FC-77 was measured relative to FC-77 that had not been distilled. After distillation, a large
negative peak was obtained that extended over the 2900 to 3100 cnr1 region of the
11-119
-------
spectrum. It was concluded from this result that the volatile component contamination and
a portion of the hydrocarbon background contamination were distilled out of the solvent.
Distilled FC-77 was used as the collection solvent in an off-line extraction using CO2 at
0.500 mL/min of sand and Kaolin blank samples When the absorbance of the extract was
measured using distilled FC-77 for background correction, a flat baseline was obtained
without an interfering negative peak. Sand samples spiked with 1000 mg/kg diesel fuel
were extracted using the distilled FC-77. The spiked concentration recovery was 85%. It
is concluded from this data, that FC-77 can be used as the collection solvent for this
method.
Optimization of Extraction Parameters
The goal of this stage of the method development was to optimize the extraction parameters
(Tables 1&3) so that this method could be used with as broad a spectrum of environmental
matrices and hydrocarbon mixtures as possible. Due to the lack of standard reference
materials (SRMs) with known "native" TPH concentrations, spiked samples were used in
the development of this method. Using the recovery of spiked analytes to prove quantitive
extraction of "native" analytes is an uncertain comparison method. There is no way of
determining how spiked compounds compare to native pollutants in their interactions or
absorptive qualities that result from long term association with a matrix. This work
attempted to approximate matrix absorptive interactions by tumbling at a rate of 30
revolutions per minute diesel spiked Kaolin samples (porcelain clay) for approximately 24
hours. Kaolin is a highly absorptive, fine particle matrix with a high surface area. By
tumbling a spiked Kaolin sample these characteristic qualities would enhance the
absorbance of the spiked compounds. It was speculated that tumbling agitation over an
extended period of time would be a more realistic approximation of a "real world"
environmental sample than a sample that has been spiked and immediately extracted. Both
spiking techniques were used to investigate the ability of SFE to yield quantitative recovery
of petroleum hydrocarbons. Representative results from each approach are discussed
below.
Spiked samples (immediate extraction")
The initial evaluation of extraction parameters was conducted with the aid of statistical
experimental design software (Design-Easeฎ). Two Plackett-Burman designs were used to
screen the effects of the major extraction parameters (see Tables 1&3) using Freon-113ฎ as
the collection solvent. A Plackett-Burman experimental design is a special class of
fractional factorial design. This design was used to screen variables for further study by
isolating strong main effects. Interactions between variables were not considered.
Sand was the matrix throughout the first Plackett-Burman experimental design. All
samples were spiked with a 100 mg/kg TPH mixture consisting of 33% isooctane, 24%
xylenes, and 42% hexadecane, by weight. As Table 2 shows, TPH recoveries were high
throughout most of the range of conditions. This demonstrates the ease with which
hydrocarbons can be extracted when matrix interactions are minimal.
In the second Plackett-Burman in which Kaolin and sand were the variables, the matrix
was indicated as a main effect (see Table 4). When Kaolin was the matrix a mechanical
problem was discovered during the extraction . Due to the small partical size of this matrix,
periodic blockage of the extraction vessel frits resulted in fluctuation of the CC*2 flow rate.
11-120
-------
This fluctuation resulted in unacceptably low precision with replicate extractions
(53%-120%). A steady vigorous CO2 flow rate was obtained when a paper cartridge
containing the Kaolin was extracted. A paper cartridge was made out of folded filter paper
into which the Kaolin was directly weighed. The paper was folded and rolled so that the
cartridge could fit into the extraction chamber. Samples of 2 g of Kaolin in paper cartridges
were spiked to a 1000 mg/kg concentration of diesel fuel and extracted immediately with
approximately 0.7 mL/min CO2 (400 atm, 60ฐC, 10 min. static, 15 min. dynamic) using
FC-77 as the collection solvent. Recoveries of 80%-85% were obtained. A higher
pressure was needed in these extractions to improve the extraction effeciency of the higher
molecular weight components of diesel fuel.
Table 1. Plackett-Burman #1*
Factors
Mass of sample
Pressure
Oven temperature.
Type of extraction
Equilibrium time
Extraction time
Orientation of
extraction cell
Positive
Version of Factor
5
360
60
Dynamic (Dyn)
10
20
Vertical (Vert)
Negative
Version of Factor
1
150
35
Static
5
10
Horizontal (Horz)
units
Grams
Atmospheres
Degree-C
Minutes
Minutes
All samples were spiked to a 100 mg/kg concentration of the TPH mixture. Samples
were extracted using a 5 mL extraction vessel and a 600 mm length of 32 micron ID fused
silica restricter. Initial equilibrium time was a static step prior to the dynamic extraction
period.
Table 2. Plackett-Burman #1 Results
Factors
Mass of
sample
Pressure
Oven
temperature
Type of
extraction
Equilibrium
time
Extraction
time
Orientation of
extraction cell
Recoveries
run order
Standard
Order
Units
grams
Atmospheres
Degree-C
Minutes
Minutes
Percent
1
8
1
150
35
Static
5
10
Horiz
66.5
2
3
1
150
60
Dyn
10
10
Vert
66.6
3
5
1
360
35
Static
10
20
Vert
137
4
7
5
360
35
Dyn
5
10
Vert
76
5
1
5
360
60
Static
10
10
Horiz
84.9
6
4
5
150
35
Dyn
10
20
Horiz
84.8
7
6
5
150
60
Static
5
20
Vert
97.6
8
2
1
360
60
Dyn
5
20
Horiz
80.4
1-121
-------
Table 3. Plackett-Burman #2*
Factors
Pressure
Oven temperature
Equilibrium time
Extraction time
Mass of sample
Type of extraction
Matrix of sample
Orientation of
extraction cell
Positive Version
of Factor
360
60
5
20
1
Dynamic (Dyn)
Kaolin
Vertical (Vert)
Negative Version
of Factor
150
35
10
10
3
Static
Sand
Horizontal (Horz)
* All samples were spiked to a 100 mg/kg concentration of the '.
were extracted using a 5 mL extraction vessel and a 600 mm lengtl
silica restricter. Initial Equilibrium time was a static step prior to the
Units
Atmospheres
Degrees-C
Minutes
Minutes
Grams
fPH mixture. Samples
i of 32 micron ID fusee
: extraction period.
Spiked. Tumbled Samples (delayed extractions)
Thirty grams of Kaolin was spiked with 3 mL of 5000 mg/L diesel standard to give a 500
mg/kg concentration. This sample was tumbled at a rate of 30 revolutions per minute for
approximately 24 hours. Two gram aliquots of this sample were extracted off-line using
approximately 0.7 mL/min supercritical CO2 (400 atm, 60ฐC) using FC-77 as the
collection solvent. The initial static step was 10 minutes followed by a dynamic step of 15
minutes. Using these extraction conditions the best recoveries achieved were 62%.
Pressure was increased to 420 atm. and the temperature of the oven was decreased to 40ฐC
to achieve a greater supercritical fluid density. The liquid CO2 flow rate was increased to
approximately 0.8 mL/min. The recoveries did not change appreciably. Maintaining the
same pressure the extraction oven temperature was increased to 100ฐ C. This also did not
significantly change the recoveries of the diesel fuel. It was thought that possibly water
could displace the spiked hydrocarbons from the Kaolin and allow them to be swept out by
the supercritical CO2 and into the collection solvent (7). Water was added to the extraction
chamber prior to an extraction in three different cases in 200 uL, 300 uL and 1 mL
volumes. Recoveries decreased as the volume of water increased.
Method validation is in progress and the data is not available at this time. Validation will
consist of detection limit studies, performance comparision with Soxlet, sonication, and
Soxtecฎ extractions and precision studies involving several soil matrices.
Summary
Quantitative extraction and analysis of petroleum hydrocarbons was demonstrated by
immediate extraction of spiked samples. Because of the limitations that are inherent to
spiked samples in estimating recoveries of native components spiked samples were tumbled
for 24 hours. This was to allow greater mixing and enable the matrix to absorb the spiked
compounds. Some general comments can made from the development of this method:
1-122
-------
1) Fluorinertฎ FC-77 provides an adequate solubility limit for this application and with
pre-extraction distillation has an acceptably low hydrocarbon background contamination
level.
2) Though the approximate limit of hydrocarbon solubility in FC-77 is 5000 mg/L, lower
molecular weight hydrocarbons are more soluble than higher molecular weight
hydrocarbons. This should be taken into consideration when extracting higher molecular
weight hydrocarbon mixtures such as motor or crude oil.
3) When the matrix interaction with the analytes of interest is minimal, as in the case of
sand and immediatly extracted spiked samples, recoveries can be quantitatively high.
When the matrix interaction with analytes are increased, as with the tumbled spiked
samples, extraction of petroleum hydrocarbons is less efficient and might be improved by
using higher pressures than those explored in this study.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge Suprex Corporation for their help and cooperation and
the 3M Corporation for the generous supply of Fluorinertฎ FC-77 and FC-72.
REFERENCES
1) Industrial Safety & Hygiene News, 01/88.
2) NASA's Innovators, NASA Tech Briefs Vol.15, No.4, 1991, pp.114.
3) Danielson, R.D., Understanding Fluorocarbons, Electronic Packaging and Production,
July 1976, pp. 78.
4) Hawthorne, S.B, .Analytical-Scale Supercritical Fluid Extraction, Analytical Chemistry,
Vol. 62, No. 11, pp. 633A-642A, 1990.
5) Anderson, M.R., Swanson, J.T., Porter, N.L., Richter, B.E., Supercritical Fluid
Extraction as a Sample Introduction Method for Chromatography, Journal of
Chromatographic Science, Vol.27, pp. 371-377, 1990.
6) King, J.W., Fundamentals and Applications of Supercritical Extraction in
Chromatographic Science, Journal of Chromatographic Science, Vol. 27, pp. 355-364,
1989.
7) Hayes, P.C.Jr, Bruce, M.L., Stevens, M.W., Test Method to Extract TPHs from Soil,
American Environmental Laboratory, 12/90, pp. 25-28.
i-123
-------
Table #4. Plackett-Burman #2 Results*
Factors
Pressure
Oven
temperature
Equilibrium
time
Extraction
time
Mass of
sample
Type of
extraction
Type of
matrix
Orientation of
extraction cell
Recoveries
run
order
Standard
Order
Units
Atmosphere
Degree-C
Minute
Minute
Gram
Percent
1
9
360
60
10
10
1
Static
Sand
Horiz
152
2
1
360
60
5
20
3
Dyn
Kaolin
Horiz
40
3
6
150
35
5
20
1
Dyn
Sand
Horiz
154
*These recoveries are not background corrected
Kaolin has been found to have a hydrocarbon co
4
2
150
60
10
10
3
Dyn
Sand
Horiz
68
5
4
150
60
5
20
3
Static
Sand
Vert
86
6
7
360
35
5
10
3
Static
Sand
Vert
38
7
3
360
35
10
20
1
Dyn
Sand
Vert
164
8
11
360
35
10
20
3
Static
Kaolin
Horiz
37
9
12
150
35
5
10
1
Static
Kaolin
Horiz
76
10
5
150
35
10
10
3
Dyn
Kaolin
Vert
59
11
8
360
60
5
10
1
Dyn
Kaolin
Vert
102
12
10
150
60
10
20
1
Static
Kaolin
Vert
150
by subtraction of native contaminates from sand or Kaolin.
ntamination of up to 40 mg/kg by other extraction methods (7).
-------
59 APPLICATION OF SUPERCRITICAL FLUID EXTRACTION OF
DIOXINS/FURANS FROM SOIL AND PUF
Jong-Pyng Hsu, Ph.D.t Joseph C. Pan, Ph.D.
Kevin Villalobos, Gregory P. Miller
Southwest Research Institute
6220 Culebra Road
San Antonio, Texas 78228-0510
The supercritical fluid extraction of dioxins/furans from soil and polyurethane foam plug (PUF) is always
interesting in the environmental application.
In this study, dioxins/fiirans spiked on soils or PUFs will be evaluated. The soils or PUFs will be
extracted with Suprex supercritical fluid extractor using carbon dioxide. The final extract will be analyzed
by a GC/MS.
The extract can also be directly transferred from the supercritical fluid extractor to a GC/MS for the
purpose of reaching lower detection limits.
For both cases, five concentrations of target compounds will be evaluated to determine the detection limit
and linearity of the entire system.
1-125
-------
60 THE APPLICATION OF SUPERCRITICAL FLUID CHROMATORAPHY
TO THE ANALYSIS OF HERBICIDES AND PESTICIDES
IN TCLP EXTRACTS
Charles R. Hecht. Senior Chemist Methods Development, Dr. Peter A. Pospisil,
Manager Methods Development, Matthew A. Kobus, Chemist Methods
Development, Dr. Mark F. Marcus, Director of Analytical Programs, Chemical
Waste Management, Inc. 150 West 137th Street, Riverdale, IL 60627
ABSTRACT
Supercritical fluid chromatography with electron capture detection was used to
consolidate two hazardous waste methods for pesticides and herbicides, in TCLP
extracts, into a single cost effective protocol.
Herbicide and pesticide analysis of TCLP extracts of hazardous waste is currently
performed using SW-846 Methods 8150 and 8080 respectively. Both methods utilize
different sample preparation techniques. Method 8150 incorporates ether
extraction, caustic hydrolysis and diazomethane esterification, while a selection of
different preparatory methods can be utilized for Method 8080. The overall
methodology required for the analysis produces a turnaround time of up to two
days, for a group of 6 to 8 samples.
A single SFC-ECD analytical method, with a consolidated sample preparation
procedure, can be applied to the analysis of both the herbicides and pesticides in
TCLP extracts. The technology is rugged enough to handle the compound type
distribution of both the analytes and the typical interferences found in hazardous
waste samples. Chromatograms, response factors, and the SFC mass spectroscopic
data used to confirm the identity of the peaks will be presented. The application of
SCF-ECD technology to the consolidation of these methods clearly reduces the
costs and sample turnaround times.
INTRODUCTION
Pesticides and herbicides are analytes of major concern to regulatory agencies. The
analysis of the pesticides Lindane, Heptachlor, Chlordane, Endrin, Methoxychlor,
and Toxaphene, coupled with the chlorophenoxy herbicides 2,4-D and Silvex, are
now required for TCLP extracts. Two different methods are used for these analytes:
SW-846 Method 8080 for the pesticides and Method 8150 for the herbicides. The
sample preparation procedures for both methods are significantly different, labor
intensive, and hazardous. Both methods require multiple separatory funnel
extractions using ethyl ether or methylene chlonde. Increasingly health, safety and
environmental concerns are being raised from the usage and disposal of these and
other hazardous solvents. Also of prime concern is the usage of diazomethane as a
methylating agent for the chlorophenoxy acid herbicides. The Merck Index, Edition
10 lists diazomethane as a very toxic, insidious poison, that may explode upon
heating or contact with rough glass surfaces.
1-126
-------
page 2
PURPOSE
The purpose of this work is to develop a single streamlined method utilizing SFC-
ECD, which can determine both pesticides and herbicides in TCLP extracts. The
simplification of the extraction, hydrolysis and derivatization steps will increase
analyst safety, reduce hazardous solvent usage, and significantly reduce the
analytical costs and sample turnaround time.
SCF THEORY
The molecules of a liquid are bonded by electrostatic forces. The energy required to
break these associative bonds, as the liquid becomes a vapor, is known as the heat of
vaporization. When a liquid, in equilibrium with its vapor, is sealed in a tube and
heated, the pressure of the closed system rises and the liquid's density and heat of
vaporization decrease. When the associative forces reach zero, the liquid and gas
phases become one. This temperature and the corresponding pressure are unique
for each liquid and termed the critical constants, which for CU2 are 31ฐC and 73
atm., respectively.
A supercritical phase has the solvency of a liquid. It can dissolve, and thus partition
the analyte(s) between the mobile and stationary phase. It has the low viscosity and
high diffusion coefficient of a gas, resulting in low column pressure drops and rapid
mobile/liquid phase equilibration. The chromatographic efficiencies approach those
of GC. The technique is not thermally driven thus it is also possible to analyze
thermally labile and non-volatile materials. A supercritical fluid, therefore combines
me best qualities of a gas and liquid in a single process. Carbon dioxide is the most
popular material used for extraction because of its low critical temperature,
inertness, safety and ease of purification.
The solvency of the mobile phase can vary and is a function of its density. Density
programming has the same effect on an SFC separation as temperature and solvent
composition have on GC and LC. When utilizing density programming to improve a
separation, the system controller must vary the pressure to linearize the density.
INSTRUMENT DESCRIPTION
The supercritical fluid instrumentation used for this work was purchased from the
Lee Scientific Company, the mass spectrometer from the Finmgan Company, and
the electron capture detector from the Hewlett Packard Company. Two instruments
were used for the study and were configured as follows:
Instrumental SFC-MS
Lee Scientific Model 600 SFC pump, oven and controller interfaced
to a Finnigan INCOS-50 Mass Spectrometer via a heated transfer line.
Transfer line manufactured by Lee Scientific.
11-127
-------
page 3
Restrictor - Fifty micron fused silica frit restrictors producing a carbon
dioxide linear velocity of 0.6 cm/sec at 75ฐC and 75 atm were used for
all SFC-MS work. Restrictors were purchased from Lee Scientific.
Instrument #2 SFC-ECD
Supercritical Fluid Chromatograph - Lee Scientific Model 600 SFC pump
and controller interfaced to a Hewlett Packard 5890 GC equipped
with an BCD detector.
Restrictor - Fifty micron fused silica frit restrictors producing a carbon
dioxide linear velocity of 1.8 cm/sec at 75ฐC and 75 atm were used for
all SFC-ECD work.
Liquid Carbon Dioxide - Supercritical grade. The tank must contain a dip tube to
deliver liquid product (Scott Specialty Gases).
EXPERIMENTAL PROCEDURE AND DATA REVIEW
EQUIVALENCY SUPPORT DATA
Analyte Recovery from The Empore
Solid Phase Disk
In order to evaluate the performance of the method, initial studies were done using
blank TCLP extraction fluid, and extracts from a variety of sample matrices.
Recovery data was generated for the pesticides, herbicide acids, and their respective
methyl esters. The Empore disk extraction, hydrolysis, and esterification efficiencies
were determined. Recovery data is presented in Figure 1. The analyte recovery
range of 59% to 117% shows that the Empore Solid Phase disk is an acceptable
means of analyte concentration.
Included in the study were sample types such as: filter press cake, oil dry + gas,
hydrocarbon contaminated soil, dimethyl disulfide spill debris, terminal plant
sludge, grease, spent carbon filter media, and others.
Analyte Chromatographv and Mass
Spectrometric Verification
In order to determine that no analyte alteration, reaction or modification occurred
during the supercritical fluid chromatographic procedure, the chromatograph was
linked to a mass spectrometer for spectral confirmation. The column selected for
this work was a Phenyl-5, 10 meter, 50 micron ID, haying a 0.25 micron film
thickness. A Finnigan INCOS-50 was used for the confirming spectrometer. Figures
2 and 3 show the SFC-MS total ion chromatogram and library matched spectra
produced by the run. The NIST library searches for all of the analytes produced
library match factors of 800+. The data show that no analyte alteration, reaction or
degradation occurred in the system.
1-128
-------
page 4
EQUIVALENCY METHOD DATA
CWM Combined Herbicide - Pesticide Sample Preparation
The first part of the method involves the two part streamlined sample preparation
procedure. Both procedures are compared with the corresponding SW-846 methods
in Figures 4 and 5.
1 - One hundred mL of TCLP extract, including the surrogate, is made alkaline
using a KOH solution. The extract is stirred at 70ฐC for one hour. The
extract is reacidified to a pH of 2 and passed through an Empore solid phase
extraction disk, which removes the herbicide acids. The acids are eluted from
the disk with methanol. Methane sulfonic acid is added to the extract and the
extract solution is heated for one hour to esterify the acids and reduce the
solution volume.
2- A second 250 mL volume of TCLP extract, including surrogates, is
neutralized with KOH and passed through a second Empore solid phase
extraction disk. The pesticides are eluted using acetone. The extract solution
volume is reduced using nitrogen blowdown.
The two extract solutions are combined, Aldrin is added as an internal standard and
the analysis is performed on the combined extract.
SFC-ECD Chromatography
An SFC-ECD chromatogram of the herbicide methyl esters and pesticides is shown
in Figure 6. All of the peaks are sharp, baseline resolved and of approximate equal
intensity.
Initial studies were performed using a mass spectrometer but to simplify the method
even further, final studies were done using the ECD detector. All work was done
under isothermal, density programmed chromatographic conditions. They are as
follows:
Oven temperature 100ฐC isothermal
Initial density 0.13 (equivalent to 74 atm)
Hold 2 minutes
Ramp at .006 to 0.42 (equivalent to 178 atm)
Ramp at 0.2 to 0.75( equivalent to 390 atm)
Final Hold Time 7 minutes
Response factor, relative retention time and method detection limits are listed in
Figure 7.
1-129
-------
page 5
EPA Equivalency Study Data Generation
To begin formal EPA Equivalency spike recovery studies have been done on a wide
variety of sample types. To this date six samples have been completed in duplicate.
The six sample types include incinerator ash, wastewater treatment sludge, plating
filter cake, soil contaminated with 1,1,1-TCE, grease/oil debris, and copper
reclamation tailings.
Recovery data from these six sample types is listed in Figure 8. The data quality is
high and meets formal EPA Equivalency requirements.
CONCLUSIONS
1 - A single SCF method requiring less than 5 hours has been developed for the
analysis of both herbicides and pesticides in TCLP extracts, at detection
levels well within the regulatory range.
2- The data generated shows that it meets the EPA's formal Equivalency
criteria.
3 - The use of ether and methylene chloride is eliminated. No explosive
derivatization agents are used, and general solvent use is reduced by a factor
of 5.
4- The analyte recovery and SFC-MS support data show that there is no
chemical reactivity between the analytes and the supercritical phase.
11-130
-------
FIGURE 1
EQUIVALENCY SUPPORT DATA
% RECOVERIES
U
LINDANE
HEPTACHLOR
g-CHLORDANE
a-CHLORDANE
ENDRIN
METHOXYCHLOR
TOXAPHENE
2,4-D
SILVEX
TCMX(SURR)
DBC(SURR)
2,4,5-T(SURR>
Extract 1 Extract 2
91 92
55 56
76 80
82 84
99 98
85 88
81 77
98 100
55 56
54 57
101 102
87 81
Extract 3 Extract 4
92 91
59 49
105 88
108 92
110 100
114 96
148 126
85 118
93 68
61 47
124 133
87 110
Extract 5 Extract 6
94 92
58 60
73 75
78 80
103 93
" 93 87
123 140
91 88
51 61
50 79
98 95
88 107
Extract 7 Extract 8
94 92
65 63
76 77
81 80
102 90
106 93
139 102
95 106
79 84
57 51
99 95
121
Extract 9 Extract 10
92 91
60 62
75 71
78 77
99 98
96 89
115
90 99
95 99
48 48
96 92
AVERAGE
92
59
80
84
99
95
117
97
74
55
103
97
-------
1C QATfti Slซt2 II
12/tl/ป 13i37iซ CM.Ii CM.TM 13
SffTUEl E8UIU STB 12739-2^27 St-TWtf
SCMS UN T023M
1N.*-
tatt
3276M.
FIGURE 2
SFC-MS
PESTICIDES
43il9
47t39
SCON
urn
RIC ปTte SSli files SOWS 9M TO 14ซ
SMVLEt WMH MMOT B1PW4VL FULL FRIT
CQKK.i l.WL LOOP 1GMDW CT/OIP IH^.1* IHJ
WMZl C 1,1X3 UซLl H ป. ซ,ป QUNNi ซซ, l.i J BOSEt U 21,
1M.I
SFC-MS
TOXAPHENE
131C
18M
36il6
11W
44t2ซ
1298
48i22
13M
52i24
14W S0*i
56.23 TIIC
11-132
-------
1*8
I 288*9
HID LIBMRV SEMCH (UMNmซ>
12/18/38 UiXiM * 39i2C
Smil EOUIU STD 57-19WW 82739-2^27
COWS, i 13NOW PWซ.-Vtl ซTQป RET 0
11818 TO 11025 9MB) - 113*1 TO 11814
DOTfli S1831 11821
CM.Ii CM.TMI 3
BOSC
own
V2 38
L1WMC
SMVLE MINUS LIBMtY
A ,li i 11 iJ ซ MI J - ,1.1
191
FIGURE 3
SFC-MS
LIBRARY MATCH
SPECTRA
399
HID LIMWtt SEMCH (LIBWTM) MTfli S787 8112*
Ili26i88 * 27i99 CM.li COLTAi 3
K 11739-17-9
COKB.I 1788BW PtOm.-9 18KTB RET GflP
SMflf
., 1 .. ll IL.
K/2i 199
RICl 349184.
C9.H8.03.CL2
I 21788
FUR 722
2,4-0 nmm. Esmt
ll II, 1. II, i, ll
SM*U MINUS LIMRY
II p
IM l.ll
11-133
-------
FIGURE 4
FLOWCHART COMPARISON BETWEEN
EPA METHOD 8150 and CWM HERBICIDE METHOD
FOR TCLP EXTRACTS
EPA METHOD 8150
1) ETHER EXTRACT AT pH=2
1 time at 150 mL
2 times at 50 mL
2) HYDROLYSIS
2mLof37%KOH
3) ETHER EXTRACT AT pH = 11
2 times at 20 mL
4) ETHER EXTRACT AT pH=2
1 time at 20 mL
2 times at 10 mL
2mLof(l:3)H2SO4
CWM METHOD
NONE REQUIRED
TCLP EXTRACT HYDROLYSIS
2mL37%KOH
NONE REQUIRED
NONE REQUIRED
EMPORE EXTRACT AT pH=2
21mLofMETHANOL
2mLof(l:l)H2SO4
EMPORE ANALYTE ELUTION
15mLofMETHANOL
5) EXTRACT TRANSFER TO K-D
1 time at 30 mL
6) INITIAL K-D
7) MICRO K-D
8) DIAZOMETHANE ESTERIFY
NONE REQUIRED
MSA ESTERIFICATION
1001L of MSA
9) GC-ECD ANALYSES
SFC-ECD ANALYSES
COMBINED EXTRACT
(See Figure 6)
11-134
-------
FIGURES
FLOWCHART COMPARISON BETWEEN
EPA METHOD 8080 and CWM PESTICIDE METHOD
FOR TCLP EXTRACTS
EPA METHOD 8080
1) SEPARATORY FUNNEL EXTRACT
3 times at 60 mL MeCb
NEUTRALpH
2) K-D TO 5 mL
3) NITROGEN BLOWDOWN
4) SOLVENT EXCHANGE
9mLHEXANE
5) GC-ECD ANALYSES
CWM METHOD
EMPORE DISK EXTRACT
20 mL ACETONE
NEUTRAL pH
ANALYTE ELUTION
15 mL ACETONE
NITROGEN BLOWDOWN
NONE REQUIRED
SFC-ECD ANALYSES
COMBINED EXTRACT
11-135
-------
AN .ป! 132964820. BNC
RUN ป 168 flAY 13. 1991 19I19I24
START
FIGURE 6
SFC-ECD
CHROMATOGRAPHY
11.331
~ 11.142
ฃ9.ias Z.'t-O Methyl Ester
1.093
39.722
44.720
30.957 Tetra-Chloro-M-Xylene (Surr)
31.923 Silvex Methyl Ester
33.013 2,4,5-T Methyl Ester (Surr)
34.778 Lindane
38.394 Heptachlor
40.849 Aldrin (interns: 31.)
""~~ 42.398 Ga 1MB-Chi".' !3fl^
43. 899
43.372 Endrin
48. a is Methoxychlor
40.084 Dibutylchlorendate (Surr)
TIMETABLE STOP
11-136
-------
FIGURE?
RESPONSE FACTORS , RELATIVE RETENTION TIMES
AND METHOD DETECTION LIMITS(NG/ML)
TARGET COMPOUNDS
RF RRT MDL
2,4-D Methyl Ester .267 .712 10
Silvex Methyl Ester .949 .781 10
Lindane .981 .851 0.5
Heptachlor 1.20 .939 0.2
g-Chlordane 1.13 1.05 0.2
a-Chlordane 1.24 1.07 0.2
Endrin .992 1.11 1.0
Methoxychlor .525 1.19 0.6
Toxaphene(4 Peaks) .140 1.13 75
l'.26
1.31
SURROGATES
2,4,5-T Methyl Ester .885 .808 NA
Tetra-chloro-m-xylene 1.06 .758 NA
Dibutylchlorendate .866 1.17 NA
INTERNAL STANDARD 0.5 PPM ALDRIN
11-137
-------
FIGURE 8
TCLP EXTRACT SPIKE RECOVERY DATA
DUPLICATE RESULTS
00
LINDANE
NEPTACHLOR
g-CHLOftDANE
a-CHLOROANE
EHDRIN
NETHOXYCHLOR
TOXAPHENE
2.4-D
SILVEX
TCHX(SURR)
DBC(SURR)
2,4,5-T(SURR)
Uastewater
Treatment Sludge
94 99
86 93
92 94
92 95
111 106
108 108
121 126
93 96
86 81
75 87
101 102
83 96
Soil Cont.
With 1,1,1-TCE
106 98
88 68
96 78
100 84
108 92
121 110
118 90
48 82
92 66
86 68
110 100
38 75
Plating Filter
Cake
94 94
76 79
89 88
90 88
100 98
110 110
59 104
94 84
85 88
59 70
100 100
90 93
Incinerator
Ash
99 97
75 84
86 92
90 95
98 98
104 104
108 102
97 84
86 94
76 78
104 102
95 94
Grease/Oi I
Debris
76 71
76 98
86 82
90 91
88 89
102 100
78 128
76 83
81 93
82 84
94 90
81 88
Copper Reclaimation
Tailings
80 90
77 84
80 84
78 78
78 86
92 100
84 99
94 83
84 86
76 88
88 94
90 91
AVERAGE
92
82
87
89
96
105
101
84
85
77
98
84
-------
61
PROBLEM SOLVING IN THE ORGANIC EXTRACTIONS LABORATORY;
HERBICIDES
John Doeffinger, Teresa Wittwer, Jim Giannella, Chris Lott,
Larue Stanton, Erik Alverson, Sean Fitzgerald, Kristine
Klinger, Deborah Smith. Ph.D.. IEA, Inc. - New Jersey, 628
Route 10, Whippany, New Jersey 07981
ABSTRACT
In the commercial laboratory, it is often difficult to set
aside time and resources to improve and optimize execution of
acceptable methods without a dedicated "special projects"
group. At IEA, Inc. - New Jersey, method development for
sample preparation is carried out by the Organic Extractions
Group as a whole, from experimental design through data
interpretation, within the normal flow of production
laboratory work. This problem-solving process, with
supporting data, including matrix spike and surrogate
recoveries of real samples, is illustrated for the extraction
of herbicides.
INTRODUCTION
The "routine" extraction of samples for herbicide analysis by
SW846 methods has presented difficulties for the laboratory.
Problems including the use of the reagents diazomethane and
ethyl ether in a production environment, the inconsistency of
spike and surrogate recoveries, and the elaborate sample
manipulations required, result in a procedure that is time-
consuming and frustrating. IEA, Inc. - NJ has optimized an
existing trial USEPA method1 which addresses the compounds
2,4-D, Silvex, and 2,4,5-T. We present here results of this
optimized IEA method for water and leachate samples and show
evidence that it is adaptable to a variety of matrices.
METHOD DEVELOPMENT
Preliminary trials using the above mentioned USEPA method as
written yielded poor surrogate and spike recoveries. However,
we were interested in pursuing optimization of this method
because of the many advantages it offered. A general meeting
of the entire extractions staff, QA, and laboratory management
was called to organize a systematic approach for method
optimization. Data from extractions using the trial USEPA
method were discussed and all agreed upon the next course of
action, which was a limited experiment to be conducted along
with normal extraction batches. The group reconvened the next
11-139
-------
week to discuss the results of the experiment. This process
evolved into an on-going program of weekly meetings followed
by limited experiments, which resulted in method definition
and refinement.
EXTRACTION OF HERBICIDES FROM AQUEOUS SAMPLES
Poor performance (shown in Table 1.0) of the trial USEPA
method forced critical examination of the variables displayed
in Table 2.0.
TABLE 1.0
TYPICAL SURROGATE AND SPIKE RECOVERIES
USING USEPA TRIAL METHOD1
SAMPLE
LEACHATE 1
LEACHATE 2
LEACHATE 3
LEACHATE 4
LEACHATE 5
LEACHATE 6
LEACHATE 7
2,4-DB
SURR.
59
73
78
97
100
30
54
2,4-D
60
52
36
SILVEX
66
54
59
2,4,5-T
49
45
91
11-140
-------
TABLE 2.0
VARIABLES INVESTIGATED FOR OPTIMIZATION
OF HERBICIDE EXTRACTION
1.0 Acid-Washing of all glassware and materials
2.0 pH (2, 3, <2)
3.0 Use of microsnyders
4.0 Concentration of extract to dryness prior to
esterification
5.0 Use of Na2SO4 column cleanup
6.0 Presence of acetic acid in TCLP leachates
7.0 Use of BC13 versus BF3
8.0 Temperature for esterification
9.0 Methylene Chloride volume versus sample volume
10.0 Alkaline hydrolysis required?
We found that acid-washing of all equipment was most crucial
to the success of the extraction, and that aqueous samples
should be taken to pH 1. Neither use of microsnyders nor
concentration to dryness improved recoveries. Also, we found
Sodium Sulfate column clean-up to be unnecessary. Since
leachates performed better than "plain" aqueous samples, we
investigated the addition of acetic acid to the water samples,
but the recoveries were unaffected. Best recovery occurred
when esterif ication was carried out with BF3 at 60ฐ C. We are
still optimizing sample and solvent volumes, and investigating
the necessity of the alkaline hydrolysis. The IEA method is
summarized in Table 3.0:
11-141
-------
TABLE 3.0
METHOD FOR HERBICIDE EXTRACTION OF WATER SAMPLES
1
2
3
4
5
6
7
8
9
0
0
0
0
0
0
0
0
0
Acid wash all glassware and materials
Adjust pH of sample to < 2.0
Extract 500 ml sample three times with 60/40/40 ml
methylene chloride
Concentrate extract to 4.0 ml
Solvent exchange with hexane
Concentrate extract to 1.0 ml
Esterify with BF3 at 60ฐ C for 10 minutes
Dilute to 5.0 ml with hexane
Add 10 ml 7% NaS0, vortex
. ?4,
10.0 Collect 1.0 ml hexane extract for analysis
We tested this IEA method for extraction of the most recent
USEPA WS series proficiency samples; the results are shown in
Table 4.0.
TABLE 4.0
PROFICIENCY RESULTS: WS027
ANALYTE
2,4-D
SILVEX
REPORTED
45.9
17.9
TRUE
46.3
18.1
ACCEPTANCE LIMITS
15.1 - 59.1
7.47 - 24.4
The majority of aqueous herbicide analyses requested are for
TCLP leachates; surrogate recovery results for leachate
blanks and samples are presented in Figures 1.0 and 2.0; spike
and surrogate recovery data are presented in Table 5.0.
Results of similar analyses of a series of spiked reagent
blanks are shown in Table 6.0. Note that leached samples
consistently performed better than water samples.
11-142
-------
TABLE 5.0
FREE ACID SURROGATE AND SPIKE RECOVERIES
TCLP LEACHATES
SAMPLE
LEACH BLANK
LEACH BLANK SPIKE
LEACHATE X3
LEACHATE X3 MS
LEACHATE BLANK
LEACH BLANK SPIKE
LEACHATE WP1
LEACHATE WP1 MS
LEACH BLANK
LEACH BLANK SPIKE
LEACHATE WP9
LEACHATE WP9 MS
LEACH BLANK
LEACH BLANK SPIKE
LEACHATE D3
LEACHATE D3 MS
LEACHATE D4
LEACHATE D4 MS
LEACH BLANK
LEACH BLANK SPIKE
LEACHATE 191
LEACHATE 191 MS
LEACHATE 380
LEACHATE 380 MS
2,4-DB
100
104
106
109
95
97
98
93
116
101
114
120
75
89
78
85
94
104
85
79
80
77
70
70
2,4-D
83
95
83
82
88
105
65
64
72
81
75
72
SILVEX
85
94
83
77
89
110
71
75
74
66
69
58
2,4,5-T
66
78
67
66
76
89
59
57
61
55
54
53
1-143
-------
TABLE 6.0
FREE ACID SURROGATE AND SPIKE RECOVERIES
REPLICATE BLANK ANALYSIS
SAMPLE
BLANK
BLANK SPIKE 1
BLANK SPIKE 2
BLANK SPIKE 3
BLANK SPIKE 4
BLANK SPIKE 5
BLANK SPIKE 6
2,4-DB
99
104
79
108
112
104
114
2,4-D
61
44
67
53
60
68
SILVEX
83
58
82
75
80
91
2,4,5-T
49
29
51
38
42
53
Although there were incidents of low spike and surrogate
recoveries, the results were generally good. Lower recoveries
were consistent throughout a complete batch, indicating that
an isolated extraction procedure, not the method, had
performed poorly. The compound 2,4,5-T was the poorest
performer throughout the aqueous studies, with the lowest
recoveries noted in spiked reagent blanks.
EXTRACTION OF HERBICIDES IN ESTER FORM
We suspected that herbicides in various ester forms would not
be converted efficiently to the methyl ester, since the
alkaline hydrolysis step was omitted4. An extraction of
reagent water spiked with the propylene glycol butyl ether
ester of Silvex (Silvex PGBE) was carried out without
performing an alkaline hydrolysis step. The results of this
preliminary investigation are presented in Table 7.0 with an
example chromatogram shown in Figure 3.0.
11-144
-------
TABLE 7.0
PERCENT RECOVERIES OF ESTERS OF HERBICIDES
(BLANK SPIKES)
SAMPLE
BLANK
BUTYL ESTER,
BUTYL ESTER,
SILVEX
SILVEX DUP
2,4-DB
(SURR.)
109
SILVEX
26
24
The recovery of Silvex PGBE as a methyl ester was marginal,
but we feel that additional development work will result in
improved recoveries. Further experiments are in progress to
determine the conversion efficiency of a greater variety of
esters.
EXTRACTION OF HERBICIDES FROM ORGANIC MATRICES
IEA, Inc. - NJ is frequently called upon to perform the TCLP
on organic matrices, but we had been unable to carry out
herbicide analysis on this matrix type. We tried an approach
similar to a BNA partition 3. First, the sample is washed
with a basic aqueous solution to separate all acids into the
water layer. The water fraction is then acidified and
extracted with methylene chloride. The methylene chloride
extract is then subjected to the IEA herbicide procedure. The
results presented in Table 8.0 indicate that analysis of
herbicides in organic matrices is meaningful. This capability
is important to clients because it permits complete sample
characterization; inability to test the herbicide fraction
allows potential classification of a sample as a hazard.
1-145
-------
TABLE 8.0
PERCENT RECOVERY OF HERBICIDES IN AN ORGANIC MATRIX
SAMPLE
BLANK
BLANK SPIKE
SAMPLE 1
SAMPLE 2
SAMPLE 3
2 , 4-DB
96
82
69
96
82
2,4-D
64
61
88
52
SILVEX
59
48
54
47
2,4,5-T
52
51
57
44
In this example:
Sample 1 -
Sample 2 -
Sample 3 -
One gram of motor oil was
with methylene chloride,
during dilution.
One gram of motor oil was
with methylene chloride,
during dilution.
One gram of motor oil was
with methylene chloride,
diluted to 10 ml
spike was added
diluted to 25 ml
spike was added
diluted to 25 ml
spike was added
directly to motor oil, prior to dilution.
EXTRACTION OF HERBICIDES FROM SOIL/SEDIMENT MATRICES
Most of our non-TCLP requests for herbicide analysis are for
soils; therefore we wanted to extend the IEA herbicide
procedure to soil analysis. Data from preliminary trials of
the IEA method are shown in Table 9.0. Note that the soil
sample showed better spike recovery than a blank sand matrix.
Further development is in progress.
1-146
-------
TABLE 9.0
FREE ACID SURROGATE AND SPIKE RECOVERIES
SOILS
SAMPLE
BLANK (SAND)
BLANK SPIKE
SOIL Dl
SOIL D2
SOIL Ul
SOIL Ul MS
SOIL Ul MSD
SOIL U2
SOIL BK
SOIL Rl
SOIL R2
2,4-DB
37
56
49
82
73
75
81
77
66
98
79
2,4-D
52
98
94
SILVEX
60
115
113
2,4, 5-T
93
91
86
CONCLUSIONS
Because the demand for herbicide analysis fluctuates and the
scope of requested analytes is limited, a method that can be
performed without a period of fine-tuning required to achieve
acceptable recovery is needed. The IEA method described here
has the advantages of simplicity, use of routine reagents, and
improved reproducible analyte recoveries. Production is
doubled. Along with the development of the extraction
procedure, there were many intangible benefits to working on
the problem as a group. However, the ultimate success will be
in the achievement of full method approval.
1-147
-------
SUMMARY
The concept of using regular group problem-solving sessions
has resulted in development of a simple, effective methylene
chloride extraction for the determination of herbicides.
Generally acceptable spike and surrogate recoveries of
commonly requested herbicide analytes in aqueous, leachate,
soil, and organic matrices were achieved. This method is
easily implemented and results in increased capacity.
REFERENCES
Analysis of Pesticide Residues in Human and Environmental
Samples, June 1980, "Determination of Some Free Acid
Herbicides in Water", USEPA.
Analysis of Pesticide Residues in Human and Environmental
Samples, June 1980, "Sampling and Analysis of Water for
Pesticides", USEPA.
Test Methods for Evaluating Solid Waste, Volume IB, SW846
3rd Edition, Method 3650 "Acid-Base Partition Cleanup",
USEPA
Analysis of Pesticides in Water, Volumes I - III, CRC
Press, 1986.
1-148
-------
o
u
K
U
hJ
C
130
140
130
120
110
100
90
80
70
60
50
40
30
FIGURE 1.0
2,4DB Surrogate Recovery
Leach Blank/Leach Blank Spike
2,4-DB Surrogate Recovery
Samola/SamDle Soike
UHU I I I I
20
40
60
Sample No.
11-149
-------
l'\
FIGURE 3
SILVEX
BUTYL ESTER
CN
s
CN
>Q GO O> CNOQD J I CD O O II f\ h^ป
cH sO O ซH |^> I \ XT CD
?H
rH rH CN CN CN
-------
CN
r\
FIGURE 4
HERBICIDES
ORGANIC MATRIX
-------
52 Infrared Microsampling for the Qualitative Analysis of
Organics Extracted from Soil Samples
M.P. Fuller and F.J. Weesner
Nicolet Instrument Corporation, Spectroscopy Research Center,
5225 Verona Road, Madison, WI 53711
Infrared spectroscopy provides a unique "fingerprint" that can often
be used to identify the structure of unknown compounds. Often
spectral library searches are used to make this type of identification.
The primary difficulty in determining the identity of unknowns in
this manner is usually spectral contamination caused by the
absorbances of components other than the target compound. It is
possible to separate many organic compounds using thin layer
chromatography (TLC) techniques. The relative elution distance can
then be related to separations of standard mixtures and the structure
of unknown compounds elucidated. Many times, however, it is
impossible to absolutely identify components in this manner.
Recently we have investigated the use of TLC separations combined
with infrared microspectroscopy for the identification of organic
compounds extracted from soil samples. The soil extract is
separated using standard TLC procedures and the resulting "spots"
are analyzed both directly on the plates and after extraction with
appropriate solvents. The results of these experiments will be
described and compared with GC/FT-IR measurements obtained on
the soil extracts.
11-152
-------
63
A PERFORMANCE COMPARISON STUDY OF DIFFERENT TYPES OF DEVICES FOR SOLID PHASE
EXTRACTION
Y. Joyce Lee. E. Neal Amick, Jack A. Serges, Lockheed Engineering & Sciences
Company, Las Vegas, Nevada
Gary L. Robertson, U.S. Environmental Protection Agency, Environmental
Monitoring Systems Laboratory, Las Vegas, Nevada
ABSTRACT
Solid Phase Extraction (SPE) is rapidly becoming an alternative to separately
funnel extraction and continuous liquid-liquid extraction for the isolation of
organic compounds from environmental samples. SPE can provide analytical data
in a timely manner for decision-making during site inspections, remediations,
and emergency removal activities. It is especially useful if there is
knowledge regarding potential matrix interference at the site. SPE utilizes a
compact manifold that can process multiple samples simultaneously. Solvent
usage is minimized and the sample preparation can be performed rapidly. The
extraction can be accomplished by using glass cartridges, plastic cartridges,
or extraction disks. Characteristics of these extraction devices, including
recovery, capacity, interferences, and contamination for polycyclic aromatic
hydrocarbons, phenols, pesticides, and Aroclors have been compared and will be
discussed. The performance results presented are data generated as part of
the Superfund Contract Laboratory Program Quick Turnaround Method development
and validation process.
Notice: Although the research described in this article has been funded
wholly or in part by the United States Environmental Protection Agency through
contract number 68-CO-0049 to Lockheed Engineering & Sciences Company, it has
not been subjected to Agency review and therefore does not necessarily reflect
the views of the Agency and no official endorsement should be inferred.
11-153
-------
STANDARD REFERENCE SPECTRA for MS/MS QUALITY ASSURANCE,
PERFORMANCE EVALUATION, and PROFICIENCY TESTING:
XC rf1Q TANDEM MASS SPECTROMETERS
Richard I. Martinez, Research Chemist, Chemical Kinetics and
Thermodynamics Division, National Institute of Standards and
Technology, Gaithersburg, Maryland 20899
ABSTRACT
The collisionally-activated dissociation (CAD) of the
acetone cation (m/2 58) can be used for quality assurance,
performance evaluation, and proficiency testing of CAD
measurements in tandem mass spectrometry ( MS/MS) instruments
which use rf-only multipole collision cells. The absolute
branching ratios (product distributions) of the CAD fragment
ions, when measured as a function of the center-of-mass
collision energy E , can provide an objective basis for
quality assurance wnenever MS/MS methods are used (viz., to
validate how well the target thickness, ion containment
efficiency, and collision energy are being controlled in
various instruments).
INTRODUCTION
Tandem mass spectrometry (MS/MS) instruments which use rf-
only muljj^inole collision cells are complex ion-optical
devices. Such MS/MS instruments are denoted hereinafter
by the generic symbol XIrf1Q, where Q denotes a quadrupole
mass filter, Irf1 denotes an rf-only multipole collision
cell used for collisionally-activated dissociation (CAD),
and X can be either a Q or a sector analyzer (denoted by EB
or BE) . There are several types of Xtrf]Q MS/MS instruments
(e.g., QqQ, BEqQ, QoQ, QhQ, etc.; here q, h, and o denote,
respectively, rf-only collision cells which use quadrupole,
hexapole, and octopole rod assemblies). There are currently
more than 400 XCrf]Q instruments worldwide, representing a
capital investment of more than $200M.
In this note we discuss an objective basis for quality.
assurance of CAD measurements in "dynamically-correct"
X[rf]Q instruments. The practical tuning criteria and
guidelines herein can be used routinely (e.g., on a daily
basis) to check instrument and/or operator performance once
it has been certified ( cf. section 4a of ref. 10) that
1-154
-------
Q
dynamically-correct product distributions can be measured
in any particular instrument.
Technical Background:
To study ion-neutral reaction mechanisms in XIrf]Q
instruments, it is crucial that one measure dynamically-
correct product distributions which are instrument
independent. Otherwise one may measure a distorted
representation of the reaction dynamics, which consequently
can lead to incorrect conclusions about the pertinent
reaction mechanisms. The dynamical prerequisites for
obtaining dynamically-correct product distributions
(branching ratios) within Xtrf] Q instruments have been
detailed elsewhere.
It follows, therefore, that to develop an instrument-
independent database (or library) for MS/MS measurements
within X( rf]Q instruments one must obtain substantially the
same representation for any reaction [e.g., CAD] occurring
within any such instrument (i8e.0 no discrimination effects;
see Appendix of reference 9) . '
A measurement protocol was developed at the National
Institute of Standards and Technology (NIST; formerly
National Bureau of Standards) to provide a basis for precise
and accurate (ฑ10%) instrument-independent, dynamically-
correct measurements within XqQ instruments. The precepts
of the NIST protocol should also be applicable to other
types of Xtrf) Q tandem mass spectrometers which have strong
focusing properties (e.g., QhQ, QoQ, etc.), so long as the
collision energy range is the same as for XqQ instruments.
The NIST protocol10 was validated by the recent NIST-EPA
International Round Robin which indicated that at least
50% of the QqQ instruments which have been sold and are
currently in the field can provide an instrument-
independent, dynamically-correct representation of any ion-
neutral reaction mechanism when this kinetics-based
measurement protocol is used. Hence, the NIST protocol can
be used to develop an instrument-independent database of CAD
spectra for dynamically-correct X[ rf] Q tandem mass
spectrometers ' (and/or to study the kinetics and
mechanism of ion-neutral reactions). The NIST protocol is
to be incorporated into EPA' s SH-846 Test Methods for
Evaluating Solid Haste as an 8000 series tuning procedure
for dynamically-correct XI rf 1 Q instruments.
11-155
-------
DISCUSSION
After one has proven that a given XIrf]Q instrument is
capable of measuring dynamically-correct product
distributions ( cf. section 4a of ref. 10), it becomes
imperative that standardized'operating conditions be
maintained to,ensure precision and accuracy of the CAD
measurements, especially each time CAD spectra are to be
taken for inclusion in a NISI standardized database.
Maintaining Standardized Operating Conditions:
The CAD of the acetone cation is especially well suited for
the requisite quality assurance, performance evaluation, and
proficiency testing applications because:
(a) it provides a relatively simple test case (if one
cannot generate instrument-independent CAD spectra for
the acetone cation, then it may not be possible to
develop an instrument-independent database under any
conditions), and
( b) there is a wealth of information about the unimolecular
and collisionally-activated dissociation of the acetone
cation. T9 37
and, most importantly, because of the following unique
characteristies:
(c) there are distinct differences in the energy
dependences of the branching ratios obtained under
single-collision (SO vs. multiple-collision
conditions . Therefore, one can readily determine
whether or not the target thickness is within the
single-collision regime. Comparison of the {SO and
data in Tables 1 and 2 of ref. 9 indicated that
the control of the target thickness becomes extremely
critical if one hopes to measure instrument-independent
product distributions (i.e., CAD spectra).
(d) the production of 1 5 is a significant decomposition
channel. This allows one to gaugeghow well the
reaction-induced mass discrimination ' due to CAD is
controlled in various Xtrf1Q instruments (i.e., how
well one can compensate for differences in ion
containment efficiencies, especially for low-mass
11-156
-------
daughter ions (here md9ปghter/!Tparent= 1 ^5?= ซ ซ) 1 .
If one cannot measure^aynamicaliy-correct branching
ratios vs. E for 15 , then one obtains an incorrect
representatiSB of the low-energy CAD mechanism for
Me2CO .
(e) the energy dependence of the branching ratio foe.,
production of 15 goes through a sharp maximum. This
allows one to gauge how well the collision energy is
controlled in various instruments.
It is proposed, thereforeg7that the standard spectra for the
CAD of the acetone cation be used by the MS/MS community
to periodically recheck the performance of dynamically-
correct Xtrf]Q instruments (viz., how well the key MS/MS
parameters such as target thickness, ion containment
efficiency, and collision energy are being controlled in a
XC rf]Q instrument). These standard CAD spectra can also be
used to test the proficiency of Xfrf1Q operators of varying
skill levels, thus providing an objective basis for quality
assurance whenever one uses MS/MS methods such as the EPA' s
SH-846 method.
Reference Spectra for the CAD of the Acetone Cation:
Table 1 shows the absolute branching ratios (from ref. 37)
for the CAD of Me-CO [generated by 70 eV electron
ionization (EIX or acetone!. They were measured with the
NIST protocol in NIST's dynamically-correct QqQ
instrument under single-collision conditions ( Ar target)
at the center-of-mass collision energies (E ) indicated.
The E of Table 1 were selected iteratively to optimize the
information about competitive reaction channels (including
the absolute maximum branching ratio for Me production at
E = 32.6 eV) . The reader is referred to the EXPERIMENTAL
s8
-------
The ketene cation (m/z 42; branching ratios of 0.02-0.06 for
E = 1-60 eV) is a minor CAD fragment.
cm
MS/MS Quality Assurance:
One's ability to reproduce the dynamically-correct branching
ratios shown in Table 1 for the CAD of the acetone cation
should indicate that one's XCrfJQ instrument is functioning
properly, and is ready to measure standardized CAD spectra.
This provides an objective basis for quality assurance of
CAD measurements in "dynamically-correct" XCrf]Q
instruments.
Branching Ratios and Target Thickness:
One should be able to replicate the values in Table 1 to
within the maximum uncertainty indicated by the bracketed
values in Table 1 for branching ratios >0. 01. This would
ensure that the Ar target thickness is within the single-
collision regime.
Collision Energy:
It was shown in ref. 37 that the complementary energy
dependences for production of MeCO and Me are due to a
competition between three fast, primary (direct) reactions,
each of which opens sequentially at its respective threshold
energy [viz., (1), (2), and (3)1.
Me2CO*-ป MeCO* + Me- (X 2A"2> AH= 0. 82 eV (1)
ป Me* + Me- + CO AH= 4.24 eV (2)
* MeCO* + Me- ( B, 1 2A'1) AH= 6.55 eV (3)
That is, the maximum in the branching ratio vs. E curve
for Me production at E = 32. 6 eV corresponds toฐS'he
opening of reaction (3) when the collisionally-activated
Me2CO has acquired an internal excitation E. .- 6.55 eV.
This E. . is corroborated by the increased pro-auction of 42
for E X32.6 eV, which was attributed (in ref. 37) to the
opening of a new direct reaction channel [(5) or (6)]1 for
production of H2C=C=0 .
Me2CO+- > H2C = C = 0+ + CH4 AH= 0.89 eV (4)
~ป H2C=C=0* + H + Me- AH= 5.43 eV (5)
1-158
-------
ป H2C=C = 0 + CH2 + H2 AH= 5.69 eV (6)
Hence, E = 32.6 eV corresponds to E- . - 5.43-6.55 eV [for
reactionim( 3) , (5), (6)]. That is, afi uncertainty in E. .
of ca. 1 eV (=6.55-5.43) corresponds to an uncertainty In
Ecm of ca. 5-6 eV at Ecm= 32/6 eV.
SUMMARY
The absolute branching ratios (product distributions)
for the CAD of the acetone cation (measured as a function of
E ) provide an objective basis for quality assurance,
performance evaluation, and proficiency testing of CAD
measurements in dynamically-correct tandem mass
spectrometers which use rf-only multipole collision cells.
That is, by replicating NIST' s standard reference spectra,
an operator can determine that the key MS/MS parameters
(e.g., the target thickness, ion containment efficiency, and
collision energy) are under control, and that one's Xfrf]Q
instrument is functioning properly and is ready to measure
standardized CAD spectra .
ACKNOWLEDGMENTS
R.I.M. gratefully acknowledges the funding of this
work, in part, by the US Environmental Protection Agency
[the Atmospheric Research and Exposure Assessment Laboratory
(AREAL)] under Interagency Agreement IAG #DH-13934363-1 , and
helpful discussions with Dr. L. D. Betowski ( EMSL - Las
Vegas).
11-159
-------
REFERENCES
1. P. H. Dawson, (a) Quadrupole Mass Spectrometry and its
Applications, Elsevier, Amsterdam, 1976; ( b) Adv.
Electronics Electron Phys. 53. 153 (1980).
2. S. C. Davis and B. Bright; Rapid Commun. Mass Spectrom.
4., 186 (1990), and references therein.
3. P. H. Dawson, Int. J. Mass Spectrom. Ion Phys. 20. 237
( 1976) .
4. P. H. Dawson and J. E. Fulford, Int. J. Mass Spectrom. Ion
Phys. 42. 195 ( 1982) .
5. P. H. Dawson, J. B. French, J. A. Buckley, D. J. Douglas,
and D. Simmons, Org. Mass Spectrom. (a) 17. 205 (1982);
( b) 17. 212 ( 1982) .
6. (a) K. L. Busch, G. L. Glish, and S. A. McLuckey, Mass
Spectrometry/Mass Spectrometry: Techniques and
Applications of Tandem Mass Spectrometry, VCH, New York,
1988; ( b) P. H. Dawson and D. J. Douglas, in Tandem Mass
Spectrometry ( ed. by F. H. McLaf f erty) , p. 125, Hiley, NY
(1983); ( c) R. A. Yost and D. D. Fetterolf, Mass Spectrom.
Revs. 2_, 1 (1983).
7. B. Shushan, D. J. Douglas, H. R. Davidson, and S. Nacson,
Int. J. Mass Spectrom. Ion Phys. 46. 71 (1983).
8. R.I. Martinez, Rapid Commun. Mass Spectrom. ( a) 2., 8
( 1988) ; ( b) 2_, 41 ( 1988) .
The term "dynamically correct" was coined (see Appendix
of reference 9) ' to indicate those branching ratios
measured in Xtrf]Q instruments which correspond to the
distribution of reaction products which, in principle,
would be observed at the scattering center of an
idealized crossed molecular beam machine (if one were
able to integrate over all angles the ion intensities of
each reaction product channel). This correspondence is
attributed to the strong focusing properties of rf-only
multipoles which provide high ion-containment
efficiencies for ions scattered through a broad range of
angles. Hence, dynamically-correct branching ratios
are those which have been appropriately corrected for
discrimination effects, and, therefore, provide an
11-160
-------
instrument-independent representation of the primary
ion-neutral interaction of A +B.
9. R.I. Martinez and B. Ganguli, Rapid Commun. Mass
Spectrom. 3., 427 (1989).
10. R.I. Martinez, (a) J. Res. Natl. Inst. Stand. Tech. (US)
94. 281 (1989); ( b) J. Am. Soc. Mass Spectrom. 1_, 272
(1990).
11. R.I. Martinez, Rapid Commun. Mass Spectrom. 3., 127
(1989).
12. R. I. Martinez, "WORKSHOP: Instrument-Independent CAD
Database", Proceedings of the 37th ASMS Conference on
Mass Spectrometry and Allied Topics (p. 1560), Miami
Beach, Florida, May 21-26, 1989.
13. R.I. Martinez and B. Ganguli, Rapid Commun. Mass
Spectrom. 1, 377 (1989).
14. D. J. McAdoo, F. H. McLafferty, and J. S. Smith, J. Am.
Chem. Soc. 92. 6343 (1970).
15. J. H. Beynon, R. M. Caprioli, and R. G. Cooks, Org. Mass
Spectrom. 9_, 1 ( 1974) .
16. C. C. van de Sande and F. H. McLafferty, J. Am. Chem. Soc.
97. 4617 (1975).
17. D. J. McAdoo and D. N. Hitiak, J. Chem. Soc., Perkin II,
770 (1981).
18. C. Lifshitz, (a) J. Phys. Chem. 87. 2304 (1983); ( b)
Int. J. Mass Spectrom. Ion Phys. 43, 179 (1982).
19. F. H. McLafferty, D. J. McAdoo, J. S. Smith, and R.
Kornfeld, J. Am. Chem. Soc. 93. 3720 (1971).
20. J. Diekman, J. K. MacLeod, C. Djerassi, and J. D.
Baldeschwieler, J. Am. Chem. Soc. 9J_, 2069 (1969).
21. I. Powis and C. J. Danby, Int. J. Mass Spectrom. Ion
Phys. 32. 27 (1979).
22. R. Stockbauer, Int. J. Mass Spectrom Ion Phys. 25. 89
(1977).
1-161
-------
23. D. J. McAdoo and C. E. Hudson, Int. J. Mass Spectrom. Ion
Proc. 59. 77 ( 1984) .
24. G. Bouchoux and Y. Hoppilliard, Can. J. Chem. 60. 2107
( 1982).
25. A. J. Stace and A. K. Shukla, Int. J. Mass Spectrom. Ion
Phys. 37. 35 < 1981) .
26. H. J. van der Zande, D. P. deBrui jn, J. Loss, P. G.
Ristemaker, and S. A. McLuckey, Int. J. Mass Spectrom Ion
Proc. 67. 161 (1985).
27. C. Lifshitz and E. Tzidony, Int. J. Mass Spectrom Ion
Phys. 39. 181 (1981).
28. P. J. Derrick and S. Hammer urn, Can. J. Chem. 64. 1957
(1986).
29. C. S. T. Cant, C. J. Danby, and J. H. D. Eland, J. Chem. Soc.
Faraday Trans. II, 71. 1015 (1975).
30. D. M. Mintz and T. Baer, Int. J. Mass Spectrom. Ion Phys.
25. 39 ( 1977) .
31. R. Bombach, J. -P. Stadelmann, and J. Vogt, Chem. Phys.
72. 259 (1982).
32. A. K. Shukla, K. Qian, S. L. Howard, S. G. Anderson, K. H.
Sohlberg, and J. H. Futrell, Int. J. Mass Spectrom. Ion
Processes 92. 147 (1989).
33. K. Qian, A. Shukla, S. Howard, S. Anderson, and J.
Futrell, J. Phys. Chem. 93. 3889 (1989).
34. K. Qian, A. Shukla, and J. Futrell, J. Chem. Phys. 92.
5988 (1990).
35. A. K. Shukla, K. Qian, S. Anderson, and J. H. Futrell, J.
Am. Soc. Mass Spectrom. 1_, 6 (1990).
36. N. Heinrich, F. Louage, C. Lifshitz, and H. Schwarz, J.
Am. Chem. Soc. 1 10. 8183 (1988).
37. R.I. Martinez and B. Ganguli, J. Am. Soc. Mass Spectrom.
2_, xxx (1991); submitted for publication.
1-162
-------
37
In this paper we describe the kinetics and mechanism
of the CAD of the acetone cation which indicates that
there is a correspondence between the distribution of
internal energies accessed by the photoionization of
acetone [viz., the PEPICO data of ref. 29 and 311 and
the energy deposition function accessed by collisions!
activation of acetone cations formed by 70 eV El [viz.,
our CAD data]. That is, the low-energy CAD of the
acetone cation involves electronic transitions (rather
than vibrational excitation), ' and dissociation
occurs primarily from the same electronic states in both
the CAD and PEPICO experiments.
37
The concordance of our findings with those of
PEPICO ' and,molecular beam experiments
indicates again that the HIST kinetics-based protocol
developed in this laboratory makes it possible for one
to measure dynamically-correct product distributions
which have been appropriately corrected for
discrimination effects. That is, one can obtain an
undistorted (instrument-independent) representation of
ion-neutral interactions (e.g., CAD). This is essential
for the development of a standardized, instrument-
independent MS/MS database for XIrfJQ instruments. The
data in Table 1 constitute some of the first elements of
such a database.
38. R.I. Martinez, Rev. Sci. Instrum. , 58. 1702 (1987).
10
11-163
-------
Table 1. Branching Ratios1 vs. E for the CAD of 58* from Acetone
Parent Ion: C-H..O* (m/z 58) Source Compound: 2-Propanone (>99.7Jt)
lonizat
cm
( ซT_1
1. 2
4. 1
10. 6
32. 6
44. 9
61. 2
Bcn,
( eV)
1. 2
- j b
ion Mode: 70 eV electrons Target Gas: Ar (single
a . Branchi no Ratios for the CAD of 56
UiL
24
[101
31
(15)
32
[151
34
(251
35
(101
34
[201
(14*)
0. 0000
0.0000
0. 0000
0. 0128
1751
0. 0000
0. 0000
(15*)
0. 0153
(8]
0. 0151
(101
0. 1581
( 31
0. 2561
[101
0. 0444
( 10]
0. 0278
[ 10]
Branch! nor
(31*)
0. 0000
(39*)
0. 0000
collision)
fr26*) <27*) (28*) (29*)
0. 0000 0.
0. 0000 0.
0. 0005 0.
[50] [
0. 0054 0.
(30) I
0. 0059 0.
(1001 (
0. 0000 0.
[
Ratios for
(40*) (
0. 0000 0.
0000 0.
0000 0.
0046 0.
20] (
0578 0.
71 [
0882 0.
15J [
0195 0.
20]
the CAD
41*1 I
0000 0.
0000 0.
0000 0.
0033 0.
15] (
0062 0.
251 t
0059 0.
100] C
0000 0.
[
of 58*
42*) (
0195 0.
0000
0000
0054
15]
0308
151
0738
20]
0124
351
43*)
965
4. 1 0. 0000 0. 0000 0. 0000 0. 0000 0. 0241 0. 961
(20] (4]
10.6 0.0008 0.0000 0.0000 0.0000 0.0177 0.810
[351 [251 [51
32.6 0.0020 0.0029 0.0014 0.0047 0.0218 0.598
(601 (501 [501 (301 (15) (2]
44.9 0.0059 0.0036 0.0012 0.0084 0.0323 0.730
(1001 (35] (1001 [301 (201 [31
61.2 0.0000 0.0120 0.0084 0.0169 0.0580 0.845
(20] (301 (201 (151 (51
Thฃ CAD^of 58* produces only the fragment ions indicated (e.g.,
14 , 15 , etc.) and a complementary neutral fragment (not shown).
Numbers in square brackets represent maximum possible uncertainty
in the cross section a and in the branching ratios, expressed as a
percentage of each o and of each branching ratio. Hence, at
E =44.9 eY, the branching ratio for 58 ป43 is 0.730 (ฑ0.02 max),
wfiFle that for 58 ป31 is 0.0059 (ฑ0.0059 max).
11-164
-------
*e IMPROVED TECHNIQUES FOR FORMALDEHYDE ANALYSIS
BY HPLC USING AUTOMATED SAMPLE PREPARATION
AND DIODE ARRAY DETECTION
by Brian Goodby. Smita Vasavada, Jim Carter and Larry Schaleger
B C Analytical, 801 Western Avenue, Glendale, California, 91201
ABSTRACT
Formaldehyde, one of the more widely produced intermediates in
the U.S. chemical industry, is formed by combustion and biological
processes, making its presence in the environment ubiquitous.
Because formaldehyde is a probable carcinogen, reliable analytical
methods for identifying trace levels of this analyte must be found.
In this study, we examined some of the difficulties involved in a
common HPLC method, exemplified by draft EPA Method 8315 and
California Air Resources Board (CARB) 430, and considered
approaches for reducing systematic error. Using spectral
confirmation, we also investigated the frequency at which
interferences or false positives occur.
The common method relies on pre-column defivatization of the
aldehydes with 2,4-dinitrophenylhydrazine (DNPH) followed by
reversed-phase HPLC analysis. As others have noted, this method
of determining formaldehyde at trace levels presents a major
problem: contamination, which results in elevated blank levels and
increased detection limits. Causes of contamination include solvents
and solid-phase extraction columns as well as the exposure of
reagents and samples to ambient air. Introduction of contamination
counteracts the advantages of concentrating the sample through
extended preparation procedures, such as liquid/liquid and solid-
phase extraction. Because concentration is not needed to meet
detection limits on the order of 20-50 ppb, which are satisfactory
for most regulatory purposes, we have investigated the application
of automated pre-column derivation using the Hewlett-Packard
1090 Series II HPLC system. The paper presents details of this
procedure.
1-165
-------
The application of the diode array UV-visible detector to this
analysis provides many advantages over fixed-wavelength analysis.
Shifts in retention time during the course of analyzing batches of
real-world samples are quite common, and lead to misidentification
in the case of single-wavelength, single-column detection. With
spectral information available from the diode array detector, false
positives can be virtually eliminated. The paper provides examples
of chemical interferents and misidentifications observed in
environmental analysis.
INTRODUCTION
Formaldehyde analysis of environmental samples is becoming more
common in the analytical lab. The detection of this compund plus
other carbonyls is important due to the health hazards and possible
role they play in environmental reaction pathways. These
compounds are formed by incomplete combustion and atmospheric
photoxidation of hydrocarbons. Formaldehyde is a very common
ingredient in cosmetics, building materials, as well as a key
chemical for chemical synthesis. The fact that formaldehyde is
universally present leads to positive detection by any analytical
procedure. The blank levels that are observed can vary
dramatically when an extensive sample preparation procedure is
used. This paper will discuss the application of two slightly
different approved procedures, EPA 8315 and CARB 430, plus
compare these methods to an on-line HPLC sample preparation
procedure that controls the sample contamination problem.
EPA Method 8315 involves the analysis of aqueous and solid
samples by DNPH derivatization followed by HPLC detection. This
research did not investigate the application of 8315 to solid
samples. Aqueous samples are mixed with a derivatizing solution
of DNPH in ethanol plus acetic acid. The pH for this reaction is
adjusted to around 5 and derivatization is allowed to proceed for at
least a half hour. This solution is then extracted by either
liquid/liquid or solid phase extraction. The liquid/liquid procedure
partitions the DNPH hydrazone product into methylene chloride.
The SPE process uses a CIS column to separate the derivatized
product. In both preparations the final extract must be exchanged
for one that is suitable for HPLC analysis, acetonitrile (ACN) or
1-166
-------
methanol. The reader is referred to the actual methods for more
details on this preparation.
GARB 430 is the California approved air sampling method for
airborne formaldehyde. It involves the use of DNPH impinger
solution that is composed of approximately 2N hydrochloric acid
(HCL). This acidic solution traps and derivatizes immediatley all
carbonyl species that bubble through. The sample solution is then
extracted with 70/30 vol/vol % hexane/methylene chloride. The
extract is again exchanged to a suitable HPLC solvent. Therefore,
the final extracts that are obtained by these two methods are
identical and all HPLC conditions are the same.
Both approved methods and many journal publications mention the
high blank levels obtained by these sample preparation procedures.
CARB 430 outlines a very detailed purification procedure involving
multiple recrystallizations of DNPH followed by storage in a
nitrogen purged dessicator. All impinger solution must be checked
for contamination prior to use with 48 hours as the maximum time
between preparation and use. 8315 also mentions that blank levels
are a major problem but the only precaution mentioned is to use
the highest quality reagents. In both methods it is suggested that
the blank level is subtracted from all sample data. 8315 mentions
this blank subtraction procedure only in the context of the
establishing of the method detection limit. The work reported here
has centered around trying to clean up this blank problem and
therefore not do blank subtraction when reporting data.
EXPERIMENTAL
Reagents and standards.
All organic solvents were of HPLC UV spectral grade. Many
vendors were consulted during solvent selection and no guarantee
of low ( <50 ppb) formaldehyde levels could be confirmed. The
results obtained here indicate that the purity of the DNPH and
water is more critical then that of the organic solvents. DNPH was
purchased from ChemService (West Chester, PA) and recrystallized
twice in pure ACN according to the CARB 430 method. Impinger
solution was prepared from 90 ml of concentrated HCL (Baker
Analyzed) to which .250 grams of pure DNPH is added. After the
11-167
-------
crystals have dissolved organic free water is added to produce 500
ml of solution. The water used throughout this work was distilled
then passed through a Barnstead Nanopure II system, heated to
near boiling and purged for a hour with helium, and finally charcoal
filtered. This impinger solution was used directly as the on-line
(autoprep) derivatization solution in the Hewlett Packard HPLC.
Formaldehyde for calibration and quality control solutions was
obtained also from ChemService as a 70% aqueous solution. Known
concentrations (1000 ppm) of a aqueous stock were prepared
according to method 8315. This procedure uses a pH titration
procedure to establish the concentration of formaldehyde in the
stock. CARB 430 recommends the production of pure formaldehyde
hydrazone crystals for calibration. Our experience with this
procedure has never produced quantitative information that
compares favorably with 8315. Even though the melting point
observed for the product appeared to be acceptable these crystals
always produce standards that gave very low responses. All
calibration standards were prepared exactly like as a analytical
sample for all three methods. The autoprep technique involved the
production of spiked water at 4 or 5 calibration concentrations
which were placed on the instrument and derivatized on-line.
Instrumentation.
The HPLC instrumentation consists of a Hewlett Packard 1090
series II with a diode array detector. Computer control is provided
with a HP ChemStation with a Pascal based operating system. This
instrument is a binary gradient low pressure mixing system. The
mobile phase consists of 0.01 molar phosphoric acid (channel A)
and pure ACN (channel B). The selection of a weak acidic mobile
phase provides two benefits. First it assures that the DNPH reaction
proceeds to completion plus it stops the columns from becoming
clogged. Without a acidic mobile phase the reverse phase columns
used in our lab have stopped functioning after as few as 30
samples. With the introduction of a phosphoric acid mobile phase
the current column has completed over 300 analytical runs. Due to
the nature of the samples we analyze (high organic contamination),
inexpensive C18 columns are purchased from Alltech (Deerfield,
IL). The column used for this work is a Econosphere C18 with 5
micron particle packing (150mm x 4.6mm). The gradient
1-168
-------
conditions are shown in Table 1 The spectroscopy parameters are
also presented.
The procedure for on-line derivatization is presented in Table 2.
The first step (line 1) prompts the autosampler to draw up into the
sample loop 2.5 micro liters (ul) of DNPH impinger (vial #1)
solution. Next 10 ul of sample is drawn up followed by a syringe
rinse in solvent (ACN). Another plug of DNPH is then drawn up to
sandwich the sample Line 5 is the mixing step which moves the
sample plus reagent back and forth through the heated reaction
furnace which is at 50C. After mixing a final 10 ul of solvent is
drawn in order to optimize the position of the sample in the
reaction furnace. The autosampler then waits two minutes before
performing the analytical injection.
RESULTS AND SUMMARY
Method development on the autoprep derivatization technique
involved experimenting with different reagent combinations. The
starting point used the reagents suggested by EPA 8315. Thus, the
derivatizing solution was a combination of 5M acetic acid plus
saturated DNPH/ethanol. This combination did not give satisfactory
results. The blank values observed were very high (at least 100
ppb). Also, the sensitivity was not comparable to standards
prepared by Method 8315. This lack of sensitivity was most likely
due to the use of a weak acid for derivatization. It had been
observed previously in our lab that this reaction was not complete
in a few minutes. In order for the autoprep technique to be time
effective different reagents had to be selected. The application of a
strong acid such as HCL or sulfuric was suggested to us by a fellow
researcher. Because we always are preparing clean HCL impinger
solution for Method CARB 430 this was the easiest reagent to use.
It was proposed that the injection of 5 ul's of this acid solution
could be tolerated by the HPLC.
Over the course of approximately 3 months a calibration
comparison between the autoprep procedure and the approved
sample preparation procedures was performed. The results of this
study is graphically displayed in Figure 1. Each calibration curve
has a least squares fit equation and line associated with it. The
1-169
-------
equation in the top of the each plot is associated with the top line
on the right hand border. The linearity (correlation coefficient RA2)
of these calibration curves are all > 0.990 with only one of the
autoprep coefficients being < 0.995.
There is no criterion for linearity in EPA Method 8315. GARB 430
mentions that linearity through the origin may be assumed if an
RA2 of 0.999 is obtained. As this data indicates an RA2 of 0.999 does
not guarantee that the calibration curve passes through the origin.
In fact, in our experience these calibration curves rarely include the
origin. If one tries to force the orgin as a data point then the RA2
goes down. The fact that these calibration curves behave in this
manner is not surprising. One must remember that contamination
by formaldehyde can occur during any stage of the standard
preparation. It is this random contamination that leads to the
spread in these calibration plots. By comparing the autoprep data
to that obtained by the two approved methods it is seen that all
three techniques are fairly comparable.
One point that is not obvious from the calibration data is the
comparison of blank levels obtained by each technique. In the
cases of 8315 and CARB 430, the blank response is frequently as
high as the lowest calibration standard (100 ppb). The blanks
obtained on clean water by autoprep show substantially less
response. Assigning an absolute quantifiable number to the
autoprep blank is difficult, because preparation of calibration
standards at these low ( <100 ppb) levels is nearly impossible.
Basically, the peak areas obtained for the autoprep blanks are 4 to
5 times less than those obtained through normal sample
preparations. This lack of low level quantification information also
makes it difficult to establish a method detection limit (MDL) for
the autoprep technique. The approved methods however both use
blank subtraction to establish their MDL's. If one looks at the
absolute instrumental response from the HP1090 it can be
estimated that a detection limit of around 5-10 ppb should be
obtainable.
Figure 2 depicts the chromatgraphic response of a high level
formaldehyde standard. The peak shape for the formaldehyde
derivative is obviously not symmetrical. However, with the
1-170
-------
spectral power of the diode array it can be demonstrated that this
peak is pure. On a brand new column the standards do initially
appear symmetrical. But after a few real world samples the peak
shapes deteriorate. This deterioration is not due to column overload
because even low level standards are asymmetrical. In order to
check the performance of the used columns a pesticide mixture was
run under the same mobile phase. The gradient program was
slightly different. As can be seen in Figure 2B the chromatography
of this mix displays better peak shape. From the point of view of a
production laboratory a new column for formaldehyde analysis
cannot be justified.
The analytical results for a number of different types of water
samples are presented in Table 3. In every case the normal 8315
preparation of these samples produced higher results than the
autoprep technique. Not only did the normal preparation produce
contaminated sample extracts this preparation requires a lot of
labor. The normal preparation involves liquid/liquid separatory
extractions in triplicate followed by concentration and solvent
exchange. Reagent consumption for the normal preparation is
hundreds of milliliters compared to ul by autoprep. The only
manipulation done for the autoprep procedure is to load a
autosampler vial with sample and place .it on the HPLC.
Another advantage of the autoprep technique is due to the fact that
the HPLC injects the sample directly rather then a concentrated
extract. The solvent extraction processes of 8315 and CARB 430
extract all the organic components in the sample and these then can
cause many types of chromatographic problems. Not only does it
appear that the extraction process contaminates the samples but
chromatographic interferences can be promoted by the
concentration procedure. The HPLC chromatograms generated by
autoprep are less likely to produce complicated peak shapes,
retention time shifts, or UV/Vis spectral complications. An
example of the type of problems observed for a sample that has
been prepared according to Method 8315 is shown in Figure 3. The
top portion of this figure shows the chromatogram and the bottom
portion contains selected UV/Vis spectral scans. It is fairly obvious
from the chromatogram that the formaldehyde derivative
(retention time = 7.0) peak is not a single component. If one is
11-171
-------
performing this test with a single beam UV/Visible instrument no
further insight into this coelution is possible.
Because the diode array system gathers and stores complete
spectral scans during each peak the presence of an interfering
species can be confirmed. Three spectral responses are shown for
1- the first peak max at 6.7 min., 2 - the peak max at 6.9 min., and
3- the tailing edge at 7.3 min.. Spectra 1's response is indicative of
pure formaldehyde derivative which quickly becomes convolved
with a coeluting species spectra shown in spectra 2. Even though
the interfering species peak maximum is at a different wavelength
(250 nm) than that used for formaldehyde (360 nm) the
absorbance is strong and probably adds to the total peak area. The
three spectra shown in the figure are all scaled to the same axis.
Nothing about absolute intensity is displayed. It was observed that
the absorbance from the interfering species was approximately
three times as strong as that seen for formaldehyde. The presence
of this interfering species remains throughout the remainder of the
chromatographic peak. Spectra 3 still displays the low wavelength
absorption due to its presence. Because of the diode array
information these analytical results could be reported to the client
as elevated due to this interferent.
1-172
-------
2000
It
:
1000
8315 Calibration
y ป 0.28190 X +113.87
R*2 = 0.996
y = 0.23850 X + 33.865
RA2 = 0.999
2000
1000-
I
0.
CARB 430 Calibration
y = 0.31547 X + 30.224
fi*2 = 0.999
y = 0.2483 X +67.438
R*2 = 0.997
1000 2000 3000 4000 SOOO SOOO
Concentration (ppb)
1000 2000 3000 4000 SOOO 6000
Concentration (ppb)
1MO
1000
t
500
Impinger Autoprep Calibration
y = 0.25544 X - 21.179
R*2 - 0.991
y = 0.20985 X + 88.569
Hป2 = 0.999
2000
1000-
Composite Calibration Data
ป IMP prep
art* 430
e*to 430
8315
a S31S
0 1000 2000 3000 4000 5000 SOOO 0 1000 2000 3000 4000 SOOO 8000
Concentration (ppb) Concentration (ppb)
Figure 1 - Comparison of Calibration Data
11-173
-------
ซBB-
3BB-
2BB-
tee-
DNPH
FORMflLBEHTOE
DERIVATIVE
A
B B IB
Time (min.1
20-
13-
E IB-
SPIKE
B
SIHRZINC
Jb
5URFLBN
a ซ e e te 12
T I me ( m 1 n . )
Figure 2 - HPLC Chromatograms
A) 5000 ppb formaldehyde std.
B) Pesticide test mix
11-174
-------
IT
E
2-
6.8
6. 5
7.0 7.5
Ti me (m i n. )
8. 0
8. 5
TJ
I)
0
0
M
I
E
1 - Front edge spectra
/ 3 Tall edge spectra
2 - Peak max spectra
300 -400
Have length (ntn)
500
600
Figure 3 - Example of a Chromatographlc/
Spectral Interferent.
1-175
-------
LIQUID CHROMATOGRAPH
initial parameters
Flow : 0.500 ml/nin
Solvent A : 40.0 %
B : 60.0 %
Max Pressure :
Stop Time :
Post Time :
Injection Volume :
400 bar
14.00 min
0.00 min
10.0 ul
Min Pressure :
off
LIQUID CHROMATOGRAPH
Tin
(Bin)
1.00
5.00
7.00
8.00
10.00
14.00
Solvent
Solvent
Solvent
Solvent
Solvent
Solvent
A:
A:
A:
A:
A:
A:
40
30
0
0
40
40
.0
.0
.0
.0
.0
.0
B:
B:
B:
B:
B:
B:
60.0
70.0
100.0
100.0
60.0
60.0
DIODE-ARRAY DETECTOR
SIGNALS ABC
Sample (nm)
Wavelength : 360 340 380
Bandwidth 80 80 80
Reference (nm)
Wavelength 560 560 560
Bandwidth 40 40 40
signals & spectra
Store Spectrum
Threshold
PeaKwidth
Stop Time
Post Time
Prerun Balance.
peak controlled
0.1 mAU
0.150 min
14.00 Bin
0.00 min
Yes
about 896 Records acquired during Run
Sampling Interval 960 OB
Spectrum Range from : 220 nm
to : 600 nn
step : 4 MI
Table 1 - HPLC Chromatography and
Diode Array Parameters
11-176
-------
INJECTOR PROGRRM
Slowdown Draw & Eject : 2
Mix : 2
Hold after Draw Be Eject : 0 seconds
Linett
1
2
3
4
5
6
7
8
Function
ggHB 2-s
Draw
Draw
Draw
Mix
Drav
Wait
1D.D
D.D
2.5
1D.O
1D.D
2. DO
Inject
ul
ul
u)
u]
u]
u]
from :
from :
from :
from :
Vialtt :
Sample
Vial* :
Vial* :
cycles : ID
from : Vial* :
1
D
1
D
minutes
25.D ul accumulated in Syringe with Line* 5
Table 2 - Autoprep Injector Program
1-177
-------
TABLE 3 - COMPARISON OF ANALYTICAL RESULTS
(ALL RESULTS ARE PPB)
SAMPLE DESCRIPTION
8315 RESULT
AUTOPREP RESULT
WASTE WATER
SEWER COMPOSITE #1
SEWER COMPOSITE #2
GROUND WATER #1
#2
#3
#4
#5
#6
#7
#8
AQUEOUS SAMPLE
104
226
230
110
45
62
87
94
95
82
113
389
53
<25
<25
<25
11
it
it
<25
-------
AN INTERLABORATORY COMPARISON STUDY OF SUPERCRITICAL FLUID
EXTRACTION FOR ENVIRONMENTAL SAMPLES; Tammy L. Jones. U.S.
Environmental Protection Agency, Environmental Monitoring Systems
Laboratory - Las Vegas, Las Vegas, NV 89193, Tom C.H. Chiang,
Lockheed Engineering and Sciences Company, Las Vegas, NV 89119
The U.S. Environmental Protection Agency recently conducted
a multilaboratory evaluation of a supercritical fluid extraction
(SFE) protocol. SFE is a relatively new technique which can be
used to extract compounds of environmental interest from solid
matrices (soils, sediments, fly ash, etc.) by using supercritical
C02. Ten laboratories participated in this study that was designed
to evaluate the feasibility and applicability of a protocol
developed for the extraction of environmentally significant
analytes from environmental matrices.
The efficiency of analyte (polynuclear aromatic hydrocarbons
and phenols) recoveries, using SFE, from three solid matrices (two
standard reference materials and one spiked sand) was studied. The
analyses of the resulting extracts from all the laboratories were
performed by a single laboratory using gas chromatography/mass
spectrometry (GC/MS). The data were evaluated in terms of
precision, accuracy, and the intra-laboratory and inter-laboratory
variations within this technique. In general the percent
recoveries of the analytes from the various laboratories ranged
from poor (< 40%) to very good (> 90%). There was a trend noticed
that those laboratories who performed satisfactorily on one sample
matrix also continued to do so on the other two. NOTICE: Although
the research described in this article has been supported by the
1-179
-------
U. S. Environmental Protection Agency, it has not been subjected
to Agency review and therefore does not necessarily reflect the
views of the Agency. This document is intended for internal Agency
use only. Mention of trade names or commercial products does not
constitute endorsement nor recommendation for use.
1-180
-------
67
AN ANALYTICAL MANUAL FOR PETROLEUM PRODUCTS IN THE ENVIRONMENT
M.W. Miller, M.M. Ferko, F. Genicola, H.T. Hoffman and A.J. Kopera
New Jersey Department of Environmental Protection
Office of Quality Assurance
CN 027, Trenton, New Jersey 08625
Abstract
The New Jersey Department of Environmental Protection (NJDEP)
Analytical Chemistry Manual for Petroleum Products in the
Environment was drafted to help project managers select appropriate
analytical methods. Eight NJDEP programs administer regulations
concerning petroleum products. The analytical methodologies for
these programs have not been codified within federal or state
regulations, and several method variants exist.
Preparatory to drafting the Manual, we conducted an extensive
review of the regulatory programs. The methods and standards
reviewed include those of federal and state agency departments, as
well as those of the American Public Health Association, American
Society for Testing and Materials, and American Petroleum
Institute. Selected methods and procedures for free product,
aqueous matrices and nonaqueous matrices were edited to establish
a Department Manual. Methods for volatile petroleum products
(e.g., gasoline, jet fuel, kerosene, solvents) and semivolatile
petroleum products (e.g. diesel, fuel oils #2-#6) are presented.
The analytical laboratory methods contained in the Manual will
become part of the revised NJDEP Regulations Governing Laboratory
Certification and Standards of Performance, N.J.A.C. 7:18.
The paper discusses a survey method, two quantitative methods
and one fingerprint method. These methods are representative of
the fifteen methods in the first edition of the Department Manual.
A gas chromatography-photoionization-flame ionization detector (GC-
PID-FID) survey method is presented for volatile petroleum
products. A quantitative GC-PID-FID method is discussed for
volatile petroleum products. Gas chromatography-mass spectroscopy
is discussed for the identification and quantification of specific
semivolatile compounds in petroleum contaminated soil. The
identification of specific petroleum products in contaminated water
and soil or free product is accomplished by GC-PID-FID
fingerprinting.
Each method contains calibration procedures for petroleum
products, and quality control requirements. The manual also
contains a users guide for environmental professionals.
1-181
-------
QQ EVALUATION OF LIQUID/SOLID EXTRACTION FOR THE ANALYSIS OF
ORGANOCHLORINE PESTICIDES AND PCB'S IN TYPICAL GROUND AND SURFACE
WATER MATRICES
Anne D. O'Donnell. Group Leader, Denise R. Anderson, Group Leader, John
T. Bychowskl, Manager, Semi-Volatile Organics Department, WMI
Environmental Monitoring Laboratories, Inc., 2100 Cleanwater Drive,
Geneva, Illinois, 60134; Craig G. Markell, Research Specialist, I&E Sector
New Products Department, Donald F. Hagen, Corporate Scientist, Corporate
Research Analytical Laboratory, 3M Corporate Research Labs, 3M Center,
Bldg. 201-1S-26, St. Paul, Minnesota, 55144
ABSTRACT
Method 608/8080 is used for the analysis of organochlorine pesticides and
PCB's in water and wastewater. The main features of the method are
liquid/liquid extraction (LLE) with methylene chloride, removal of the
methylene chloride to concentrate the analytes, a solvent exchange into
hexane, and gas chromatographic analysis with electron capture detection.
Because several of the 500 series drinking water methods are being updated
with the inclusion of liquid/solid extraction (LSE), a similar
modification was evaluated for Method 608. The LLE steps were replaced
with a solid-phase disk (47mm C18) extraction, elution of analytes with
ethyl acetate, and direct GC analysis of this eluate.
The LSE evaluation study was performed with reagent water and composites
of typical ground and surface waters, including groundwater composites
with very high particulate content. The single organochlorine pesticides
and the multicomponent mixtures were all spiked at two concentration
levels, a "validation" level and an "MDL" level. Elution efficiency was
determined for all sample types.
The recovery efficiencies, %RSD's, and method detection limits obtained
demonstrate that LSE is at least equivalent to LLE for the Method 608/8080
analytes, and, in most cases, an improvement. The LSE disk modification
was successfully applied to all water matrices typically encountered in
our laboratory. Disk LSE provides a clear advantage in terms of time and
cost per analysis and solvent use and disposal.
INTRODUCTION
The use of LSE instead of LLE for the isolation and concentration of
organic components in environmental water samples is becoming more
extensive because of the time and cost benefits it provides. LSE is less
labor-intensive, uses substantially less glassware, and significantly
reduces the volume of hazardous and costly solvents required. The solid
phase used most frequently is octadecane (C18) chemically bonded to porous
silica particles. It is commonly packed into disposable plastic
cartridges, producing LC mini-columns. SPE using these cartridges is an
alternative sample preparation procedure cited in the Drinking Water
Methods 506, 525.1, and 550.1. Method 525.1 addresses many of the same
1-182
-------
analytes covered by Method 608/8080 (1).
Recently 3M introduced membrane disks, Empore, as an LSE medium (2).
Instead of being packed in a cartridge, the CIS-bonded silica particles
are enmeshed in PTFE fibrils. The large diameter, thin disks (47mm x
0.5mm) provide a large cross-sectional area with low back pressure. The
primary advantage of the disk technology is the speed of extraction
possible with equivalent extraction efficiency. Efficiency is achieved
with a smaller particle size, a uniform, high density packing, and an
effective low linear velocity through the disk at high sample flow rates.
The 47mm diameter disks fit standard glass filtration assemblies, allowing
the extraction to be carried out in an all glass/PTFE environment. One
of the principal disadvantages of the LSE cartridges is the amount of
trace contaminants contributed by plastic housings. Disk LSE is
designated as an approved technique for Methods 506/8061, 513, 525.1, and
550.1.
The organochlorine pesticides and PCB's determined by Method 608/8080
present good candidates for LSE. They are extracted at a neutral pH, are
insoluble in water (large capacity factor, k', for CIS/water
reversed-phase conditions, soluble in organic solvents (easily eluted),
and are relatively non-volatile. In addition, the 608/8080 Method is very
susceptible to interferences from trace level contamination because of the
high sensitivity of the electron capture detector. A procedure that
significantly reduces both the amount of glassware that must be kept
scrupulously clean and the volume of solvent concentrated for the final
extract will also significantly reduce contamination interferences.
The evaluation of Empore disks for LSE of the Method 608/8080 analytes
was performed using reagent water, composites of "average" ground and
surface waters, and composites of groundwater samples with a very high
total suspended solids (TSS) content. The technique was to be challenged
with all the types of water samples normally encountered. The single
organochlorine pesticides and the multicomponent mixtures were all spiked
at two concentration levels, a "validation" level and an "MDL" level.
Elution efficiency was determined for all analytes in all sample types.
EXPERIMENTAL
Materials. Empore Extraction Disks, C18, 47mm (Varian Sample Preparation
Products, Harbor City, CA, Cat. #1214-5004). Whatman Multigrade GMF 150
graded density glass microfibre filters, 37mm (Cat. #1841-047, Clifton,
NJ).
Apparatus. Glass filtration apparatus, 47mm, 300mL funnel, lOOOmL flask
Nuclepore Cat. #410502 (Pleasanton, CA). Millipore (Bedford, MA)
vacuum/pressure pump (Cat. #XX55 000 00). A tee with a pinch clamp is
placed in the line between the filtration assembly and the pump to allow
fine control of the vacuum for the preconditioning and eluting steps.
1-183
-------
Procedure. The filtration unit is assembled with the Erapore disk. The
funnel and disk are washed with lOmL of ethyl acetate (the elution
solvent), then with lOmL of methanol (preconditioning wetting agent), and
finally with two lOmL rinses of reagent water. The 1L water sample, to
which surrogate standard and 0.5% methanol wetting agent have been added,
is then passed through the disk at full vacuum (25" Hg, 85 kPa). A thin
layer of liquid is maintained on the disk from the methanol conditioning
step until the entire sample has been extracted. The disk is subsequently
eluted with two 5mL portions of ethyl acetate; the first portion is also
used to rinse the sample bottle. During the elution step, the ethyl
acetate is allowed to equilibrate on the disk for a few minutes. The
eluate is collected in a lOmL Kuderna-Danish (KD) concentrator tube.
Internal standard is added to the extract and it is made up to volume.
Na2S04 is added to dry the sample.
Liquid-liquid extraction analyses were performed using continuous
liquid-liquid extractors for an 18-hour period.
GC Analysis. Samples were analyzed on a Hewlett-Packard Model 5890A GC
with electron capture detection, using a Hewlett-Packard Model 7673A
autosampler and Fisons/VG Multichrom Data System, Version 1.8. The column
was a J&W Scientific (Folsom, CA) DB608, 30m x 0.53mm i.d., 0.83pm film
thickness (Part No. 125-1730). Helium was the carrier gas at y=A5cm/sec,
with argon-5ฃmethane make-up at 65mL/min. The injection port was at
200ฐC, and the detector at 300ฐC. The temperature program was:
isothermal at 140ฐC for 0.5 min, 140ฐ-275ฐ @ 6ฐC/min, hold 15 min. The
injection was 2 uL splitless.
RESULTS AND DISCUSSION
Table I summarizes the results obtained for Empore LSE extraction of the
608/8080 analytes from reagent water at a "validation" concentration
levels. For all experiments, the amount of analyte spiked into the sample
water was also spiked into lOmL K-D concentrator tubes containing ethyl
acetate, the "spike check" sample. The sample extracts and the
triplicate spike check samples were treated identically for analysis. The
mean of the spike check samples was the basis for the recovery efficiency
calculation.
To test the completeness of elution with the two 5mL volumes of ethyl
acetate, a second set of 5mL washes was passed through the disk, after the
first elution, and collected in a second lOmL concentrator tube. Elution
efficiency was calculated as the fraction of analyte concentration in the
first eluant compared to the total analyte concentration in both eluants.
Elution efficiency tests were run for representative Aroclors, not the
entire set.
Table I data show excellent recovery and elution efficiencies. The mean
JKRecovery for all analytes was 91.5%, and the mean elution efficiency was
0.991. The lower recovery value for Aldrin is a function of its higher
11-184
-------
volatility. The same effect is seen in the lover recovery for the
surrogate standard, which is more volatile than the rest of the analytes.
The precision of the method is also excellent. The mean %RSD for all the
analytes vas 3.0%.
Method validation data for the single organochlorine pesticides using LLE
are presented in Table II for comparison. The mean accuracy of the method
is 88.2%; the mean precision is 2.7 %RSD. Aldrin is seen to have the
lowest recovery by LLE also.
Results for the disk extraction of the 608/8080 analytes from an average
groundwater composite at validation concentration levels are compiled in
Table III. Representative Aroclors were included in this study, not the
entire list. The data indicate that very good accuracy and precision can
be expected for disk LSE applied to actual samples. For these groundwater
composites, the mean ^Recovery was 92.6, the mean precision was 4.4 %RSD,
and the mean elution efficiency was 0.993.
Summary Table IV contains the results for the disk LSE of all the analytes
from reagent water at MDL concentration levels. The MDL's calculated from
the Empore data and the current laboratory MDL's for LLE are also listed.
Several of the Empore MDL's would have to be rerun at a lower level to
meet the requirements of 40 CFR, Part 136, Appendix B. Accuracy (%R)
values at these levels are good; the mean ^Recovery is 91.9%. Except for
a contaminant interfering with endrin aldehyde, the precision data are
also good, with the mean at 5.3 %RSD. Comparison of the MDL values for
LSE and LLE shows lower results for LSE in all cases except the endrin
aldehyde. The lower level of contaminants accounts for the lower MDL's
by LSE, a factor more evident in the MDL differences for multicomponent
analytes (Chlordane, Toxaphene, and the PCB's).
Table V shows analysis results for disk LSE of the analytes from actual
groundwater composites at the low MDL concentration levels. MDL's
calculated from these data compare favorably with the MDL's determined in
reagent water. Mean accuracy was 81.3% Recovery, and precision was 7.1
%RSD.
A liter of reagent water could be processed through the disk in an average
time of 7-8 min. The processing time for the "average" groundwater
composites ranged from 8-18 min. In production, several samples could be
extracted simultaneously. Many of these experiments were run using a
manifold with four extraction stations.
High Particulate Samples
Composites of samples with very high paniculate content were prepared to
study the procedure modifications that might be necessary to handle these
sample types. The composites had TSS contents in the range of 1.8-18 g/L.
In contrast, the "average" groundwater composites had TSS contents of 1-
5 mg/L. The high particulate samples took several hours to process
through the disk, even after allowing the particulates to settle and
11-185
-------
decanting most of the sample volume.
The problem of excessive filter time was managed with a pre-filter
positioned on top of the Empore disk. Five different pre-filters of
varying pore size were evaluated. The best results were achieved with the
Whatman graded density filter. With this prefilter, and decanting most
of the sample volume before transferring the bulk of the particulates, the
high particulate composites could be extracted in approximately 20-40 min.
The more critical problem presented by high TSS content samples is
effective recovery of analytes sorbed on the particulates. It requires
efficient elution of the filter cake of particulates that results from
sample filtered through the Empore disk and prefilter. Consistent
recoveries were obtained by adding ImL of methanol to the disk and
collected particulates with the first 5ml of ethyl acetate eluant, mixing
the particulates so that they were well dispersed in the eluant mix, and
than allowing some time for equilibration ("3 min). A larger K-D
concentrator tube is used to allow for the larger water/methanol layer (1-
4mL) in the total collected eluant.
Table VI shows Empore LSE results on two of the very high TSS content
composite groundwaters. Data for LLE, using continuous liquid-liquid
extractors, were also obtained for comparison. Recovery data are
consistently good for the LSE analyses. Precision data obtained for LSE
and LLE on these high particulate samples are very comparable.
The slightly higher total average recovery values for LSE compared with
LLE are the result of poorer recoveries for selected analytes by LLE:
Aldrin, Heptachlor, Methoxychlor, and the 4,4'-DDT, -DDE, and -DDD.
Various mechanisms may be at work contributing to the loss/degradation of
these analytes light and temperature conditions during the 18-hour
extraction, or particulate surface reaction effects.
Results on elution efficiency tests for four different high particulate
groundwater composites are compiled in Table VII. These data indicate
that the procedure used adequately eluted the analytes from the collected
particulates, prefilter, and Empore disk. More exhaustive elution is not
required.
CONCLUSIONS
The experimental data clearly validate the substitution of LSE using
Empore CIS disks for LLE in the analysis of the organochlorine pesticides
and PCB's tested. This study went beyond the required validation and MDL
determination in reagent water; the method was validated in the types of
water sample matrices typically encountered in an environmental
laboratory. LSE is not only equivalent to LLE, it is preferred because
of its time and cost benefits, and especially because of its environmental
benefits. It represents a substantial reduction in the volume of
hazardous solvents required for sample preparation.
1-186
-------
ACKNOWLEDGEMENT
The authors acknowledge, vith appreciation, the technical assistance of
Laura K. Bartoszek in conducting these extraction studies.
REFERENCES
(1) Determination of Organic Compounds in Drinking Water by Liquid-Solid
Extraction and Capillary Column Gas Chromatography/Mass Spectrometry,
J. W. Eichelberger, T. D. Behymer, W. L. Budde, Revision 2.1 (1988).
(2) Hagen, D. F.; Markell, C. G.; Schmitt, G. A.; Blevins, D. D. Analyt.
Chim. Acta 1990, 236. 157-164.(1) Method 525.
11-187
-------
TABLE I
EMPORE EXTRACTION OF ORGANOCHLORINE PESTICIDES AND
PCB's FROM REAGENT WATER AT VALIDATION CONCENTRATION LEVELS
SPIKE
SAMPLE ANALYSIS
Analyte
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
Chlordane
4,4'-DDD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
Methoxychlor
Toxaphene
PCB-1016
PCB-1221
PCB-1232
PCB-1242
PCB-1248
PCB-1254
PCB-1260
Ug/L
1.27
1.30
1.33
1.24
1.33
12.44
1.33
1.19
1.36
1.41
1.27
1.22
1.13
1.36
1.38
1.34
1.60
1.33
4.21
6.31
5.86
11.83
7.95
7.13
5.75
2.51
2.08
Mean8
%R
77
93
93
94
93
94
93
90
92
92
93
94
93
95
93
93
87
93
91
90
96
94
88
87
89
88
96
%RSD
3.9
3.3
2.9
3.3
3.0
4.0
2.7
3.5
2.8
2.7
3.0
2.5
2.3
3.4
3.6
2.3
2.4
2.8
3.1
6.0
3.9
1.8
1.5
2.0
3.1
3.5
2.6
Elution
Efficiency1*
0.990
0.995
0.992
0.996
0.995
0.991
0.994
0.993
0.992
0.993
0.991
0.989
0.992
0.991
0.987
0.992
0.992
0.992
0.992
0.995
0.988
N.A.
N.A.
N.A.
0.990
0.986
0.996
an=8, R=recovery (accuracy)
bA/(A+B) A=analyte concentration in first lOmL eluant
B=analyte concentration in second lOmL eluant
1-188
-------
TABLE II
LIQUID-LIQUID EXTRACTION OF ORGANOCHLORINE PESTICIDES
FROM REAGENT WATER AT VALIDATION CONCENTRATION LEVELS
Analyte
SPIKE
UE/L
SAMPLE ANALYSIS
Mean3
%R %RSD
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
4,4'-ODD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
Methoxychlor
an=4, R=recovery (accuracy)
241
236
251
093
246
220
214
177
248
264
240
225
196
268
308
Oil
263
4.000
65
84
91
93
85
88
87
91
88
86
88
90
95
80
85
113
86
90
3.4
3.8
1.5
3.0
3.6
1.7
1.8
0.9
2.6
2.3
1.9
2.0
2.6
3.7
2.1
6.8
2.8
2.1
11-189
-------
TABLE III
EMPORE EXTRACTION OF ORGANOCHLORINE PESTICIDES AND
PCB's FROM COMPOSITE AVERAGE GROUNDWATER
AT VALIDATION CONCENTRATION LEVELS
SPIKE
SAMPLE ANALYSIS
Analyte
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
Chlordane
4,4'-ODD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
Methoxychlor
Toxaphene
PCB-1016
PCB-1260
Ug/L
0.83
0.81
0.87
0.52
0.84
12.36
0.97
0.85
0.99
0.96
0.85
0.87
0.82
1.00
0.94
0.98
0.94
0.88
3.43
5.51
6.23
2.18
Mean*
_*IL
84
96
96
96
96
82
94
90
95
96
95
94
96
100
90
96
86
96
95
93
86
88
%RSD
6.2
4.7
4.4
4.6
4.6
6.0
4.5
3.9
4.3
4.6
4.4
4.1
4.2
4.6
3.6
4.3
5.4
4.5
4.4
5.0
2.4
1.7
Elution
Efficiency
0.991
0.997
0.994
0.997
0.997
0.978
0.995
0.993
0.993
0.994
0.994
0.993
0.994
0.993
0.988
0.995
0.993
0.994
0.993
0.985
0.996
0.995
an=8
11-190
-------
TABLE IV
EMPORE EXTRACTION OF ORGANOCHLORINE PESTICIDES AND
PCB'S FROM REAGENT WATER AT MDL CONCENTRATION LEVELS
SPIKE SAMPLE ANALYSIS MDLb. Ug/L
Mean3
Analvte ug/L %R %RSD LSE LLE
Aldrin 0.019 89 7.7 0.004 0.011
a-BHC 0.010 100 5.1 0.001 0.009
b-BHC 0.020 95 4.1 0.002 0.013
d-BHC 0.010 95 6.3 0.002 0.006
g-BHC (Lindane) 0.012 96 4.1 0.002 0.004
Chlordane 0.313 96 3.8 0.035 0.081
4,4'-DDD 0.017 93 4.9 0.003 0.006
4,4'-DDE 0.015 93 3.9 0.002 0.007
4,4'-DDT 0.019 84 6.0 0.004 0.006
Dieldrin 0.019 104 11.0 0.006 0.007
Endosulfan I 0.019 93 4.1 0.002 0.005
Endosulfan II 0.021 102 5.1 0.004 0.006
Endosulfan sulfate 0.020 82 4.0 0.006 0.020
Endrin 0.021 95 3.5 0.003 0.006
Endrin aldehyde 0.026 89 23.5 0.017 0.011
Endrin ketone 0.019 91 3.5 0.002 0.015
Heptachlor 0.024 90 4.1 0.004 0.005
Heptachlor epoxide 0.019 93 3.9 0.002 0.004
Methoxychlor 0.078 89 5.1 0.011 0.036
Toxaphene 0.378 100 3.0 0.034 0.093
PCB-1016 0.583 87 4.6 0.070 0.21
PCB-1221 0.613 84 2.4 0.048 0.37
PCB-1232 0.431 90 4.2 0.049 0.11
PCB-1242 0.246 101 5.7 0.042 0.096
PCB-1248 0.307 78 3.3 0.024 0.11
PCB-1254 0.303 86 2.4 0.019 0.098
PCB-1260 0.130 87 3.8 0.013 0.016
an=8
b40CFR, Part 136, Appendix B. The minimum detection limit (MDL) is
defined as the minimum concentration of a substance that can be
measured and reported with 99% confidence that the analyte
concentration is greater than zero.
11-191
-------
TABLE V
EMPORE EXTRACTION OF ORGANOCHLORINE PESTICIDES AND
PCB's FROM COMPOSITE AVERAGE GROUNDWATER
AT MDL CONCENTRATION LEVELS
Analyte
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
Chlordane
4,4'-DDD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
Methoxychlor
PCB-1016
PCB-1260
SPIKE
Ug/L
SAMPLE ANALYSIS
Heana
%RSD
0.016
0.011
0.024
0.011
0.014
0.317
0.018
0.022
0.017
0.018
0.018
0.029
0.022
0.024
0.027
0.019
0.027
0.019
0.076
0.563
0.127
109
82
71
80
76
80
77
59
90
90
89
62
74
76
82
83
71
84
85
91
95
8.4
6.1
6.5
7.6
6.8
11.2
5.7
5.2
5.6
14.0
5.7
10.3
8.1
5.4
12.9
6.2
6.5
6.0
6.0
2.5
2.7
MDL
Ug/L
0.004
0.002
0.003
0.002
0.002
0.035
0.002
0.002
0.003
0.007
0.016
0.006
0.017
0.003
0.008
0.016
0.004
0.003
0.012
0.038
0.010
ln=8
1-192
-------
TABLE VI
COMPARISON OF LIQUID-SOLID AND LIQUID-LIQUID EXTRACTION OF
ORGANOCHLORINE PESTICIDES FROM COMPOSITE HIGH PARTICULATE
GROUNDWATERS AT VALIDATION CONCENTRATION LEVELS (lug/L)
Analyte
to
CO
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
4,4'-ODD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptatchlor epoxide
Methoxychlor
AVERAGE:
GROUNDWATER #1: TSS = 18 e/L
5/9 EMPORE
Mean %R
n=5
72
78
80
77
80
79
78
72
81
78
76
81
82
66
83
72
81
79
77
LSE
%RSD
6.5
8.5
8.0
9.3
8.3
5.6
5.8
5.4
7.6
7.5
7.5
8.4
7.1
6.5
8.7
6.6
7.4
5.6
7.2
4/25
Mean ฃR
n=3
31
79
94
84
82
35
31
26
54
61
64
75
56
77
78
38
64
36
59
LLE
%RSD
6.9
4.2
3.3
3.6
4.1
4.4
5.7
3.6
6.2
5.4
5.4
4.2
4.8
3.8
5.5
4.5
4.8
0.7
4.5
GROUNDVATER #2
5/14 EMPORE
Mean %R
n-5 %.
59
73
77
75
76
62
60
65
73
74
72
77
77
77
77
61
75
68
71
LSE
RSD
8.8
4.9
3.6
5.6
4.7
6.1
6.8
5.7
3.8
4.2
4.4
4.5
3.4
6.0
4.3
5.0
3.6
2.2
4.9
: TSS = 15
5/14
Mean XR
n=4
45
89
101
94
93
43
35
32
67
74
72
84
71
89
85
37
74
48
68
s/L
LLE
2RSD
13.6
0.4
2.7
1.1
3.1
9.8
8.3
8.6
6.2
6.4
6.7
6.1
6.0
2.1
3.9
10.2
5.1
7.7
6.0
-------
TABLE VII
EMPORE EXTRACTION OF ORGANOCHLORINE PESTICIDES FROM
COMPOSITE HIGH PARTICULATE GROUNDWATERS AT
VALIDATION CONCENTRATION LEVELS (lug/L)
MEAN ELUTION EFFICIENCIES: A/(A+B)
TSS (g/L)
n =
Analvte
Aldrin
a-BHC
b-BHC
d-BHC
g-BHC (Lindane)
4,4'-DDD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
Methoxychlor
18
5
GW n
0.925
0.991
0.980
0.987
0.991
0.929
0.926
0.928
0.955
0.954
0.956
0.968
0.956
0.973
0.975
0.932
0.959
0.923
15
5
GW #2
0.864
0.977
0.964
0.972
0.976
0.861
0.856
0.873
0.923
0.930
0.925
0.937
0.926
0.953
0.948
0.898
0.935
0.873
2.4
6
GW #3
0.930
0.977
0.956
0.973
0.974
0.933
0.922
0.917
0.946
0.941
0.942
0.949
0.943
0.948
0.958
0.925
0.946
0.910
1.8
3
GW #4
0.959
0.988
0.976
0.986
0.982
0.968
0.962
0.954
0.957
0.967
0.970
0.976
0.969
0.965
0.980
0.952
0.968
0.954
Mean
0.920
0.983
0.969
0.980
0.981
0.923
0.917
0.918
0.945
0.948
0.948
0.957
0.949
0.960
0.965
0.927
0.952
0.915
AVERAGE: 0.956 0.922 0.944 0.968 0.948
1-194
-------
69 IMPROVING THE ANALYSIS OF SEMI-VOLATILE POLLUTANTS
Christine Vargo. Applications Chemist, Neil Mosesman, Technical Marketing Manager,
and Gary Barone, Research Chemist, Restek Corporation, Bellefonte, Pennsylvania 16823
ABSTRACT
Complex resolution and monitoring requirements established in EPA Method 8270
demand the use of capillary columns that have high inertness, efficiency, and thermal
stability. Recent polymer technology has been developed that substantially improves
capillary columns used in the analysis of semi-volatile pollutants. Columns produced with
this technology exhibit increased response factors for active compounds such as 2,4-
Dinitrophenol and 4-Nitrophenol and increased thermal stability, resulting in faster
analysis times and lower column bleed. Data and chromatograms will be shown
comparing the analysis of semi-volatile pollutants on conventional capillary column and
columns made with new technology. Direct comparisons will be shown of response
factors, analysis times, and column bleed between the columns.
INTRODUCTION
The complex resolution and monitoring requirements established under RCRA, SARA,
and SDWA, demanded improvements in existing analytical methodology. The EPA
responded to this need with the development of method 8270, a GC/MS method for the
analysis of semi-volatile pollutants. This method utilizes high resolution capillary
chromatography. Capillary columns have the required inertness to allow acidic, basic,
and neutral compounds to be analyzed simultaneously, the efficiency to separate highly
complex mixtures, and the thermal stability to analyze high molecular weight compounds.
With the widespread use of method 8270, it has become evident that not all capillary
columns have the necessary inertness for trace analysis of active compounds. Others do
not have the efficiency to resolve isomer pairs which cannot be distinguished by their
spectra alone. Still other columns do not have the thermal stability essential to analyze
high molecular weight compounds and reduce analysis times.
Recent polymer technology has been developed that yields a column with substantially
improved inertness, efficiency, and thermal stability for the analysis of semi-volatile
pollutants - the XTI-5.
The response of phenols is an excellent indication of capillary column inertness. Figure 1
shows a total ion chromatogram of fifteen phenols and six internal standards on a XTI-5
capillary column. The phenols show excellent peak symmetry and response at 50ng/w/,
indicating the inertness of the column.
Figure I - Phenols Look Exceptional With GC/MS Analysis on an XTI-5
Phenols
1. Phenol
2. 2-chlorophcnol
3. 2-methylphenol
4. 4-raelhylphcnoI
S. 2-nitrophenol
6. 2.4-diinethylphenol
7. Benzoie acid
8. 2,4-dichlorophenol
9. 4-chloro-3-melhy!phenol
10. 2,4,6-irichlorophcnol
11. 2.44-trichlorophcnol
12. 2,4-dinitrophenol
13. 4-nitropbenol
14. 4.6-dinitro-2-methylphcfiol
15. Penlachlorophcnol
IS
IS I. l,4-dichlorobenzene-d4
IS2. N.phthalene-d8
1S3. Accnaphthalcne-dlO
IS4. Phenanthracene-dlO
IS5. Cnrysene-dl2
IS6. Perylcne-dl2
30m, 0.25mm ID, 0.25fim XTI-5 (cat.# 12223).
l.Ojil splilless injection. 40ng of phenols and IS mix.
Oven temp.: 40ฐCio350ฐC 3 15ฐC/min. Hold ISmin.
Inj-temp.: 350ฐC Del.: MS (TIC)
Scan rale: 1.5scan/sec. Scan Range: 35-400AMU
11-195
-------
Many EPA and CLP methods require minimum response factors and linear calibration
curves over a concentration range of 20 to 160ng. Linear response factors are another
indication of column inertness and critical for environmental analyses. CLP protocols list
nineteen semi-volatile compounds as having minimum Relative Response Factors (RRF)
criteria of 0.010. These low response factors are due to these compound's poor linearity
and sensitivity. Figure 2 shows calibration curves on the XTI-5 for two erractically
performing compounds, 2,4-dinitrophenol and 4-nitrophenol. The calibration curves of
these phenols on the XTI-5 is very linear, even over a concentration range of 20 to 160ng.
Figure i- XTI-5 Calibration Curve
l"
S. U
I *
ac
01
30m.Oi23mmlD.OJ5|un XTM
Table 1 shows the response factors and percentage of Relative Standard Deviations (RSD)
calculated for 2,4-dinitrophenol, 4-nitrpphenol, pentachlorophenol, and benzoic acid. The
response factors were calculated by using the internal standard that elutes closest to the
compound (ie., dlO-phenanthracene for Pentachlorophenol and dlO-acenaphthene for 2,4-
dinitrophenol). Five data points at concentrations of 20, 40, 80, 120, and 160ng/w/ were
plotted for the calibration curve. The linear plots of the phenols clearly indicate the highly
inert nature of the XTI-5 column. All response factors meet or exceed the
criteria and all RSD percentages are well below the maximum deviation criteria of 20.5%.
Table ( - Response Factors are Linear for XTI-5 Capillary Columns
^Compound.*.-
2.4-dhuucfilwol
mm. CLP RF- 01
rn.CLPRF-.OI
mv.CLFRF-.05
ILCLPRF-WA
CoU
1
2
3
4
1
2
3
1
2
3
4
1
2
3
4
20ttf
0.402
0.418
0.38S
0.372
0.186
0.1S8
0.06S
0.127
0.184
0.181
0.160
0.275
0.302
0.321
OJ14
0.359
Stag
0.503
0.502
0.475
0.480
0.194
0.196
0.110
0.138
0333
0323
0.182
0243
0.453
0.453
0.426
0.451
-Star
0.523
0.550
0.496
0.498
0.208
0-223
0.122
0.157
0253
0.242
0.203
0.260
0.466
0.494
0,465
0.539
120nt
0.553
0.494
0.524
0.445
0325
0308
0.144
0163
0.268
0.233
0319
0272
0.491
0.523
0.520
0.433
' 160ng- ttmetn-
OJ61
O560
O.SSO
0.444
0^32
0.167
0.147
0133
0376
0315
0336
0267
0.504
0.583
0.485
0.489
0.508
0.507
0303
0447
0309
0.190
0.122
0144
0343
0319
0300
0263
0.443
0.475
0.442
0.454
stider. MtSD
0.057
0X150
0X65
0043
0.017
0.024
0.023
0014
0.033
0.021
0.027
0011
(XOBg
0.071
0.060
113%
9.9%
11.0%
96%
8.1%
12.8%
18.7%
97%
13.6%
9.6%
13.4%
4 3%
16.4%
105%
18.0%
Thermal stability is of extreme importance when analyzing high molecular weight
compounds, such as PNA's found in semi-volatile pollutant analyses. Column bleed can
present several problems when analyzing environmental samples. The rise in baseline
associated with column bleed can lead to inaccurate quantitative results, confuse spectral
interpretation and, in extreme cases, cause misidentification. Figure 3 shows total ion
chromatograms bleed profiles of the XTI-5, the conventional RV^> ^4 a ccunpctitors,
environmental column. MSD test results clearly show the XTI-5 exhibits the lowest bleed.
11-196
-------
Figure 3 - XTI Shows Lowest Bleed of any Column
Competitors Env.
Analysis Column
Rt-5
Time -
Oven temp.:
Inj. temp.:
Scan rate:
MS temp:
40ฐCio350<'C 3 15ฐC/min. Hold 15min.
350ฐC Del.: MS (TIC)
1 Jscan/sec. Scan Range: 35-400AMU
270ฐC
Figure 4 shows the analysis of the semi-volatile compounds monitored in EPA's Contract
Lab Program on a 30m, 0.25mm ID, 1 .Own XTI-5. Analysis times is complete in 45
minutes and bleed is minimal at 325ฐC.
Figure 4 - Semi-Volatile Pollutant Analysis on 30m, 0.25mm ED, l.Oum XTI-5
The new XTI-5 capillary column can improve the consistency and reliability of your semi-
volatile pollutant data. The technology used to produce these columns yields capillary
columns with improved inertness, increased efficiency, and higher thermal stability.
11-197
-------
File: A:\0701002.D
Operator:
Date Acquired: 3 Jan 91 5:38 pro
Method File Name: benzphen.M
Sample Name: 40ng std
Misc Info: rtx5 xti 30,.25,.25 40(1)-3JO @ 10/min
Bottle Number: 7
Abundance TIC: 0701002. D
3000000-
2800000-
2600000-
2400000-
2200000-
2000000-
1800000-
1600000-
1400000-
1200000-
1000000-
800000-
600000-
400000-
200000-
0-
rime ->
IS5
46
3
IS1
2
1.
i
5
7.
IS3
10
9
,
11
12
IS4
15
14
I13
1
IS
i. . u
5
IS6
J ~
-trfMtviirfftiU^.
^
5.00 10.00 15.00 20.00 25.00 30.00 35.00
11-198
-------
Figure 2 - XTI-5 Calibration Curve
CO
CO
fe
0.8
0.6
0.4
50
100
Nanograms
150
~*~ 4-nitrophenol ""*" 2,4-dinitrophenol
Co1# 18946B 30m, 0.25mm ID, 0.25nm XTI-5
200
-------
N>
8
Table*! - Response Factors are Linear for XTI-5 Capillary Columns
Compound
ColJ
2pซg
SOiig
8%&
^/JlOftgl;!;
160ng
mean ซ
jtd. dev.
f *fe iFfc C3tr*1i
/V m%,K?K,Jp
2,4-dinitrophcnol
min.CLPRF-.Ol
4-nitrophenol
min.CLPRF-.Ol
Pentachlorophenol
min. CLP RF-.05
Benzoic Acid
min. CLP RF-N/A
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
0.402
0.419
0.389
0.372
0.188
0.158
0.085
0.127
0.184
0.181
0.160
0.275
0.302
0.321
0.314
0.359
0.503
0.502
0.475
0.480
0.194
0.196
0.110
0.138
0.233
0.223
0.182
0.243
0.453
0.453
0.426
0.451
0.523
0.550
0.496
0.498
0.208
0.223
0.122
0.157
0.253
0.242
0.203
0.260
0.466
0.494
0.465
0.539
0.553
0.494
0.524
0.445
0.225
0.206
0.144
0.163
0.268
0.233
0.219
0.272
0.491
0.523
0.520
0.433
0.561
0.560
0.550
0.444
0.232
0.167
0.147
0.133
0.278
0.215
0.236
0.267
0.504
0.583
0.485
0.489
0.508
0.507
0.503
0.447
0.209
0.190
0.122
0.144
0.243
0.219
0.200
0.263
0.443
0.475
0.442
0.454
0.057
0.050
0.055
0.043
0.017
0.024
0.023
0.014
0.033
0.021
0.027
0.011
0.073
0.088
0.071
0.060
11.3%
9.9%
11.0%
9.6%
8.1%
12.8%
18.7%
9.7%
13.6%
9.6%
13.4%
4.3%
16.4%
18.5%
16.0%
13.2%
-------
Figure*^ - XTI Shows Lowest Bleed of any Column
3
c
JO
700000-
650000-
Competitors Env.
Analysis Column
RtK-5
XTI-5
Time ->?.oo 10.00
Oven temp.:
Inj. temp.:
Scan rate:
MS temp:
i
15.00 20.00
25.00
30.00
35.00 40.00
40ฐC to 350ฐC ฎ 15ฐC/min. Hold 15min,
350ฐC Del.: MS (TIC)
1.5scan/sec. Scan Range: 35-400AMU
270ฐC
-------
File: D:\DATA\40NGPP.D
Operator:
Date Acquired: 1 Apr 91 12:39 pm
Method File Name: PP.M
Sample Name:
Misc Info:
Bottle Number: 1
Abundance
1800000-j
1700000-
1600000:
1500000-
1400000:
1300000-
1200000:
1100000-
1000000-
TIC: 40NGPP.D
rime -> 15.00
20.00
25.00
30.00
35.00
40.00
11-202
-------
yn Electrospray Combined with Ion Trap Mass Spectrometry
'v for Environmental Monitoring
Robert 0. Voyksner and Hung-Yu Lin
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, NC 27709 USA
Environmental monitoring of many non-volatile or thermally unstable
polar organics relies on the development of sensitive, specific and cost
effective LC/MS techniques. Electrospray can meet these goals since it
has the capability to generate molecular ions from low pg quantities of
most environmentally relevant compounds. The coupling of electrospray
with an ion trap mass spectrometer (ITMS) offers the potential to gain
structural information through MS/MS without the additional cost of
multiple mass analyzers, as well as achieving better sensitivity than
conventional quadrupole mass analyzers. This paper reports on the
coupling of a commercial electrospray interface to an ITMS. The system
was evaluated for its use for environmental monitoring.
The electrospray source was interfaced to a second analyzer mounted in
the ITMS vacuum chamber with minimal changes to either commercial unit.
The use of a second analyzer in the ITMS minimized switch time between El
and electrospray operations. Ions formed in the electrospray interface
were gated into the ITMS analyzer, using the 180 V gating circuit employed
for El operations, with good efficiency and minimal losses from
collisional activation. The determination of numerous pesticides,
herbicides, dyes, and potential DMA adducts proved that the electrospray
ITMS combination could acquire high fg to low pg full scan spectra. These
sensitivities were 10-30 times superior to those obtained by electrospray
1-203
-------
on a quadrupole mass analyzer. The electrospray spectra of these
compounds usually only consisted of [M+H3* and/or [M+Na] ions and no
fragment ions. No thermal decomposition products were detected for the
thermally labile compounds analyzed. Occasionally other adduct ions were
detected, such as [M+NH41+ and [M+m-triethylamine]"1" when buffers such as
ammonium salts or triethylamine were used in the LC mobile phase. The use
of collisional activation decomposition in the ITMS analyzer proved useful
in generating MS/MS spectra from the protonated molecular ion or adduct
ion for each compound, resulting in fragment ions for identification or
confirmation.
Although the information described in this article has been funded
wholly or in part by the Environmental Protection Agency under contract
68-02-4544 to Research Triangle Institute, it does not necessarily reflect
the views of the Agency and no official endorsement should be inferred.
11-204
-------
Recent Advances In the Use of Supercritical Fluid Extraction for Environmental
Applications
J.M. Levy*, A.C. Rosselll, D.S. Boyer, M. Ashraf-Khorassani
Suprex Corporation, 125 William Pitt Way, Pittsburgh, PA 15238
One of the distinct advantages In using supercritical fluid extraction (SFE) is the
ability to achieve selective extractions based upon differences in threshold
solubilities of different analytes. Different threshold solubilities can be attained by
varying extraction pressures and/or temperatures. The addition of modifiers to
the supercritical fluid extracting phase has also enhanced the extraction
efficiency of specific analytes. Depending on the sample matrix, the
enhancement of solubilities could be offset by diffusion enhancement or by the
displacement of analytes from the outer or Inner surface of matrix particles.
There also Is the possibility of performing chemical reactions, such as acid
hydrolysis and functional group derivatization, during the SFE step thereby
achieving distinct ctivity enhancements for specific analytes in complex
matrices. Further selectivity enhancements can be achieved by utilizing different
adsorbents which are added to the extraction vessel with the sample or are
packed into secondary extraction vessels which are placed down stream of the
sample extraction vessel, in this work, each of these enhancements will be
investigated and demonstrated using a newly developed directly coupled
SFE/GC field portath system with environmental matrices such as soils, marine
sediments, drilling muds, sludges, and ashes.
II-205
-------
-79 USING SUPERCRITICAL FLUID EXTRACTION TO SEPARATE
'^ DIESEL FROM SOIL MATRICES
Carl A. Craig. Ph.D., Suman Prashar, and Jennifer Cunningham
BC Analytical, 1255 Powell Street, Emeryville, CA 94608;
Bruce E. Richter, Ph.D., and Amy Rynaski Lee Scientific, 4426
South Century Drive, Salt Lake City, Utah 84123
ABSTRACT
The initial and arguably the most crucial step in most
environmental analytical analyses is the separation of
analytes from the sample matrix. Separating or extracting
organic compounds from soil is currently accomplished by
several methods: heating, purging or by solvent extraction.
Of these, solvent extraction is the principal method used.
Because there is interest in reducing dependance on solvent-
use extraction technologies, we have examined Supercritical
Fluid Extraction (SFE) as a method to separate semi-volatile
organic analytes from soil matrices. Our initial data indicate
that SFE works very well to extract diesel from soil samples
and reduces the amount of solvent required by a factor of ten
compared to Ultrasonic Extraction (USE). This paper will
present the results of a direct comparison study of diesel
levels in soil extracts; where each soil sample was submitted
for both USE and SFE extraction and the extracts analyzed by
gas chromatographic methods.
INTRODUCTION
Supercritical fluid extraction of analytes from environmental
sample matrices is generating interest in the analytical
laboratory. The interest stems from reports of initial success
in separating organic analytes from samples.1"8 In addition to
the technical feasibility, there is a genuine concern among
the environmental lab community to reduce the amount of toxic
and hazardous solvents in the work-place. These two
influencing factors guided our efforts to investigate the use
of SFE. Our efforts have focused upon comparing the extraction
techniques (USE and SFE) directly. We accomplish this by
submitting soil samples to both extraction methods and
comparing the results of the GC-FID quantitation for the
extracts. The soil samples that we investigated were actual
soil samples which had been submitted by clients to BC
Analytical for diesel hydrocarbon analysis.
11-206
-------
Diesel hydrocarbons are herein defined as a class of C12-C25
hydrocarbons. A GC chromatogram of a common diesel standard
is provided in Figure 1. To quantitate the diesel in a sample
extract, integration of peak areas over the entire diesel
spectrum range was done. In addition to the integration,
diesel must be qualitatively identified by a characteristic
chromatographic fingerprint. This is necessary to insure that
diesel not gasoline or mineral spirits has been quantitated
by the integration.
EXPERIMENTAL
Supercritical fluid extractions were all performed using a
Dionex Model 703 Extraction System. All of the extraction
cells were 5 cm X 9.4 mm I.D.; the end caps contained
stainless steel frits (0.5 urn pore size). Eight extractions
were run simultaneously, the outlet of each cell was connected
to a separate temperature controlled restrictor. Collection
of the analytes involved a unique dual chamber vial and
double-sided Teflon coated septa. Approximately 10 raL of
methylene chloride were placed into each vial. The collection
system was then electronically cooled to 5ฐC. Extractions were
carried out with 100% CO2 that contained 1500 psi of helium
headspace (Scott Specialty Gases). All extractions were run
at 75ฐC and 300 atmospheres for a total of 15 min. The
restrictors were heated to 150ฐC to eliminate restrictor
plugging. The flow rate was on average 250 mL/min as gas.
Three one gram samples of each soil were weighed and placed
in separate vials. The sample size was selected for ease of
handling and extraction. No effort was taken to optimize the
weight of the sample extracted by SFE". Two of the three Ig
samples were extracted under the conditions detailed above.
The third Ig sample was used to determine the moisture content
of the soil.
Each SFE extract (approximately 10 mL) was reduced to less
than 5 mL total volume under a flush of high-purity grade
nitrogen gas. The extracts were then transferred to 10 mL
concentrator tubes via pasture pipets that were packed with
anhydrous sodium sulfate. Each extract vial was then rinsed
with 1-3 mL of methylene chloride, and the rinsate added to
the concentrator tubes. The extracts were then reduced to 1
mL final volume under a flush of nitrogen. The 1 mL extracts
were quantitatively transferred to 1.5 mL vials with Teflon
lined septa and screw caps. These extracts were stored at
4 ฐC prior to GC analysis.
Ultrasonic extraction was carried out in accordance with EPA
method 3550A. Flow diagrams of the USE and SFE methods are
shown in Figures 2 and 3.
11-207
-------
Gas chromatographic identification and quantitation of diesel
in the USE and SFE extracts was accomplished with a HP 5890
gas chromatograph using a 30 meter DB5 capillary column. Prior
to each daily run, column and septa conditioning was conducted
at 300ฐC for 90 minutes. The oven temperature was brought to
40ฐC and allowed to stabilize. The instrument was calibrated
using 50, 100, 250, 500, and 1000 ppm diesel standard. Data
were analyzed with Nelson 2600 software.
RESULTS AND DISCUSSION
This investigation involved analyzing extracts from 42 soil
samples. The soil samples were submitted by clients to BC
Analytical for analysis of diesel hydrocarbons. Once the study
was under way, all of the samples submitted to BCA for diesel
analysis were extracted by both USE and SFE methods. Many of
the soil samples investigated have less than the reporting
detection limit of diesel in the USE extract. Thirty two of
the 42 samples were reported as "not detected" for diesel
hydrocarbons by ultrasonic extraction and GC-FID quantitation.
(See Table 1.) The remaining ten samples had reported
quantities of diesel in the 3550A extract ranging between 1-
2800 ppm.
Similarly, 32 of the 42 SFE extracts were confirmed not to
contain diesel hydrocarbons. Supercritical extracts were
reported to contain diesel if duplicate extracts contained
diesel. (See Table 1.) In several cases, a duplicate extract
was unavailable; the data were then based upon a single
replicate.
Both extraction methods yielded ten soil extracts with diesel
hydrocarbons identified. Seven of these ten extracts were for
the same soils independent of the extraction method. (See
Table 2.) For three soils, a diesel quantity was reported
using extraction method 3550A and no confirmation was reported
in the SFE extracts. Alternatively, three SFE extracts gave
reportable levels of diesel when the 3550A extract had no
diesel hydrocarbons reported. However, there is generally good
agreement between the incidence of soil extracts containing
diesel hydrocarbons using these two extraction methods.
During the investigation, the accuracy of the SFE extraction
was not measured using surrogates or spiked soils. Therefore
the "true" value of analyte in the soil is taken to be the
value reported for the 3550A method. Of the seven soil samples
for which both extraction methods gave positive results for
diesel analysis, only two results were significantly
different. Sample 32 was reported to give 700 ppm diesel via
method 3550A while the same soil gave only 115 ppm in the SFE
11-208
-------
extract. However, the converse was also observed. For soil 18,
the reported quantity of diesel in the 3550A extract was 46
ppm and the SFE extract gave 535 ppm.
Most of the GC-FID chromatograms of the SFE extracts had a
characteristic pattern which was not attributed to hydrocarbon
in the soil. (See Figure 4.) However this pattern was not
observed in the sample blank. This interference was manually
subtracted from each chromatogram to allow for quantitation.
The source of the contamination is unknown and is currently
under investigation. As a result of the contamination, the
reporting detection limit (RDL) for the SFE extracts was set
at 5 ppm. The problem that one encounters with a 5 ppm RDL is
that a hydrocarbon pattern (such as the mineral spirits
identified in Figure 4) may be observed but not reliably
quantitated because of the high level of interference.
SUMMARY
Supercritical fluid extraction will find wide use in the
environmental laboratory because of the ability to extract
organic analytes from soil matrices without the use of large
volumes of hazardous solvents. However, before this will
happen, the utility of SFE techniques on actual field samples
must be demonstrated. Our results indicate that SFE works very
well to extract diesel hydrocarbons from soil matrices. In
fact these results indicate that SFE is as efficient as
ultrasonic extraction in removing diesel from sample matrices.
Of the 42 samples that were analyzed, SFE gave similar results
to those obtained using USE. In addition the differences that
were observed were not one sided. In three cases diesel was
identified in the ultrasonic extract and was not confirmed in
the corresponding SFE extracts. Similarly, there were three
sets of SFE extracts where reportable levels of diesel were
identified, yet no diesel was found in the USE extracts. In
these six instances, the reported quantity of diesel in the
soil was less than 31 ppm. The differences observed between
these extracts might very well be attributed to non-
representative sample sizes, or sample inhomogeneity. During
the next phase of the research we plan to investigate the
effect of sample size on extraction optimization. In addition,
a reduction in the interference observed by the FID detector
in the SFE extracts must be eliminated in order to lower the
detection limit to levels equivalent to current methods.
Supercritical fluid extraction does significantly reduce the
amount of solvent required for the extraction of diesel
hydrocarbons from soil samples from approximately 400 mL for
EPA method 3550A to approximately 20 mL for the SFE. Because
of these initial successes we plan to continue to investigate
the use of SFE to separate diesel from soil matrices.
11-209
-------
REFERENCES
1.) Hawthorne, S.B. Anal. Chem. 1990, 62, 633.
2.) Yu, X.; Wang, X.; Bartha, R.; Rosen, J.D. Environ. Sci.
Technol. 1990, 24. 1732.
3.) King, J.W. Journal of Chromatographic Science 1989, 27,
355.
4.) Ndiomu, D.P.; Simpson, C.F.; Analytical Proceedings 1989,
26, 393.
5.) Janda, V.; Steenbeke, G.; Sandra, P. Journal of
Chromatography, 1989, 479, 200.
6.) McNally, M.P.; Wheeler, J.R. Journal of Chromatography
1988, 435, 63.
7.) Wright, B.W.; Frye, S.R.; McMinn, D.G.; Smith, R.D. Anal.
Chem. 1987, 59, 640.
8.) Hawthorne, S.B.; Miller, D.J. Anal. Chem. 1987, 59, 1705.
11-210
-------
C\J
Figure 1 . Capillary gas chromatogram of a 250 ppm
diesel hydrocarbon standard
-------
Ultrasonic Extraction (SW-846 Method 3550A)
General Tasks Analysis Specific Tasks
CO
Weigh 30g of sample
into a 400 mL beaker
'Add 150 mL of extraction/
solvent to beaker /
I
Sonicate sample
for 5 minutes
I
r Decant solvent over NaฃO4/
into evaporator
1
Add 100 mL of extraction
to beaker
Sonicate for 5 minutes.
I
Rinse beaker with 50 mL
of extraction solvent
I
Decant solvent over
into evaporator
I
Distil off solvent
Evaporate remaining solution x
to correct final volume
(Transfer sample to vial)
Pesticides. PCB. .. Hexane/Acetone
Semivotetie organics.Diesel.. .
Total volume of
solvent required
400 mL
Solvent exchange to hexane
for pesticide and PCS analysis
1 mL for ofesel analysis
2 mL for semivolatle organics
10 mL for pesticides and PCB
Plods! cleanup... Pesticides
Acid cleanup.. .PCB
Figure 2. Flow diagram of ultrasonic extraction
1-212
-------
Supercritical Fluid Extraction
General Tasks Analysis Specific Tasks
Weigh 1g of sample
into an extraction thimble
T
Load eel Wo extractor
i
Extract for
15 minutes
I
Collect analyte h solvent trap
Pesticide. PCS. ..Hexane
Semivolatite organics, Diesel.
Transfer solvent with pipet anc
NagS(j into concentrator
Rinse vial with
5 mL of solvent
Total volume of
solvent required
20 mL
Evaporate remaining solution/
to correct final volume
I
("Transfer sample to vlaf)
1 mL for diesel analysis
2 mL for semivolatfle orgaracs
10 mL for pesticides and PCB
Fkxisl cleanup. . . Pesticides
Acid cleanup. ..PCB
Figure 3. Flow diagram of Supercritical fluid extraction
11-213
-------
03
O
CO
CO
-t'
o
10
C\J
IB
(J
ID
n
r~
I
Low level (3. 2 ppm) mineral spirits
A I I I I \ l_
_1 I L
Contaminants
U
CTi
tr
sc
a
*
ซ
CB
J*^,
1 1 1 1
Figure 4. Capillary gas chromatogram of supercritical fluid extract
for sample number 31
C\
-------
TABLE 1. RESULTS OF DIESEL HYDROCARBON ANALYSES
SAMPLE
NUMBER
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
BY EPA 3550/8015 BY SFE/8015
(mg/kg) (mg/kg)
REPLICATE 1 REPLICATE 2
2800 2280
<1 17
<1 <5
<1 <5
<1 <5
<1 8.5
<1 <5
<1 7.0
<1 <5
<1 <5
<1 <5
<1 <5
<1 <5
<10 <10
<1
<1
<1 <5
46 510
560 635
5 <5
<1 <5
<1
6
<10 <10
<10 <10
<1 <5
<1 <5
1 <5
<1 <5
<1
<1 <5
700 118
<1 <5
140 65
<1
<1 <5
30 <5
<1 <5
30 55
15 6.6
<1 <5
<1 <5
<5
12
<5
<5
9.5
<5
7.8
10
<5
<5
<5
<5
<10
<5
<5
<5
560
581
<5
<5
17
5.2
<10
<10
<5
<5
14
<5
<5
<5
112
<5
167
<5
<5
<5
<5
15
<5
<5
<5
PERCENT
MOISTURE
19
20
23
19
12
19
4.7
21
14
6.0
14
22
14
12
10
11
17
6.6
14
7.2
11
19
13
11
15
15
21
23
25
20
22
22
19
20
25
26
18
16
15
18
19
11-215
-------
TABLE 2. EXTRACTS OF SOILS CONTAINING DIESEL HYDROCARBON
LEVELS GREATER THAN THE REPORTING DETECTION LIMIT
SAMPLE
NUMBER
1
6
8
18
19
20
22
23
32
34
37
39
40
BY EPA 3550/8015
(mg/kg)
AVERAGE
2800
<1
<1
46
560
5
<1
6
700
140
30
30
15
BY SFE/8015
(mg/kg)
OF 2 REPLICATES
2280a
9.0
7.4
535
608
<5
17a
5.2
115
116
<5
35
<5
a) Single extract analysis (no replicate available)
1-216
-------
70 CREATIVE REVIEW OF "TENTATIVELY IDENTIFIED COMPOUND"
' DATA USING THE RETENTION INDEX
WILLIAM P. ECKEL
VIAR AND COMPANY
300 NORTH LEE STREET
ALEXANDRIA, VIRGINIA 22314
1-217
-------
INTRODUCTION
Identifying non-target compounds in gas chromatography-mass spectrometry (GC/MS)
data is a difficult and time-consuming task, even for those trained in interpretation of mass
spectra. It is rare that a compound can be "tentatively identified" with confidence when the
only information available is the mass spectrum and the computer library matches.
Aside from prior knowledge of what types of compounds to expect in a particular
sample, the only other piece of data available to the data reviewer when interpreting a
spectrum is the GC retention time. Because absolute retention times are dependent on a large
number of experimental conditions, the retention index was developed to express retention
data relative to a standard set of compounds. The original system, called the Kovats index,
dates from 1958 and uses the normal alkanes as the retention index standards.
A retention index system was developed in 1979 by Lee and co-authors (1) for use in
identifying polycyclic aromatic hydrocarbons (PAH). In the "Lee" retention index system, the
retention index standards and their retention indices are naphthalene (1=200.00), phenanthrene
(1=300.00), chrysene (1=400.00) and picene or benzo(ghi)perylene (1=500.00). Beeause the
perdeuterated analogs of the first three of these are used as internal standards in the several
variations of EPA method 625 for extractable compounds, the Lee retention index can be used
NOW by data reviewers in identifying unknown compounds. The Lee retention indices of
several hundred compounds of environmental interest are available in the literature (1-5).
1-218
-------
CALCULATION OF THE LEE RETENTION INDEX
The Lee retention index, I, is defined as:
I = 100 ( rtunk - rtz / rtz+1 - rtz ) + 100 (Z).
where rtunk is the retention time of the unknown compound, rtz and rtz+j are the retention
times of the bracketing retention index standards, and Z is the number of benzene rings in the
retention index standards. To summarize:
Standard Retention Index Z
naphthalene 200.00 2
phenanthrene 300.00 3
chrysene 400.00 4
For unknowns which elute before naphthalene or after chrysene, the retention index is
"projected" using the two retention index standards closest to the unknown compound.-
11-219
-------
APPLICATION OF THE LEE RETENTION INDEX TO DATA REVIEW
This paper presents some applications of the Lee retention index system to the
identification of compounds in real environmental samples.
The first example (Figure 1) is the spectrum of a dimethylphenol whose experimental I
was 194.35. The target compound 2,4-dimethylphenol was also found in this sample at
1=190.70. Judging from the known retention indices of the isomeric dimethylphenols, and the
measured bias between the experimental index and the known index for the 2,4- isomer, the
most probable identification for the unknown is 3,5-dimethylphenol.
Figure 2 shows the spectrum of a polycyclic aromatic hydrocarbon of molecular weight
252. Three of the library matches are target compounds which were also found in the sample
at different scan numbers than this unknown. Their scan numbers, and experimental and
known retention indices are shown. Based on the known retention order of these compounds
(as shown by the retention indices) and considering the experimental bias, the proximity of
the unknown to the scan number for benzo(a)pyrene identifies it as benzo(e)pyrene.
Figure 3 shows the spectrum of an unknown with an apparent molecular weight of about
190. Comparison of the experimental spectrum with the library spectra of compounds of MW
192 shows much more intense ions at m/z 190 and m/z 189 in the unknown spectrum. This
suggests that the unknown spectrum may not be that of a pure compound, but of two
coeluting compounds. In fact, the experimental retention index for the unknown lies just
between the known retention indices of the top two library matches, one of which has MW
190 and the other MW 192. Thus both cyclopenta(def)phenanthrene and 9-
raethylphenanthrene appear to be present in the unknown spectrum.
Figure 4 is the spectrum of another pair of coeluting compounds, this time of dissimilar
chemical classes. An isomer of the top library match is present, judging from the retention
index. However, a hydrocarbon pattern is also present at m/z 43, 57, 71, 85 etc. The library
search results do not suggest the presence of an alkane, but the retention index of pentadecane
is close enough to the experimental index that it can also be tentatively identified in this
spectrum.
Finally, Figures 5a through 5d show a series of normal alkanes that were found in a
sample. As is usual with alkanes, the top library matches are of widely varying chain length.
Calculation of the retention index quickly narrows the possibilities to one or two compounds
for each spectrum. Note that the tentative identification for each spectrum is not among the
top library matches.
1-220
-------
CONCLUSIONS
The Lee retention index is an easily calculated bit of information which can be
extremely valuable in the identification of unknown compounds. It is useful in identifying
specific members of homologous series with identical spectra; coeluting compounds of similar
and dissimilar chemical classes; specific positional isomers; and structural isomers. More
sophisticated applications of the existing Lee retention index data, using the principles of gas
chromatography and structure-retention relationships, are also possible.
11-221
-------
REFERENCES
1. Lee, M.L., Vassilaros, D.L., White, C.M., and Novotny, M., "Retention Indices for
Programmed-Temperature Capillary-Column Gas Chromatography of Polycyclic
Aromatic Hydrocarbons", Analytical Chemistry 51, 768-773, 1979.
2. Vassilaros, D.L., Kong, R.C., Later, D.W., and Lee, M.L., "Linear Retention Index
System for Polycyclic Aromatic Compounds. Critical Evaluation and Additional Indices",
Journal of Chromatography, 252, 1-20, 1982.
3. Willey, C., Iwao, M., Castle, R.M., and Lee, M.L., "Determination of Sulfur
Heterocycles in Coal Liquids and Shale Oils", Analytical Chemistry, 53, 400-407, 1981.
4. Rostad, C.E., and Pereira, W.E., "Kovats and Lee Retention Indices Determined by Gas
Chromatography/Mass Spectrometry for Organic Compounds of Environmental Interest",
Journal of High Resolution Chromatography and Chromatography Communications, 9,
328-334, 1986.
5. U.S. Environmental Protection Agency, Office of Water Regulations and Standards,
Industrial Technology Division, "Analytical Methods for the NationaLSewage Sludge
Survey. Method 1625, Revision C. Semivolatile Organic Compounds by Isotope Dilution
GCMS", August 1988.
-------
Figure 1
Identification of Specific Dimethylphenol Isomer
1088
SAMPLE 1
C8.H10.0
1088
MWT122
BPK 107
RANK 1
JT IN 2643
K PUR 866
CO
C8.H10.0
1088
M WT122
BPK 107
RANK 2
IN 2634
PUR 864
C8.H10.0
1088
MWT122
BPK 122-
RANK 3
IN 2645
PUR 863'
LIBRARY SEARCH
08/04/8516:34:00 + 8:02
SAMPLE:! ULCCJS7689 (8-1-85)
DATA:GH057689A16*534
ENHANCED (108 2NOT)
UNKNOWN
BASE M/E: 107
RIC:457727
lexp= 194.35
lref = 198.95
I ...l...l..,,, ,,
PHENOL,3,4-DIMETHYL- CAS* 95-65-8
lref = 201.49
PHENOL,2,6-DIMETHYL- CAS* 576-26-1
lref= 184.08
50 100 150 200 250 300
IJUMMM uiiiiiriiT-jiim,., ...,,|i,,,.,,,,. |. in i M,, , ii ,, , ,, , ,, | , i,
PHENOL,2,3-DIUETHYL- CAS* 526-75-0
Dimethylphenol
Isomer
2,6
2,4
(Target Compound)
3,5
2,3
3,4
'ref
184.08
192.69
196.53
198.95
201.49
exp
190.70
194.35
Bias
+1.99
+2.18
-------
Figure 2
Identification of PAH with Molecular Weight of 252
8
1237
SAMPLE '
C20.H12
MWT252J
3 PK 252 I
* 23964 j
C20.H12
M WT 252 j
3 PK 252 i
ป 23966 j
PUR 914 '
C20.H12
MWT252.
3 PK 252
* 23965
C20.H12
1237
MWT252 j
3 PK 252
RANK 4 j
ป 23961
C20.H12
1237
MWT252J
3 PK 252
RANK 5 J
9 23962
PUR 905 =
LIBRARY SEARCH DATA: 61 127645 ป 2758 BASE M/Z: 252
06/08/906:59:00*45:58 CALI: B1127645 1 2 RIC: 51263
SAMPLE: Lป1 12764
lexp = 447.12 |
;Ti m- i-tu -,- i '' 'fl11 V
BENZO(J)FLUORANTHENE
lre,= 440.92 |
f
... ,ll ..ill i
8ENZO(K)FLUORANTHENE
|fe( 442.56 |
BENZO(8)FLUORANTHENE
lre) = 441.74 |
^ittrrnr -, -; - '' I1' V '
BENZO(A)PYRENE
lret- 453.44 |
BENZO(E)PYRENE
lrel- 450.73 |
,1 ,||
^.^Mt.lUMt . i^.4V>ป ,, Jtt,t,V,,~,-,t ปlj-.ifi ff-r,; 4 t^XtTi '.'TK-n'r-r T>I t r r i r'l-ri'A '
1
PAH
Benzo (j)
Fluoranthene
Benzo (b)
Fluoranthene
Benzo (k)
Fluoranthene
Benzo (e)
Pyrene
Benzo (a)
Pyrene
ire(
440.92
441.74
442.56
450.73
453.44
Scan
2698
2703
2758
2769
IซP
438.03
438.79
447.12
448.79
Bias
+3.71
+3.77
+3.61
+4.65
250
-------
Figure 3
File>90868HOU
SpkAbl1258
Identification of Coeluting PAHs
INST C 2/7/90 NEAT 340 PHC/DRS CLP 5 Scan 2272
SUBAODOVC 21.72min.
UNKNOWN(S)
50 62 70 * * 109 126 152 ^
'""I1" '"'1 !! {IIM..IMI . .,.,1111 l|limH III
File >BIGDD Phenanthrene, 4-methyl-
dn\* Ak QQQQ
181
,,,,. ,,,,,,,,,,
Scan 46878
192
198
I /
)miii|imni|
O.OOmin.
&
Ol
-;
50 63 75 82 95 ^ ^ 152 ^ 174
/ s^ / ' \ 1 / ' / \\
File>BIGDB 4H-Cyclopenta(def)phenanthrene Scan 46722
SpkAb9999 0.00 min.
94 1g3
50 63 74 87 ^ ^ ^ ^ ,50 - , ^
/ ^ / ' ,| / / / .1 \
I'M1 1 'i'" .i.|..i ,,[,,.,..
File>BIGDB Anthracene, 1-msthyl- Scan 47139
Spk Ab 9999 0.00 min.
44 63 74 87 K ^ ^ ^ ,52 ^ 176
X -^ / \\ / / X ''I '
192
194
190
196
192
194
I ..,ป.|. ,.!...,
60
100
120
140
160
180
200
Sample: HQ14 INSTC injected: 2/07/90 17:02
TIC # 26 Area = 390337.0 Tentative Cone = 510.00
BEST MATCHES
1. Phenanthrene, 9-methyl- 192 C15H12
2. 4H-Cyclopenta(def)phenanthrene 190 C15H10
3. Anthracene, 1-methyl- 192 C15H12
4. 9H-Fluorene, 9-ethylidene- 192 C15H12
5. Phenanthrene, 9-methyl- 192 C15H12
6. Methyl-phenanthrene or
Methyl-Anthracene 192 C15H12
7. Anthracene, 2-methyl- 192 C15H12
PAH
2-Methyl-
anthracene
Cyclopenta(def)
phenanthrene
Unknown
9-Methyl-
phenanthrene
4-Methyl-
phenanthrene
1 -Methyl-
anthracene
'ref
321.57
322.08
323.06
323.17
323.33
'exp
322.99
MW
192
190
192
192
192
-------
Figure 4
Identification of Coeluting PAH and Alkane
Filg>90868H014
BpkAb4430
57
INST C 2/7/90 NEAT S40 PHC/DRS CLP 5 Scan 1 680
SUBADDDVC 16.66min.
UNKNOWN
168
-
-.
J
i
i
^
43
.1
51
\
' 42
51
|
6C
5C
/
iniin'ti
File>BIGD
BpkAb999
76
\ ^
Fits >BIGD
BpkAb999
Fite>BIGO
BpkAb999
65
lm|iiiW
3
9
^
9
9
69
*-"
JrniTn
"I1""
3
9
77
\
85 ^
... 153
of, 111 126 1.41 \
/ / / / \
, , , >>.i.iK ,u ,1, , . , , . . , 1 1 ,\ , . , . M . . , . , 1 1.1, , ..iii
''"' |IIIII11|IIMIIIIIII1| | l|l
1,1'-Biphenyl,2-Methyl
10? 141 153
\. 115 128 ^ \
,| ^~~-- / / \ JL Jil
s-Tetfazine,3,6,-bis(dimelhylamino)-
1
x
Benzene, 1,1'-methylenebis-
91 nQ
\ 102 115 128 i
\
188 21-~^_
| / ^
Scan 41631
0.00 min.
168
X
I
182 183
Scan 41439
0.00 min.
38 169
^
'" " I i
Scan 41627
0.00 min.
168
170
Sample: H014 INSTC Injected: 2/07/90 17:02
TIC # 11 Area = 173945.0 Tentative Cone = 200.00
BEST MATCHES
1. 1,1'-Biphenyl, 2-methyl- 168 C13H12
2. s-Tetrazine, 3,6-bis(dimethylamino)- 168 C6H12N6
3. Benzene, 1,1 '-methylenebis- 168 C13H12
4. Benzene, 1,1'-methylenebis- 168 C13H12
5. 2H-Pyran-2,4(3H)-dione,3-acetyl-6-methyl- 168 C8H804
6. Benzene, 1,1'-methylenebis- 168 C13H12
7. Benzene, 1,1'-methylenebis- 168 C13H12
120
140
160
PAH
2-Methyl-
biphenyl
Diphenyl-
methane
3-Methyl-
biphenyl
4-Methyl-
biphenyl
Unknown
Pentadecane
(MW212)
'ref
239.84
243.35
254.33
256.12
256.75
exp
256.56
-------
Figure 5a
Identification of Normal Alkanes
LIBRARY SEARCH
08/04/85 14:46:00117:46
SAMPLE: 1 UL CCK57686 (8-1-85)
DATA: GH057689A16* 1182
ENHANCED(! OB 2NOT)
UNKNOWN
BASEM/E:57
RIC:1505270
1081 q
SAMPLE j
!
C8.H10.0
1081
MWT296
BPK 43-i
RANK 1 :
IN 21792 :
PUR 798-
C8.H10.0
1081 3
M WT352 =
BPK 43-:
RANK 2
IN 25265 :
PUR 795:
C8.H10.0
1081 3
M WT 492 :
BPK 43-i
IN 25265
PUR 785 -.
J
"'I'
(
,
II
Jl
(|
h
lexp = 433.60 |
HENEICOSANE CAS# 629-94-7
lref = 347.42 j
If il (1 1 1 , r iimun T r r I.HI ,
PENTACOSANE CAS# 629-99-2
lref = 400.45 |
It M n i , .
PENTATRIACONTANE CAS# 630-07-9
lref = 538.06 |
1
,,l ,,U...,I.1 1 1 M
400
-------
N>
Figure 5b
Identification of Normal Alkanes
LIBRARY SEARCH
08/04/8514:46:00* 18:24
SAMPLE: 1ULCCS57686 (8-1-85)
DATA: GH057689A16 #1224
ENHANCED (10B2NOT)
UNKNOWN
BASE M/E: 57
R 1C:1638390
SAMPLE j
C25.H52
1122 3
MWT352 :
BPK 43 -i
RANK 1 :
IN 25265 :
PUR 766
C35.H72
1122 a
MWT492 :
BPK 57 -i
RANK 2 i
IN 29645 :
PUR 760 -.
C21.H44
1122 3
MWT296 :
BPK 43 J
RANK 3 :
IN 21792 :
PUR 756-
,,.|l
J
T.i.ffl
50
lexp = 450.61 |
I - nil MI'* iiiimn tw. '. i,....-.f !-".: ;!'
I
r JlL Jl L \\ w ..L ..... ....
PENTACOSANE CAS# 629-99-2
lref = 400.45 j
ll h ii ii i I . .
PENTATRIACONTANE CAS# 630-07-9
lref = 538.06 j
( | n ,i ...
HENEICOSANE CAS# 629-94-7
lref = 347.42 j
i II i
i i il il ,, . . ..
100 150 200 250 300 350 400
-------
Figure 5c
Identification of Normal Alkanes
LIBRARY SEARCH
08/04/8514:46:00+ 19:08
SAMPLE: 1ULCCW7686 (8-1-85)
DATA: GH057686A16* 1237
ENHANCED (108 2NOT)
UNKNOWN
BASEM/E:57
RIC:1138680
1052 q
SAMPLE J
!
C35.H72
1052
MWT352 i
BPK 43-:
RANK 2 i
IN 25265
PUR 790-
C25.H52
1052 a
MWT352
BPK 43-
RANK 2
IN 25265 =
PUR 790^
C22.H46
1052 3
M WT310
BPK 57-
RANK 3
INI 22753
PUR 784-
,,,|l
Miff
||
||
|,
1 1 1 | 1
Jl
||
||
.nil
lexp = 470.45 |
J ,
PENTATRIACONTANE CAS# 630-07-9
lref = 538.06 1
1
PENTACOSANE CAS* 629-99-2
lref = 400.45 |
|i ii ,i ...... r .
1'" ' | |lllllll IHIIJ Ijllll.l 1 III II |ll 11 ,.1111111.1,11111111111,111111111
DOCOSANE CASK 629-97-0
lref = 361.53 |
:-::: :-:.:-:.:':.:':.:':.:.:.:.:-:.:->:':-:.:.:.r.>:.:.:.:-:-:.:-:.:.:.:.:.;.:.:.
\ '' '' '1 I ' v i -i v r r
50
100
150
200
250
300
350
400
450
-------
ro
Figure 5d
Identification of Normal Alkanes
LIBRARY SEARCH
08/04/8514:46:00 + 20:01
SAMPLE: 1 ULCC*57686 (8-1-85)
DATA: GH057686A16* 1332
ENHANCED (10B2NOT)
UNKNOWN
BASEM/E.-57
RIC: 683007
SAMPLE j
C22.H4S
1146 a
UWT310
BPK 57-
RANK 1
IN 22753
PUR 751
C21.H44
1146 q
MWT296:
B PK 43-:
RANK 2
IN 21792
PUR 747:
C25.H52
1146 3
MWT352 :
BPK 43 H
RANK 3 i
IN 25265 :
PUR 742:
1, ,|l
L
ir...ff
II
Jjl
||
,,,ll
II
ooc
m-l
HEN
ml
PEN
1
lexp = 494.33 1
TilTi nltyiiifVI'ii itlii i VTJ t V n n 1 iVn i ifi ill Tun il il il nil i nil 1 1 il n nil
OSANE CASI 629-97-0
lref = 361.53 j
*
i I fti niMi i i i i^rrrTit ' < V i f |"| ii i I't i I I i*f I I I |T| I I 1 fl 1 1 1 1 1 1 I J 1 1 1 1 1 1 1 li |] I I I 11
EICOSANE CAS# 629-94-7
lref = 347.42 1
FACOSANE CAS# 629-99-2
lref = 400.45 !|
. . .I.L.ll ... IL,,.. ,1. . , .!...ซ ซ...ซ... H._..W. .. . > 4^ ป
50
100
150
200
250
300
350
-------
Alkane
n-C27
n-C28
n-C29
n-C30
n-C31
n-C32
n-C33
"ref
425.51
437.68
448.93
460.36
471 .96
484.94
499.88
'exp
433.60
450.61
470.45
494.33
Spectrum
5a
5b
5c
5d
Bias
+4.08
+1.68
+1.51
+5.55
11-231
-------
74 HIGH EFFICIENCY GPC CLEANUP OF ENVIRONMENTAL SAMPLES -
COLUMN OPTIMIZATION
Gary J. Fallick. Richard Cotter, Waters Chromatography Division, Millipore
Corporation, 34 Maple Street, Milford, Massachusetts 01757; Russell Foster,
Richard L. Wellman, Resource Analysts Inc., P.O.Box 778, Hampton, New
Hampshire 03842
ABSTRACT
Gel permeation chromatography (GPC) has been used for almost two decades to
remove unwanted high molecular weight compounds from agricultural and
environmental samples prior to final analysis. Virtually all of this work has been
done with 25mm ID x 40 to 100cm long columns packed with low efficiency 37-75
micron particles.
Using smaller diameter column packing particles produces major gains in column
efficiencies, enabling the same separation to be done with much smaller columns.
This enables the cleanup to be done in significantly less time using considerably less
solvent.
This study was done to determine the optimum grade of GPC packing material and
preferred column dimensions for environmental sample cleanup. High efficiency
100A material packed in 19mm x 30cm and 19mm x 15cm columns, as well as the
two in series, has performed the cleanup effectively while operating with less solvent
and greater throughput than the traditional column. Injecting 5ml samples
containing over 310mg of material did not overload the two column set.
INTRODUCTION
GPC clean up of environmental samples is now mandatory for preparing
semivolatile and pesticide Superfund samples according to the EPA Contract
Laboratory Program Statement of Work. Laboratories which participate in the
EPA Contract Laboratory Program and those which follow CLP protocols, doing
"CLP-like" work, must use GPC.
Another aspect of environmental testing is concern about the quantities of solvents
which are routinely consumed in environmental cleanup and testing procedures.
Recently a major refinement of the traditional GPC cleanup procedure produced
over 70% reduction in solvent usage and 69% reduction in sample processing time1,
Table 1. This was accomplished by substituting a Waters Ultrastyragel high
efficiency, high resolution GPC column for the low resolution column which has
been used in the method for almost two decades.
This high resolution column was chosen based on general properties and
requirements of the method. Since the work began, a revised set of calibration
requirements were issued for the GPC cleanup method, prompting a formal column
optimization study. This study was concerned with two main variables - grade of
column packing and column dimensions.
11-232
-------
REVISED CALIBRATION REQUIREMENTS
The original calibration mix contained corn oil, bis(2-ethylhexyl)phthalate and
pentachlorophenol (PCP). The basic requirement of the column was that it provide
85% or better resolution between the corn oil and the phthalate. The com oil
represents the low volatility, high molecular weight material which is removed from
the sample before analysis. The work reported by Bumgarner with the high
efficiency column met these requirements, Figure 1.
The revised calibration mix still contained corn oil and phthalate but Methoxychlor,
Perylene and Sulfur were added in place of PCP. Resolution of 85% or better
among each pair of compounds was also specified.
PACKING MATERIAL OPTIMIZATION
The initial high resolution columns studied contained a high efficiency packing
material with an exclusion rating of 500A. To determine whether the 500A or the
corresponding 100A material is better suited for GPC cleanup work, the relative
retention profiles of each were compared to the Bio-BeadsK packing traditionally
used in this method.
Relative retention was defined as the ratio of the retention volume of a calibration
compound to the retention volume of sulfur. As indicated in Figure 2, the 100A
packing behaved essentially identically to the Bio-Beads. This packing grade was
used during subsequent optimization of column dimensions.
COLUMN OPTIMIZATION - RESOLUTION
A 19mm ID x 30cm long column was used originally. It easily met the initial
calibration requirements and the revised requirements for all calibration pairs
except phthalate and Methoxychlor. To achieve the required level of resolution for
all pairs, the column length was increased by using a 19mm ID x 15cm long segment
in series with the original 19mm x 30cm column, Figure 3.
COLUMN OPTIMIZATION - MASS LOADING
As indicated in Figure 3, the configuration of a 15cm and 30cm column in series
more than satisfies the calibration requirements, with the phthalate and
Methoxychlor almost baseline resolved. The concentration of the calibration
markers has been increased by 2.5 times to provide the same total mass on column
as would have been loaded with a 5ml injection containing the concentrations listed
in EPA Method 3640A.
To further demonstrate the resolving power and loading capacity of this preferred
column set, a 2ml aliquot of the collected fraction was reinjected into the GPC
system, Figure 4. It is estimated that this sample contains about 8-10% of the
original mass of the collected peaks. The outstanding capacity of the column is
shown again by the similarity in resolution of this diluted fraction and the original
2ml injection.
11-233
-------
COLUMN OPTIMIZATION - VOLUME LOADING
To demonstrate volume loading capacity on the high efficiency column set, a 5ml
sample containing about 3l5mg was injected, Figure 5. The broadened peak shapes
are a consequence of the larger injection volume, resulting in more peak overlap.
Each peak was collected as a separate fraction with the valley between peaks taken
as the cut points.
About 10% of the volume of each fraction was reinjected as a separate fraction,
Figure 6. Although the detector sensed what appeared to be considerable peak
overlap in the initial injection, rerunning the individual collected fractions shows
excellent resolution, well above the calibration requirements. In all of the loading
considerations, mass or volume, the actual capacity of the high efficiency column set
surpasses the requirements of method.
SPEED. LOADING. RESOLUTION
The availability of 15 and 30cm column sections which can be used alone or in series
provides maximum flexibility for cleanup of environmental samples. Used together
they provide maximum loading capacity and resolution. With lightly contaminated,
low concentration samples the 30cm length may be used alone for maximum
throughput and solvent economy. In either case, the high resolution of these
columns provides significant gains in operating effectiveness with corresponding
reductions in solvent usage versus the low resolution column traditionally used for
GPC cleanup.
SUMMARY
High resolution GPC columns have been shown to meet the resolution criteria of
EPA Method 3640A and the EPA Contract Laboratory Program Statement of Work
for Organics Analysis. They provide major savings in solvent use and processing
time relative to the low efficiency columns traditionally used in this work.
Maximum resolution and loading capacity for performing GPC cleanup of
environmental samples with these high resolution columns is achieved with a 19mm
ID x 30cm column in series with a 19mm ID x 15 cm column.
NOTE
Bio-Beads is a registered trademark of Bio-Rad Laboratories.
REFERENCE
1. Bumgarner Jr., J., International GPC Symposium Proceedings, Boston, MA, 1989,
Waters Chromatography Division, Millipore Corp., pp.787-793,1989.
H-234
-------
Table 1
GPC CLEANUP OF ENVIRONMENTAL SAMPLES
COMPARISON OF Low RESOLUTION COLUMNS vs 500A ULTRASTYRAGEL*
TYPICAL RUN TIMES (MINUTES)
DUMP COLLECT WASH TOTAL
ro Low RESOLUTION 30 36 15 81
<ซ ULTRASTYRAGEL 12.5 8.5 4 25
TYPICAL METHYLENE CHLORIDE VOLUMES (ML)
DUMP COLLECT WASH TOTAL
Low RESOLUTION 150 180 75 405
ULTRASTYRAGEL 56.25 38.25 18 112.5
^BUMGARNER JR, J.f OCTOBER 1989 GPC SYMPOSIUM
-------
Figure 1
ULTRASTYRAGEL COLUMN RUN ON WATERS SYSTEM
SOURCE: BUMGARNER GPC SYMPOSIUM PAPER, 10/89
GPC Calibration Standard
USEPA 2/88 SOW
22.5 minute run time
4.5 ml/minute flow rate
Pentachlorophenol
Bis (2-Ethylhexyl)
Phthalate
10 12.5
Minutes
11-236
-------
Figure 2
HIGH EFFICIENCY GPC CLEANUP OF ENVIRONMENTAL SAMPLES
COLUMN OPTIMIZATION
COMPARISON OF GPC PACKINGS BY RELATIVE RETENTION
ro
00
O 100 A MATERS
500 A MATERS
BIO-BEADS S-X3
t
9.9
.*
0.5
9.4
COMPARISON OF GPC PACKING MATERIALS
USING RELATIVE RETENTION TIMES TO
SULFUR AS A TOTAL VOLUME MARKER
COBN OIL
BISO.ETHYLHEXYL)
PH'l'HALATE
PERYLENE
SULFUR
-------
Figure 3
ro
u
CD
HIGH EFFICIENCY 6PC CLEANUP OF ENVIRONMENTAL SAMPLES
COLUMN CALIBRATION AT METHOD LOAD
COLUMN: 100A 19MM x (30cM + 15cn)
SAMPLE: 2000 UL
SOLVENT: METHVLENE CHLORIDE
FLOW RATE: SHL/MIN
DETECTION: UV 9 254NM, 1.5 AUFS
PEAK ID:
1 CORN OIL, 62.5MG/HL
2 BXS(2-ETHYLHEXYL)PHTHALATE, 2.5NG/HL
3 HETHOXYCHLOR, O.SMG/ML
4 PERYLENE, O.OSNG/ML
5 SULFUR, 0.2M6/HL
s
u
n
Ul
0.
o
to
COLLECT
-------
ro
to
co
HIGH EFFICIENCY GPC CLEANUP OF ENVIRONMENTAL SAMPLES
COLUMN CALIBRATION AT METHOD LOAD
ALIQUOT OF COLLECTED FRACTION REINJECTED
COLUMN: 1QOA WMM K (30cM + 15cw)
SAMPLE: 2000 UL
SOLVENT: METHYLENE CHLORIDE
FLOW RATE: SML/MIN
DETECTION: UV 9 254NH, 1.5 AUFS
PEAK ID:
2 BzS(2-ETHYLHEXYL)PHTHALATEr 2.5HG/HL
3 METHOXYCHLOR, O.BMG/ML
4 PERYLENE, O.OSMG/HL
B
lull
-------
Figure 5
HIGH EFFICIENCY GPC CLEANUP OF ENVIRONMENTAL SAMPLES
COLUMN OPTIMIZATION
SML TEST Mix INJECTION
COLUMN: 100A 19MM x (30cM + 15cM)
1900.000
.000
220.000
10
4ง
-------
Figure 6
HIGH EFFICIENCY GPC CLEANUP OF ENVIRONMENTAL SAMPLES
COLUMN OPTIMIZATION
REINJECTED FRACTIONS FROM SML TEST Mix INJECTION
COLUMN: 100A 19MM x (30cM + 15cM>
Oft fiftfi / \ Corn Oil
28.000-j / \ FrQdion
mV
-7.
OR nnn /1 Phthalate
28-ฐฐ9l / \ Fraction
mv
-7.
28.000-
mV
-7.000-
Rerylene
Fraction
28.000-
mV
-7.000-
4-Nitrophenol
Fraction
on nnn J / \ Sulfur
28-ฐฐ.ฐH A/ \ Fraction
mv
-7.000
10 1*5 S S 35 3ง 45 S
Minutes Press Resume
-------
75
EFFICIENT AQUEOUS EXTRACTION USING AN EMULSION PHASE CONTACTOR
Kevin P. Kellv. Ph.D., Manager, Applications; Loren C. Schrier,
Applications Chemist; Kenneth C. Kuo, Senior Research Chemist, ABC
Laboratories Inc., P. O. Box 1097, Columbia, Missouri 65205
ABSTRACT
The Emulsion Phase Contactor (EPC) promotes the efficient liquid/liquid
extraction of aqueous samples with organic solvents. Use of transient
high intensity electric fields in the liquid/liquid extraction process
produces increased interfacial surface area, resulting in more efficient
mass transfer. Transient fields also induce droplet motion (distortion
and translation) and droplet coalescence, which promote phase transfer and
phase separation, respectively.
In the EPC technique, aqueous sample is introduced into a cell, where a
voltage is applied to disperse the aqueous phase into the bulk (organic)
phase. Mass transfer is facilitated by formation of small (micron-sized)
droplets. Contact time of the droplets is increased through the action of
additional charged plates, which also aid in the eventual aggregation of
dispersed droplets.
EPC data are presented which show excellent methylene chloride extraction
recoveries for many priority pollutants, such as would be analyzed using
EPA SW-846, Method 8270 (semivolatile organics by GC/MS), or the EPA CLP
(Contract Lab Program) Statement of Work (SOW).
The EPC extraction method is suitable for the automation of analytical
laboratory sample preparation, and it replaces the more labor intensive
extraction methods, such as separatory funnel extraction, or classical
liquid/liquid extraction with boiling methylene chloride.
Instrumentation to accomplish the goal of an automated EPC extraction is
currently under development at ABC Laboratories. Analytical applications
of EPC derive from technology transfer under U.S. Patent No. 4,767,515,
"Surface Area Generation and Droplet Size Control in Solvent Extraction
Systems Utilizing High Intensity Electric Fields", issued August 30, 1988,
which was developed by Scott and Wham at Oak Ridge National Laboratory.
INTRODUCTION
Traditional methods for the liquid/liquid extraction of most environmental
samples are both labor and solvent intensive, and have high potential for
exposure of laboratory workers to solvents and other hazardous substances.
Recent alternatives, such as solid-phase extraction (SPE) cartridges or
discs, may not cope well with difficult aqueous matrices. Improvement of
ordinary liquid/liquid extraction techniques by automating fluid transfers
11-242
-------
and using electrically enhanced mass transfer' provides opportunities for
superior analyte recoveries and more consistent analytical results, while
reducing solvent (and energy) consumption and decreasing the opportunities
for exposure of personnel to hazards.
Efficient mass transfer during liquid/liquid extraction depends directly
upon the availability of surface area for mass transport. Since organic
solvents tend to be essentially nonconductive, charged metal electrodes
mounted within a solvent-containing region can be used to polarize water
droplets, which leads to droplet shape distortion, rotation, translation,
and breakage (Figure 1). This greatly increases phase transfer kinetics
for extraction of organic analytes from the aqueous phase.
A device employing this principle for extraction of aqueous samples is
diagrammed in Figure 2. Aqueous sample is pumped into the bottom of the
solvent-filled chamber, where an electric field causes droplet disruption.
A second electric field, situated above the first, induces further droplet
motion and also aids coagulation of daughter droplets formed by the action
of the first field. Accumulating volume of aggregate aqueous phase (above
the two electric fields) overflows to another container. Thus the organic
solvent is in equilibrium with only a small portion of the aqueous sample
at any time. This contributes to excellent extraction efficiencies.
Efficiency of the EPC is demonstrated by extraction of various
environmental pollutants with methylene chloride, followed by analysis of
the extract using GC/FID or LC/UV detection.
RESULTS AND DISCUSSION
One liter of simulated field water2 was spiked with 2.50 mL of a methanol
solution that contained 1000 pg/mL each of priority pollutant molecules
(analytes were divided into two groups to facilitate chromatographic
analysis). The EPC extraction cell was charged with methylene chloride
(approximately 400 mL), and plate voltages were set at 15 kV (lower pair)
and 10 kV (upper pair). Aqueous sample was introduced into the EPC at a
rate of 22 mL/min. After all the sample had traversed the EPC extraction
chamber, the methylene chloride layer was drained, and one half was dried
with sodium sulfate, then concentrated to 10 mL (to minimize evaporative
losses) using a steam bath with the Kuderna-Danish apparatus.
Chromatographic data (FID) were obtained using an HP 5890 Series II
temperature programmed gas chromatograph (Hewlett-Packard) plus 3396-A
integrator, with one of the following two columns: a 30 meter x 0.25 mm
capillary (Supelco DB-5, 0.25 jim film thickness); or a 5 meter x 0.53 mm
JScott, Timothy C.; Wham, Robert M.; Ind. Enq. Chem. Res. 1989 28, 94.
Prepared by adding 24 mg KC1, 608 mg MgCl2 hexahydrate, 344 mg CaCl2
dihydrate, and 404 mg NaCl to one gallon of reagent water.
11-243
-------
Figure 1. Electric Fields in Extraction
DROP FORMATION
DROP OSCILLATIONS
DROP BREAKUP
O o
DROP-DROP
INTERACTIONS
& COALESCENCE
o o
oo
11-244
-------
FIGURE 2. EPC EXTRACTION CELL
WATER OUT
METHYLENE CHLORIDE IN
z. /WATER
PHASE BOUNDARY
10 KV
ELECTRIC FIELD
S.S. ELECTRODE
15 KV
ELECTRIC FIELD
TEFLON BODY
WATER IN
METHYLENE CHLORIDE OUT
11-245
-------
capillary (HP-1, 2.65 pirn film thickness). The recovery data (Table 1)
were developed by comparison of peak heights to calibrations with external
standards. Recoveries were good to excellent, with excellent precision.
One of the two analyte sets was also processed using a lower spiking
concentration (250 (iL of solution, Table 2) to verify adequate recovery at
trace levels. The entire extract from the low level runs was concentrated
to 5 mL for analysis.
Calculated recoveries used the average of duplicate extract injections
from each of the four extractions. Average recoveries for ten analytes,
which represented several classes of chemical pollutants, ranged from 82%
to 109%. The largest standard deviation measured was 7.1%, indicating
that in addition to good recoveries with faster turnaround time (less than
1 hour per extraction for a 1 Liter sample), EPC methodology furnishes an
enhancement in precision, in comparison with more operator dependent
techniques.
Influence of pH on the recovery of acidic analytes was assessed by spiking
water with 1.00 mL of a solution containing 1000 ^g/mL each of phenol, 4-
nitrophenol (4-NP), and 2,4-dinitrophenol (2,4-DNP) dissolved in methanol.
The pH of the sample was adjusted using 12 M HC1. The EPC extraction was
carried out at each pH, then HPLC analysis of the extracted water, with UV
detection of analytes, was used to measure the concentration of phenols
recovered in the extract and remaining in the water.
Data from those extractions at different pHs (Table 3) indicate that the
technique is efficient enough to produce reasonably good recoveries of
acidic compounds (pK, values for phenol, 4-NP, and 2,4-DNP are 9.89, 7.15,
and 3.96, respectively) with no pH adjustment of samples. Thus 2,4-DNP,
although a stronger acid than acetic acid (pK, 4.75) was recovered in 53%
yield from an unadjusted (pH 5.9) sample.
EPC development efforts are now focusing on further reduction of solvent
usage, design of particulate-tolerant extraction devices, and continued
improvements in sample throughput and analyte recoveries.
SUMMARY
The Emulsion Phase Contactor (EPC) is a very efficient extraction device
that electrically enhances mass transfer during liquid/liquid extraction.
It provides good recoveries for extraction of a wide range of semivolatile
analytes from water samples using methylene chloride, plus greater degree
of precision (reproducibility) than can be obtained using less automated
techniques. Extraction efficiency is so high that acidic analytes can be
recovered without pH adjustment of water samples. Additional advantages
are savings in labor and reduced exposure of lab personnel to hazardous
substances. Automated EPC equipment is under development for application
to SW-846, CLP, and other extraction methods.
1-246
-------
Table 1. Recovery Data (% by GC/FID) for Analytes at 2500
Analyte
bis ( 2-Chloroethyl ) ether
Nitrobenzene
bis ( 2 -Chloroethoxy) methane
Naphthalene
Acenaphthene
2-Fluorophenol
Aniline
1 , 4-Dichlorobenzene
Hexachloroethane
2-Fluorobiphenyl
Replicates
1
91
98
101
97
98
91
106
85
82
83
2
86
94
94
90
89
91
110
86
80
83
3
94
100
96
93
86
91
110
87
90
86
1
94
106
101
96
101
90
109
88
88
85
Statistics
X
91
100
98
94
94
91
109
86
85
84
a
3.8
5.0
3.6
3.2
7.1
0.5
1.9
1.3
4.8
1.5
11-247
-------
Table 2. Recoveries for Analytes at 250
Analyte
bis ( 2 -Chloroethy 1 ) ether
Nitrobenzene
bis < 2-Chloroethoxy ) methane
Naphthalene
Acenaphthene
i
100
105
101
98
88
2
101
103
104
99
105
Table 3. Influence of pH on Extraction of Phenolics
Compound
Phenol
4-Nitrophenol
2 , 4-Dinitrophenol
Recovery (%) at Indicated pH
pH = 5.9
63
59
53
pH = 3.8
80
68
75
pH = 1.6
1813
58
76
*High result due to chromatographic interference
1-248
-------
SFE PRACTICAL APPLICATIONS FOR ENVIRONMENTAL AND
INDUSTRIAL SAMPLES
.T.p. MHjR. JOE TEHRANI
Isco, Inc., 4700 Superior Street, Lincoln, NE,
68504
EADLK. MEQKV
HWS Technologies, Inc., 825 J Street, Lincoln, NE,
68501
Supercritical fluid extraction (SFE) is a technique
that has the potential to revolutionize
conventional methods of sample extraction and
analysis. When 003 is used, SFE is an exceedingly
quick, inexpensive, and environmentally safe method
of sample preparation for GC, HPLC, UV-VIS, and
TLC. SFE can also be used for percent extractable
determinations in foods and polymers.
Liquid extractions of analytes from complex
matrices often require labor intensive and time
consuming methods such as Soxhlet extraction. The
matrices must be extracted with large volumes of
environmentally hazardous solvents which must be
evaporated to the atmosphere or otherwise disposed
of.
This presentation will discuss practical SFE
applications for a variety of environmental and
industrial samples such as soils, polymers, and
foods. SFE extraction data will be presented and
comparisons will be drawn between SFE and
conventional methods of extraction.
1-249
-------
INORGANICS
-------
Microwave Sample Preparation Methods For Environmental
Analysis
H. M. Kingston, F. A. Settle*, M. A. Pleva**, Lois Jassie***, P. Walter, Jim
Petersen, and Bill Buote****, National Institute of Standards and Technology,
Center for Analytical Chemistry, Inorganic Analytical Research Division,
Gaithersburg, MD 20899.
*Dept. of Chemistry, Virginia Military Institute, Lexington, Virginia 24450
Dept. of Chemistry, Washington and Lee University, Lexington, VA 24450
Research Associate sponsored by CEM Corporation
Zymark Corporation
**
***
****
Microwave sample preparation is gaining a wide degree of acceptance and is
being applied in EPA methods for elemental analysis. The nature of standard
procedures requires that they be readily transferable and reproducible between
laboratories^. Because microwave sample preparation procedures are
quantifiable their reproducibility provides an opportunity to improve data
quality. The control and standardization of microwave methods is a matter
that must be examined. Calibration of laboratory microwave equipment is
required to transfer the methods accurately and precisely. Calibration
methods have been evaluated and the error that can be expected in
transferring these procedures has been determined.
Microwave sample preparation methods provide a platform to produce
procedures that can be used generally for sample preparation in many
environmental elemental analyses. Currently, both the RCRA and CERCLA
programs share two microwave methods, developed cooperatively, applicable
for soils, sediments, sludges, oils, and waters. These methods demonstrate
robust microwave procedures that improve precision and are more efficient
than many classical methods. These attributes arise from the direct control of
energy transfer, and the mechanisms involved in that transfer. Reaction
temperatures and their profile control the mineral acid reactions that are
necessary to release the elements for analysis. The mechanisms of the energy
transfer and the relationships that control these reactions will be discussed
relevant to the two current EPA methods.
In addition to manual methods, an automated microwave decomposition
system has been developed using a modular design^. Three separate portions
of the automated microwave sample preparation system will be described as
well as a prototype quasi expert system that has been developed to assist in
standardizing procedures for microwave dissolution. A file structure has
been devised to transfer these procedures from system to system and provides
information adaptable to the level of automation in different instrument
11-253
-------
configurations. Once a procedure has proven successful, it is stored for future
reference and can be sent to other systems where it will automatically
reproduce the method. Thus, the system has the ability to "characterize" and
"transfer" procedures. A quality control sample log-in system was developed
that guides the analyst through data entry, weighing and barcoding of the
sample, and creates a sample data file. A computer-controlled microwave
unit has been developed for use with this system. The entire system is
robotidy integrated and is being tested as the first component of a fully
integrated inorganic analysis system. Microwave sample preparation stations
following this basic design have been constructed for EPA EMSL-LV and EPA
Region 10 using commercial versions of the research system. EPA and NIST
are coordinating the testing and evaluation of the systems.
Reference
1. Binstock, David A., Grohse, Peter M., Gaskill, Alvia Jr., Kingston, H. M., and
Jassie, L. B., "Development and Validation of a Method for Determining Elements
in Solid Waste Utilizing Microwave Digestion", JOAC, 74,2,1991.
2. Walter, P., Kingston, H. M., Settle, R A., Pleva, M. A., Buote, W., and
Chrosto, J., "Automated Intelligent Control of Microwave Sample
Preparation," in Advances in Laboratory Automation and Robotics 1990, eds.
Stramaitis and Hawk, vol. 7, Zymark Corporation, 1991.
11-254
-------
70 COMPARISON OF PROCEDURES FOR TCLP EXTRACT DIGESTION;
/O CONVENTIONAL VS. MICROWAVE
V. L. Verma, PhD, Supervisor, and T. M. McKee, Director,
Browning-Ferris Industries Houston Laboratory, 5630 Guhn
Road, Houston, Texas 77040
ABSTRACT
This paper describes a study that compares and evaluates
metal data resulting from the analyses of replicate Toxicity
Characteristic Leaching Procedure (TCLP) extracts subjected
to both the metal acid digestions methods specified in
SW-846 and microwave digestion procedures. The TCLP ex-
tracts analyzed by Inductively Coupled Argon Plasma (ICP)
Spectroscopy for the metals arsenic, barium, cadmium,
chromium, lead, selenium and silver were prepared by a CEM
Model MDS 8 ID closed vessel microwave digestion and EPA
SW-846 Methods 3010 and 3020.
INTRODUCTION
Method 3010, "Acid Digestion of Aqueous Samples and Extracts
for Total Metals for Analysis by ICP Spectroscopy" and
Method 3020, "Acid Digestion of Aqueous Samples and Extracts
for Total Metals for Analysis by Furnace Atomic Absorption
Spectroscopy" found in Test Methods for Evaluating Solid
Waste. Physical/Chemical Methods. November 1986, Third
Edition, USEPA, SW-846 are commonly applied techniques to
digest metals from aqueous samples in an open vessel.
As evidenced by the numerous studies reported in the litera-
ture, closed vessel microwave acid digestion is receiving
considerable attention as a state-of-the-art metal digestion
technique. Microwave heating was first reported to speed
digestion of samples by acids fifteen years ago(!). The
technique has been used in a variety of sample preparations
since then (!~12) and is rapidly gaining recognition as a
useful tool in analytical chemistry (13). Systems designed
specifically for laboratory microwave digestion are commer-
cially available. These systems are designed to overcome
deficiencies identified by researchers (4,5,9,11,12) wno
performed their initial work with microwave ovens manufac-
tured for domestic use. Recently, advanced techniques have
become available for doing acid digestion of metals in
closed TFE vessels by microwave heating (14) .
Browning-Ferris Industries Laboratory in Houston, Texas
analyzes 150 to 200 Toxicity Characteristic Leaching Proce-
dure (TCLP) metal samples per month with a goal of from two
to ten days turnaround time, including the long manual
11-255
-------
digestion needed for samples by the traditional hot plate
techniques prescribed in SW-846. In order to reduce costs,
turnaround time, and to improve the sample handling tech-
nique while maintaining QA/QC and laboratory safety, it was
decided to investigate the microwave digestion technique.
To accomplish this task, three different TCLP waste extracts
of both extraction fluid number one and extraction fluid
number 2 were analyzed for metals. The waste descriptions
were as follows:
- Soil contaminated with gasoline, diesel and heating
oil,
Soil contaminated with hydraulic and diesel oil,
- Filter press sludge,
- Digested domestic sewage sludge,
- Reactor rake-out residue (magnesium chloride
production),
Sand/urea
All of the Toxicity Characteristic (TC) metals, arsenic,
barium, cadmium, chromium, lead, selenium and silver, were
at concentrations below their appropriate detection levels.
These six extracts were each divided into three samples and
spiked with different concentrations each of the TC metals,
except for silver. Each of the samples was then further
divided into two equal portions. One portion was split into
four 50 milliliter (mL) aliquot and microwave digestion was
performed on each aliquot in a closed TFE vessel with 3 mL
of concentrated nitric acid and 2 mL of concentrated hydro-
chloric acid for 40 minutes at 90% power (515 Watts) . The
other portions of the spiked extracts were also subdivided
into four aliquots, and each subjected to the hot plate open
vessel acid digestion procedure as prescribed by method
3010.
The same procedure was repeated for the TCLP extracts
(extraction fluid number one and extraction fluid number
two) spiked with silver at three different concentrations.
Four 50 mL aliquots of each were microwave digested in
closed vessels with 5 mL of nitric acid. Four portions each
of the same spiked samples were also digested by the conven-
tional hot plate acid digestion procedure, Method 3020.
Method 3020, utilizing only nitric acid, is used by our lab
for silver digestion to avoid silver chloride precipitation.
All samples were analyzed by ICP spectroscopy.
Comparison of the data reveals the accuracy, applicability,
and performance efficiency of each technique.
11-256
-------
SUMMARY
Results of the study suggest that the Methods 3010 and 3020
can effectively be substituted by the microwave acid diges-
tion procedures utilized.
DISCUSSION
Apparatus
Microwave Digestion System, CEM Model MDS-81D
Closed TFE digestion vessels and turntable, CEM Part
No.
600050
Capping Station, CEM Part No. 920030
Hot plates
Griffin beakers
Watch glasses, ribbed
Reagents
Hydrochloric acid, concentrated, spectrograde
Nitric acid, concentrated, spectrograde
Water, deionized
Methodology - Six different Toxicity Characteristic (TC)
samples, three utilizing extraction fluid number one and
three with extraction fluid number two, were spike with the
following three concentrations of the metals arsenic,
barium, cadmium, chromium, lead, selenium, and silver:
Concentration level 1 - Maximum regulatory limit,
except for barium which was spiked at 5.00 ppm.
Concentration level 2 - Mid-range of the maximum
regulatory limit, except for barium which was spike at 2.50
ppm.
Concentration level 3 - Five times the respective
detection levels of each metal.
The spiked solutions were then divided into two equal por-
tions. To determine accuracy, each portion was subdivided
into eight aliquots and four were digested by the microwave
method, and four by the conventional hot plate methods.
All samples were analyzed by ICP Spectrometry under SW-846
Method 6010 at the following wave lengths:
arsenic 189 lead 220
selenium 196 barium 233
chromium 205 silver 328
cadmium 214
Microwave digestion method [3010X] -
o Transfer a 50 mL aliquot of a well mixed sample to a TFE
digestion vessel. Add 3 mL of concentrated and 2 mL of
concentrated hydrochloric acid. Place the safety pressure
relief valve on the vessels and then cap to 12 ft.lbs.
torque using the capping station.
1-257
-------
o Weigh the vessel and record its weight in grains. Place it
in the MDS-81D turntable and attach the venting tube.
o Repeat the above step until the turntable contains 12
vessels. It is critical that the total volume of the
solution equals 660 mL during digestion and that each vessel
contains an equal volume of acid. This is necessary to
ensure uniform heating of all vessel solutions.
o Turn the microwave unit exhaust on to the maximum fan
speed. Activate the turntable so that it is rotating
continuously.
o Program the instrument time for 40 minutes and power at
90% (515 watts). Depress the start key and heat the sample
mixture for the programmed time.
o At the end of the digestion period, remove the turntable
from the microwave unit and allow the sample solutions to
cool to room temperature. Shake the vessels to mix the
sample solutions. Detach each venting tube and remove the
vessels from the turntable.
o Weigh and record the weight of the cooled vessel after
digestion. If there is a weight loss greater than 0.5 grams
from that recorded prior to digestion, add DI water equal to
the weight lost. If there is a significant weight loss
(e.g., two to three grams), one should discard the sample,
and repeat the digestion procedure.
o Recap the vessel using the capping station and shake the
vessel to mix the sample solution.
o Open the vessels and filter the samples to remove any
insoluble materials if necessary. Do not rinse or dilute
the digested sample.
Microwave digestion method [3020X] -
This procedure is identical to the digestion method [3010X]
described above, except that instead of adding 3 mL of
concentrated nitric acid and 2 mL of concentrated hydrochlo-
ric acid to the sample to be digested, add 5 mL of concen-
trated nitric acid.
The hot plate digestion methods utilized in this study are
those described in SW-846, method 3010 and 3020. Method
3020 is utilized for samples to be analyzed by ICP for
silver.
RESULTS
Table I contains the results of the study. The data is
divided into two parts to distinguish between the two
different TCLP extraction fluids examined. The type of
digestion employed is identified for each extract media at
the top of the table. The left margin lists the metals and
their spike concentrations. The table also lists the
concentrations of four replicates and percent recovery.
11-258
-------
Table II illustrates the average concentration of the four
replicates, and shows the relative percent deviation of the
two values obtained by microwave and hot plate digestion
methods.
CONCLUSION
The intent of the study was to evaluate the data resulting
from the metal analyses of the two TCLP extract solutions
digested by both microwave and the conventional hot plate
methods.
The data reveal that percent recoveries obtained from micro-
wave digested samples and hot plate acid digested samples
are approximately equivalent. In some instances, the lead
and barium recoveries are low, probably due to matrix
interferences, but overall, the recoveries obtained by means
of the two methods are quite comparable.
The relative percent deviation (Table II) shows that hot
plate 3010, 3020 methods can be easily substituted by
microwave acid digestion method 3010X and 3020X.
The advantages of microwave digestion procedures over those
of conventional hot plated methods are: i) it is a rapid and
safe way of preparing samples for ICP analysis, ii) the
acids do not evaporate from the closed container causing
elevated concentrations of trace acid impurities, iii) the
digestion acids apparently do not decompose under microwave
conditions, iv) there are no acid fumes, v) volatile ele-
ments are retained in the sample solution, vi) the method
requires less monitoring, and finally, vii) there is less
potential for external sample contamination. The only
limitation of this method is the time-consuming assembling
and cleaning of the digestion vessels.
ACKNOWLEDGEMENTS
The authors would like to express their appreciation to Ms.
Linda DeLeon, Ms. Nanette Skal, Ms. Jacqueline Palomino and
Mr. Charles Jui for their interest and valuable assistance
in the study and the preparation of this paper.
11-259
-------
TABLE I
COMPARISON OF PERCENT RECOVERY
MICROWAVE VS. HOT PLATE DIGESTION
EXTRACTION FLUID # 1
Arsenic at 5.000 ppm
EXTRACTION FLUID # 2
MicroW % Rec. Hot Plate
4.584 92 4.313
4.598 92 4.466
4.653 93 4.284
4.640 93 4.413
Arsenic at 2.500 ppm
MicroW % Rec. Hot Plate
2.065 83 2.257
2.066 83 2.101
2.171 87 2.018
2.038 82 2.121
Arsenic at 0.500 ppm
MicroW % Rec. Hot Plate
0.417 83 0.409
0.405 81 0.420
0.420 84 0.415
0.444 89 0.457
Selenium at 1.000 ppm
MicroW % Rec. Hot Plate
0.948 95 0.891
1.107 702 0.856
0.906 91 0.904
0.991 99 0.897
Selenium at 0.500 ppm
MicroW % Rec. Hot Plate
0.521 104 0.509
0.475 95 0.477
0.509 102 0.477
0.490 98 0.497
Selenium at 0.500 ppm
MicroW % Rec. Hot Plate
0.492 98 0.404
0.450 90 0.424
0.444 89 0.430
0.459 92 0.466
Lead at 5.000 ppm
MicroW % Rec. Hot Plate
4.767 95 4.636
4.773 95 4.621
4.703 94 4.524
4.709 94 4.627
% Rec.
86
89
86
88
% Rec.
90
84
81
85
% Rec.
82
84
83
91
% Rec.
89
86
90
90
% Rec.
102
95
95
99
% Rec.
81
85
86
93
% Rec.
93
92
90
93
MicroW
4.484
4.301
4.319
4.390
MicroW
2.509
2.456
2.223
2.432
MicroW
0.417
0.405
0.420
0.509
MicroW
1.032
1.070
1.066
1.086
MicroW
0.551
0.541
0.541
0.517
MicroW
0.547
0.522
0.528
0.549
MicroW
3.958
4.022
4.080
4.126
% Rec.
90
86
86
88
% Rec.
100
98
89
97
% Rec.
83
81
84
102
% Rec.
103
107
107
1089
% Rec.
110
108
108
103
% Rec.
109
104
106
110
% Rec.
79
80
82
83
Hot Plate
4.214
4.345
4.309
4.413
Hot Plate
2.255
2.135
2.421
2.181
Hot Plate
0.409
0.420
0.504
0.501
Hot Plate
0.906
0.910
0.862
0.897
Hot Plate
0.453
0.467
0.467
0.493
Hot Plate
0.530
0.476
0.554
0.551
Hot Plate
4.000
3.915
3.937
3.902
% Rec
84
87
86
88
% Rec
90
85
97
87
% Rec
82
84
100
100
% Rec
91
91
86
90
% Rec
91
93
93
99
% Rec
106
95
111
110
% Rec
80
78
79
78
11-260
-------
TABLE I (CONTINUED)
COMPARISON OF PERCENT RECOVERY
MICROWAVE VS. HOT PLATE DIGESTION
EXTRACTION FLUID #
Lead at 2.500 ppm
EXTRACTION FLUID # 2
MicroW % Rec.
2.277 92
2.255 90
2.278 91
2.189 88
Lead at 0.500
MicroW % Rec.
0.487 97
0.508 102
0.502 100
0.515 103
Chromium at 5.
MicroW % Rec.
4.874 97
4.844 97
4.862 97
4.844 97
Chromium at 2.
MicroW % Rec.
2.401 96
2.394 96
2.417 97
2.354 94
Chromium at 0.
MicroW % Rec.
0.283 113
0.264 106
0.257 103
0.237 95
Hot Plate
2.273
2.222
2.212
2.205
ppm
Hot Plate
0.506
0.473
0.451
0.477
000 ppm
Hot Plate
4.438
4.438
4.476
4.325
500 ppm
Hot Plate
2.210
2.170
2.187
2.240
250 ppm
Hot Plate
0.280
0.261
0.277
0.275
% Rec.
91
89
88
88
% Rec.
101
95
90
95
% Rec.
89
89
90
87
% Rec.
88
87
87
90
% Rec.
112
104
111
110
Barium at 5.000 ppm
MicroW % Rec.
4.618 92
4.466 89
4.519 90
4.560 91
Hot Plate
4.573
4.642
4.488
4.627
% Rec.
91
92
90
92
Barium at 2.500 ppm
MicroW % Rec.
2.375 95
2.355 94
2.397 96
2.261 90
Hot Plate
2.346
2.391
2.379
2.345
% Rec.
94
96
95
94
MicroW
1.914
1.872
1.918
1.939
MicroW
0.487
0.508
0.501
0.510
MicroW
4.488
4.600
4.508
5.035
MicroW
2.262
2.463
2.490
2.476
MicroW
0.283
0.264
0.257
0.238
MicroW
4.412
4.480
4.473
4.464
MicroW
2.356
2.328
2.365
2.360
% Rec.
77
75
77
78
% Rec.
97
102
100
102
% Rec.
90
92
90
100
% Rec.
90
98
99
99
% Rec.
113
1.06
103
95
% Rec.
88
90
89
89
% Rec.
94
93
95
94
Hot Plate
1.964
2.067
1.846
1.946
Hot Plate
0.506
0.473
0.451
0.477
Hot Plate
4.924
5.085
4.427
4.215
Hot Plate
2.100
2.228
2.289
2.218
Hot Plate
0.261
0.277
0.275
0.280
Hot Plate
4.287
4.403
4.296
4.295
Hot Plate
2.246
2.253
2.371
2.186
% Rec.
79
83
74
78
% Rec.
101
95
90
95
% Rec.
98
102
89
84
% Rec.
84
89
92
89
% Rec.
104
111
110
112
% Rec.
86
88
86
86
% Rec.
90
90
95
87
1-261
-------
TABLE I (CONTINUED)
COMPARISON OF PERCENT RECOVERY
MICROWAVE VS. HOT PLATE DIGESTION
EXTRACTION FLUID # 1
Barium at 0.100 ppm
Microw % Rec. Hot Plate
0.094 94 0.095
0.094 94 0.092
0.095 95 0.087
0.095 95 0.099
Cadmium at 1.000 ppm
Microw % Rec. Hot Plate
0.946 95 0.913
0.953 95 0.944
0.943 94 0.894
0.955 95 0.915
Cadmium at 0.500 ppm
MicroW % Rec. Hot Plate
0.443 89 0.429
0.431 0.86 0.419
0.439 88 0.416
0.425 85 0.415
Cadmium at 0.100 ppm
MicroW % Rec. Hot Plate
0.110 110 0.104
0.109 109 0.106
0.107 107 0.105
0.105 105 0.106
Silver at 5.000 ppm
MicroW % Rec. Hot Plate
EXTRACTION FLUID # 2
5.022
4.983
5.065
5.027
100
99.6
101
101
4.230
4.275
4.040
4.325
Silver at 2.500 ppm
MicroW % Rec. Hot Plate
2.431 97 2.200
2.420 97 2.130
2.430 97 2.340
2.542 102 2.136
Silver at 0.250 ppm
MicroW % Rec. Hot Plate
0.240 96 0.225
0.239 96 0.227
0.231 92 0.235
0.240 96 0.223
% Rec.
91
94
89
91
% Rec,
86
84
83
83
% Rec.
104
106
105
106
% Rec.
85
86
81
87
% Rec.
88
85
94
85
% Rec.
90
90
94
89
MicroW
0.899
0.909
0.913
0.913
MicroW
0.829
0.853
0.845
0.848
MicroW
0.375
0.399
0.393
0.406
MicroW
0.106
0.104
0.103
0.098
MicroW
5.022
5.088
5.049
5.093
MicroW
2.407
2.591
2.604
2.453
MicroW
0.249
0.243
0.251
0.244
% Rec.
90
91
91
91
% Rec.
83
85
85
85
% Rec.
75
80
79
81
% Rec.
106
104
103
98
% Rec.
100
102
101
102
% Rec.
96
104
104
98
% Rec.
100
97
100
98
Hot Plate
0.891
0.874
0.877
0.889
Hot Plate
0.828
0.841
0.831
0.828
Hot Plate
0.380
0.386
0.399
0.380
Hot Plate
0.101
0.098
0.103
0.100
Hot Plate
4.440
4.305
4.165
4.230
Hot Plate
2.367
2.476
2.564
2.314
Hot Plate
0.223
0.206
0.223
0.215
% Rec
89
87
88
89
% Rec
83
84
83
83
% Rec
76
77
80
76
% Rec
101
98
103
100
% Rec
89
86
83
85
% Rec
95
99
103
93
% Rec
92
82
89
86
11-262
-------
TABLE II
COMPARISON OF RELATIVE PERCENT DEVIATION
MICROWAVE VS. HOT PLATE DIGESTION
EXTRACTION FLUID # 1
Arsenic
MW H P MW H P
4.619 4.369 2.085 2.124
RPD 5.67, 1.90/.
Selenium
MW HP
0.966 0.887
RPD 8.5%
Lead
MW HP
4.738 4.602
RPD 2.9%
Chromium
MW HP
4.856 4.419
RPD 9.4%
Barium
MW HP
4.541 4.583
RPD 0.94%
Cadmium
MW HP
0.949 0.917
RPD 3.4%
MW HP
0.499 0.488
2.2%
MW HP
2.250 2.228
0.98%
MW HP
2.391 2.202
8.2%
MW HP
2.347 2.365
0.76%
MW HP
0.435 0.420
3.5%
Silver
MW H P MW H P
5.025 4.218 2.456 2.202
RPD 17.5% 10.9%
EXTRACTION FLUID # 2
MWHP MWHP MWHP MWHP
0.422 0.425 4.374 4.322 2.405 2.248 0.438 0.459
0.71% 1.2% 6.7% 4.7%
MWHP MWHP MWHP MWHP
0.461 0.431 1.064 0.894 0.521 0.466 0.536 0.528
6.7% 17.4% ll.l'/, 1.5%
MWHP MWHP MWHP MWHP
0.503 0.477 4.047 3.939 1.911 1.956 0.502 0.477
5.3% 2.7% 2.3% 5.1%
MWHP MWHP MWHP MWHP
0.260 0.273 4.659 4.663 2.423 2.210 0.261 0.273
4.88% 0.09% 9.1% 4.8%
MWHP MWHP MWHP MWHP
0.094 0.093 4.457 4.320 2.352 2.264 0.909 0.883
1.07% 3.1% 3.8% 2.9%
MWHP MWHP MWHP MWHP
0.108 0.105 0.844 0.830 0.393 0.386 0.103 0.101
2.8% 1.7% 1.8% 2.0%
MWHP MWHP MWHP MWHP
0.238 0.228 5.063 4.285 2.574 2.430 0.247 0.217
4.3% 16.6% 3.4% 12.9%
11-263
-------
REFERENCES;
1. Abu-Samra, A.; Morris, J..S.; Kortyohann, S. R., Anal.
Chero., 1975, 47, pp 1475-1477.
2. Barrett, P.; Davidowski, L. J., Jr.; Penaro, K. W.;
Copeland, T.R., Analy. Chem. 1978, 50, pp 1021-1023.
3. Nadkarni, R. A., Anal. Chem., 1984, 56, pp2233-2237.
4. White, R. T.; Douthit, G. E., J. Assoc. Off. Anal.
Chem., 1985, 68, pp 766-769.
5. Matthes, S. A.; Parrell, R. F., Mackie, A. J., Tech.
Prog. Rep., U.S. Bureau of Mines, 1983, p!20.
6. Fernando, L. A.; Heavner, W. D.; Cabrillei, C. C. ,
Anal. Chem., 1986, 58, p 551.
7. Fisher, L. B., Anal. Chem. 18986, 58, pp261-265.
8. Lamonte, P. J.; Fries, T. L.; Consul, J. J., Anal.
Chem., 1986, 58, pp 1881-1886.
9. Copeland, T. R., Work assignment for the Office of
Solid Waste, U. S. EPA, June 1986.
10. Westbrook, W. T.; Jefferson, R. J., J. Microwave Power,
1986, 21, p25.
11. Jassie, L. B.; Kingston, H. M., 1985 Pittsburg
Conference Abstracts, Paper 108 A.
12. Kingston, H. M.; Jssie, L. E., Anal. Chem., 1986, 58,
pp 2534-2541.
13. "Symposium on Microwave Techniques", Twenty-fith
Eastern Analytical Symposium, Oct. 1986, New York;
"Symposium on Microwave Techniques", Twenty-sixth
Eastern Analytical Symposium, Sept. 1987.
14. Reverz, R.; Hasty, E., Pittsburgh Conference and
Exposition on Analytical Chemistry and Applied
Spectroscopy, March 1987.
16 epapa2
11-264
-------
7Q Sample Decomposition in Closed Vessels with & Pressure Controlled
Microwave Oven.
F.flanholzer. Q. Knapp, P. Kettisch, A. Schalk
Department for Analytical Chemistry, Micro- and Radiochesitry,
Grass University of Tehcnology, Technikerstrale 4, Graz, Austria
wet-chemical sample decomposition in closed vessels is one of the
most efficient methods for trace element analysis. Temperatures of
at least 300*C are required, in order to guarantee the complete
decomposition of organic matter with nitric acid. For that purpose
the decomposition must be done under high pressure of up to 80 bar.
Thซ currently available high-pressure vessels for microwave
decomposition do not permit control of the microwave energy and
therefore unknown sample materials can cause the vessel to rupture at
excessive internal pressure. The high-pressure microwave
decomposition vessels we developed permit the control of microwave
energy by the internal pressure. Thus also unknown sample materials
can be decomposed quickly and without problems.
1-265
-------
80 STATE-OF-THE-ART OF MICROWAVE DIGESTION METHODS
FOR ENVIRONMENTAL ANALYSIS
Mark E. Tatro. SPECTRA Spectroscopy & Chromatography Specialists, Inc., P.O.
Box 352, Pompton Lakes, NJ 07442
ABSTRACT
The closed vessel microwave digestion methods approved by EPA for the preparation of
waters and soils for trace metal analysis require a two stage microwave power program
that is designed to achieve an initial target temperature of the digestion acid followed by
a slow rise to a final temperature. There are inherent problems with this method that
arise from the fact that the digestion acid temperatures are functions of vessel type,
microwave power, line voltage and the number of vessels. In-situ temperature
measurement of the acid during digestion requires the use of an expensive fiber-optic
probe device that is beyond the budget of most environmental laboratories. Therefore,
users will be "flying blind" when attempting to reach target temperatures. This paper
will present data that depicts in-situ temperature-time curves that demonstrate the effect
of vessel design and increased microwave power on target temperatures.
INTRODUCTION
The approved EPA CLP and the proposed EPA SW-846 closed vessel microwave
digestion procedures are based on very rigid formats regarding power-time
programming and the number and type of vessels used (1). The EPA methods for the
digestion of water samples requires the use of 5 vessels all containing 45 mis of water
sample and 5 mis cone, nitric acid with a two stage power program of 545 Watts for 10
minutes followed by 344 Watts for an additional 10 minutes. This program is designed
to allow the acidified samples to reach a target temperature of 160 ฑ 4 ฐC by the end of
the first 10 minutes and to allow for slow rise to 165 - 170 ฐC within the next 10
minutes. The EPA methods for the digestion of soil samples requires the use of either 2
vessels containing the samples and 10 mis cone, nitric acid with a one stage power
program of 344 Watts for 10 minutes or 6 vessels containing the samples and 10 mis
cone, nitric acid with a one stage power program of 574 Watts for 10 minutes. This
program is designed to allow the acidified samples to reach an inital target temperature of
175 ฐC in less than 5.5 minutes and remain between 170 -180 ฐC for the balance of the
10 minute time period. The above microwave programs are based on the use of 120 ml
single walled teflon vessels.
11-266
-------
If the user deviates from the above conditions, i.e. uses more than the prescribed number
of vessels, uses double walled instead of single walled vessels, uses a higher power
wattage than prescribed, then the methods as written will not work as they will not reach
the initial and final target temperatures in the prescribed time. The proposed SW-846
Methods 3015 and 3051 allow users to use higher wattage ovens to digest more than the
prescribed number of samples at one time and allow the users to use alternative vessel
designs. However, the methods allow these changes only if the user can document that
the temperature-time profiles remain unaltered. This requires the use of an expensive
fiber-optic temperature probe that requires experience and can be dangerous if not used
properly with pressurized vessels containing acid. Therefore, this author considers the
SW-846 allowed changes to be a moot point for most environmental laboratories.
The basic equation used to calculate microwave power absorbed by acid matrices (2) is
shown in Equation 1:
P(absorbed) = [(K)(Cp)(m)(dT)]/t
This author receives feedback from users that are under the impression that Equation 1
can be used to predict the final temperature reached in a specific time if they know the
number of vessels to be used. This, in fact, is not the case since this equation does not
take into account cooling effects. In fact, the power-time programs developed for the
EPA CLP and SW-846 methods had to be developed empirically using an expensive
fiber-optic temperature probe (3). Again, without such a probe, users cannot develop
their own microwave digestion methods to stay within the EPA required temperature-
time confines.
As yet, manufacturers of microwave digestion systems have not provided the information
needed to alter conditions while remaining within the temperature-time guidelines.
Therefore, this author conducted the following experiments to document the effect of
varying the types of vessels and the power on reaching the initial and final target
temperatures for the microwave digestion of water samples. This study used a Luxtron
Model 750 fiber-optic temperature probe for in-situ temperature monitoring and a Floyd
Model RMS-150 microwave digestion oven (4).
Figure l(a) depicts the temperature-time curve for the typical CLP and SW-846 type
water digestion microwave program. It uses 5 single-walled vessels, each containing 45
1-267
-------
mis deionized water and 5 mis cone, nitric acid, using the prescribed 2 stage program of
545 Watts for 10 minutes followed by 344 Watts for 10 minutes. The temperature
profile, as expected met the required target values.
Figure l(b) depicts the temperature-time curve developed empirically using 5 double
walled vessels, each filled with 45 mis deionized water and 5 mis cone, nitric acid. A 2
stage program of 480 Watts for 10 minutes followed by 234 Watts for 10 minutes met
the required target temperatures. The reduction in power to meet the temperature targets
was a result of the greater insulation afforded by the double walled vessels. Again, it is
stressed that this alteration to the EPA prescribed microwave power-time program would
have been impossible without the use of the fiber-optic temperature probe.
Figure l(c) depicts the temperature-time curve when 5 double walled vessels containing
45 mis deionized water and 5 mis cone, nitric acid, instead of 5 single walled vessels,
were used with the EPA power-time program of 545 Watts for 10 minutes followed by
344 Watts for 10 minutes. As shown, by using more insulated vessels, the temperatures
reached far exceed the EPA required target temperatures.
As stated previously, predicting target temperatures must be done empirically. This
author wanted to predict how many more doubled walled vessels could be used for water
digestions if a higher wattage (745 W) microwave oven were used. Figure 2(a) depicts
the temperature-time curve for 12 vessels all filled with 45 mis deionized water and 5
mis cone, nitric acid were heated at full power. As depicted, it required 16 minutes for
the samples to reach the initial target temperaure of 160 ฐC. From this curve, the number
of double walled vessels that can be used to reach 160 ฐC in 10 minutes using 745 Watts
of power can be predicted as follows:
Step 1: Using Equation 1, determine the actual power absorbed over the 16 minute
period. In this case, m = 624 gm [(12 x 45 ml water x 1 gm/ml) + (12 x 5 ml nitric acid x
1.4 gm/ml)]; Cp = 0.9297 (estimated from reference (2)); dT = 160 - 24.2 = 135.8 ฐC; t
= 16 minutes x 60 second/minute = 960 seconds; K = 4.184. The actual power absorbed
is therefore:
P(absorbed) = f4.184X0.9297)(624)(135.8> = 343 Watts
960
11-268
-------
Step 2: Using the empirically determined power absorbed of 343 Watts, determine how
many vessels containing 45 mis deionized water and 5 mis cone, nitric acid (52 grams
total mass/vessel) can be heated from 24 ฐC to 160 ฐC in 10 minutes:
(a) m = [(P)(t)]/[(K)(Cp)(dT)] = [(343)(600)]/[(4.184)(0.9297)(136)] = 389 grams
(b) (389 grams)/(52 grams/vessel) = 7.5 = 8 vessels
Figure 2(b) verifies that the prediction of 8 vessels is correct since the actual
temperature-time curve for 8 vessels containing 45 mis deionized water and 5 mis cone.
nitric acid using 574 Watts reaches 160 ฐC in 10 minutes.
SUMMARY
The interest by EPA in converting from hot plates to microwave ovens for the
preparation of samples for trace metal analysis is to be commended. However, the
confusion at the outset when SW-846 approves methods 3015 and 3051 is expected to be,
in this author's opinion, overwhelming. Without expensive in-situ temperature
monitoring probes and without documentation from the microwave manufacturers on
how to deviate from the rather rigid temperature-time profiles as prescribed by EPA,
users are expected to be confused. The announcement of in-situ temperature monitoring
capabilities built into the next generation of microwave ovens will go a long way to
reduce this confusion.
REFERENCES
1. Tatro, M.E., EPA approves closed vessel microwave digestion for CLP laboratories.
Spectroscopy, 5 (6), 17 (1990).
2. Kingston, H.M. and L.B. Jassie. Introduction to Microwave Sample Preparation.
Theory and Practice, Chapter 6. H.M. Kingston and L.B. Jassie, eds. American
Chemical Society, Washington, D.C. (1988).
3. H.M. Kingston, private communication (1990).
4. M.E. Tatro, From hot plates to microwaves. Environmental Lab, 3(1), 28 (1991).
1-269
-------
Rgure 1
210
200
Single- vs. Double-wall vessels
01234567891DH121314151617181920
Rgure 2.
Predicting Target Temperatures
180 ~
1 en -
isn -
13Q
4 9A -
-.120
gXUU -
i on -
30 H
20?
o^4 ^ซ
S* ^^
f ^m****^
/ ^-""""^
X -*"
^-**^
fs*^
4 S
2>
v
//
ฃ
/
r
0123456789101112131415 18
Minutes
2(a) + 2(b)
11-270
-------
g-j THE APPLICATION OF X-RAY FLUORESCENCE SPECTROSCOPY FOR RAPID
HAZARDOUS WASTE CLASSIFICATION AND SCREENING
Dr. Peter A. Pospisil. Manager Methods Development, Dr. Harold Van Kley,
Chemist Methods Development, Dr. Mark F. Marcus, Senior Director Analytical
Programs, Chandler Taylor, Chemist Analytical Lab, Dr. Nilesh Shah, Senior
Analytical Chemist Methods Development, Chemical Waste Management Inc.,
Technical Center, 150 W. 137th Street, Riverdale, IL 60627;
Dr. E. Scott Tucker, Director Chem-Nuclear Laboratory Services, 25 Woods Lake
Road, Greenville, SC 29607
ABSTRACT
X-ray fluorescence spectroscopy (XRF) with pattern recognition data interpretation
provides immediate elemental screening capabilities for the comparison of sales
sample metal composition with compositional data from loads received at hazardous
waste sites.
The proper disposal of hazardous waste through stabilization processes requires
information about the elemental composition of the waste stream. Based on the
stream's elemental composition, which is both waste code and generator specific,
processing decisions are made to select the most effective waste stabilization
procedure. Current procedures utilize SW-846 methods to generate elemental
compositional data. The analytical method turnaround time creates significant
process delays and costs for materials received at CWM sites. XRF can rapidly
provide information regarding RCRA elemental analyte composition in hazardous
waste streams, enabling processing decisions to be made on a comparatively realtime
basis and provide assurance of effective stabilization.
Advances in energy-dispersive (ED) XRF instrumentation with computer software
have greatly increased interest in the technology. Advantages for hazardous waste
analysis include minimal sample preparation, applicability to a broad range of liquid,
solid or semi-solid samples and simultaneous multi-element analysis over a broad
concentration range with no elemental carryover. The study generated elemental
pattern data for K061 and F006 wastes, produced by specific generators. Fourteen
elements were selected for the study: calcium, chromium, iron, nickel, copper, zinc,
arsenic, selenium, silver, cadmium, barium, mercury, thallium and lead. ICP
analyzed K061 wastes were used as quantitation standards.
1-271
-------
Univariate elemental analysis using a one standard deviation comparative criterion
demonstrates the effectiveness of XRF technology for distinguishing among waste
codes and generators. Multivariate analysis using the Mahalanobis distance
technique improves the comparison by utilizing specific elements or the entire X-ray
spectrum for recognition of XRF patterns. We conclude that ED-XRF is able to
provide data critical for the decision process at hazardous waste disposal sites, while
reducing the overall cost of operation.
INTRODUCTION
All wastes received at Chemical Waste Management (CWM), both for waste
disposal decisions and as received loads at our disposal sites are subjected to a
"fingerprint" analysis. The purpose of this rapid test series is to verify that the
material received for disposal matches the profile for that generator produced by the
sales samples. Screening tests are in place for nine parameters, and it would
improve the quality of the screening process and streamline disposal operations if
metals could be included in the fingerprint screening process. This study was
undertaken to determine if x-ray fluorescence spectroscopy (XRF) could provide a
reasonable and rapid metals-based fingerprint analysis to augment the current series.
An important factor in any fingerprint screen is that it be rapid and sensitive enough
to screen the hazardous parameter at the appropriate level of quantitation. XRF
is an excellent screening choice for this research, since samples require little or no
sample preparation and data can be generated within minutes.
PROJECT PURPOSE
The purpose of this project is to determine the applicability of XRF analysis to
fingerprint screening of sales and received samples. The project will proceed
through the generation of quantitative analytical data to determine the feasibility of
a univariate pattern recognition process in differentiating among waste codes and
generators. The work will continue using a multivariate process based on a
Mahalanobis distance technique, which will eliminate the need for the generation
of quantitative analysis.
TECHNICAL OVERVIEW
In XRF, electrons in the lowest energy orbitals near the nucleus of the atom are
energized by external radiation and escape from the atom. Electrons from higher
energy orbitals fill the empty orbital and the energy lost in dropping to a lower
11-272
-------
energy orbital is emitted as an x-ray. Because the emitted x-rays are always at a
lower energy than the activating radiation, the process is called "x-ray fluorescence."
Each element has characteristic electron orbitals of specific energy and therefore a
characteristic x-ray fluorescence pattern. Because inner orbital rather than valence
electrons are involved in this method, the chemical form of the element has little
or no influence. The method is applicable for qualitative and quantitative analysis
for chemical elements higher in the periodic chart than oxygen.
Although the technique of ED-XRF has been known for about 40 years, recent
advances in instrumentation and especially computer software have greatly increased
interest in the technique. XRF advantages include minimal sample preparation,
applicability to liquid, semi-solid or solid samples, simultaneous multi-element
analysis over a wide concentration range, no carryover to the succeeding sample,
rapid quantitation and potential to optimize the system for specific elements.
The XRF spectrometer records counts received in individual channels of a multi-
channel analyzer. Each channel counts a small range of energies so that a spectrum
of counts at specific energies is obtained. On the basis of known energy values for
individual elements, specific ranges are assigned to certain elements.
EXPERIMENTATION
Instrument Selection
The Kevex 770 XRF spectrometer was chosen as the most applicable instrument
based on its sensitivity, flexibility and speed of analysis. The instrument was
purchased with a DEC VAX 11-57 computer, TSX operating system and Toolbox
software.
Sample Preparation
Sample preparation for this technique is minimal. Dry powdered materials may be
placed directly in an XRF cup. Liquid or semisolid samples may be run as received
or be dried in an oven and then analyzed as a dried powder. Claylike damp
samples can be packed into the cup and tamped gently to remove the air spaces.
Instrument Calibration
As a starting point for this work the instrument was calibrated using a large set of
K061 wastes from a single generator. The set of K061 wastes were first analyzed
11-273
-------
for the following 14 elements using conventional ICP and AA methodology:
Calcium Chromium Iron
Nickel Copper Zinc
Silver Cadmium Barium
Arsenic Selenium Mercury
Thallium Lead
These data are presented in Table 1. The high values of thallium and arsenic were
found to arise from elemental interferences and were not considered in the pattern
recognition process. Additionally, values for calcium and iron were not included,
since their consistently high values overshadowed the metals of interest.
A critical consideration for this work is that the instrument was not calibrated to
produce quantitative data but consistent data among the matrices encountered.
Analytical Precision
A single K061 sample was run repeatedly under the same activation conditions to
show the precision of the method. The sample was run under the activation
conditions used for elements calcium, chromium, iron, nickel, copper, and zinc. The
sample was stirred every few runs to expose a different portion of the sample to the
analytical procedure. Results are shown in Table 2. Again close agreement for
each element is seen so that precision is acceptable for waste samples.
X-Ray Analysis
The following samples were analyzed by x-ray using the calibration curves produced
by the K061 materials.
Number of Sample Type Generator
Samples
24 K061 A
4 K061 B
7 F006 C
It was recognized that changing the sample type changed the matrix, which in turn
reduced the data's quantitative quality. Again, the project goals were not data
quantitation but pattern recognition, using elemental analysis as a guide. Varying
the calibration curves only adds an additional degree of freedom.
M-274
-------
The analytical data for Generators A, B and C are presented in Figures 1-4 a and
b, in a bar graph format. The "a" portion of the Figure presents the composition
in percent, while the "b" portion presents the ppm information. The brackets on
each bar in Figure 2 show the compositional variation of the 24 K061 wastes at one
standard deviation. In a univariate pattern recognition approach, if the elemental
concentrations for the sample fall between the brackets defined in Figure 2, there
is a 65% probability that the waste is a K061 from Generator A.
A visual comparison of the bar graphs in Figures 2 and 3 show that the overall
elemental pattern is reasonably similar for the K061 wastes from generators A and
B. But, since the lead, cadmium and barium concentrations fall outside of the
defined brackets, there is a high probability that the waste is not a K061 from
Generator A.
Comparing the elemental pattern data between Generators A and C, K061 and F006
wastes, the basic elemental pattern is extremely different. No lead appears in the
samples from Generator C, but there are significant amounts of barium and nickel.
XRF can differentiate among waste codes and generators based on the univariate
elemental pattern.
CONCLUSIONS
1. XRF coupled with univariate analysis can generate unique elemental data
patterns, within 15 minutes for specific waste codes and generators. Wastes
from an individual generator are quite characteristic.
2. The method is rapid enough to be applicable for the purpose of fingerprint
screening.
3. The data are also applicable to process related decisions which are a function
of elemental distribution.
11-275
-------
TABLE 1
ANALYTICAL DATA FOR 24 K061 WASTES FROM GENERATOR A
LEAD
CHROMIUM COPPER
CADMIUM
PPM
BARIUM
PPM
NICKEL
PPM
SILVER
PPM
N>
AVERAGE
STD DEV
RANGE
RSD68%
RSD 95%
RSD99%
1.40
1.47
1.26
1.33
1.65
1.92
1.70
1.22
1.22
1.33
1.25
1.40
1.15
1.36
1.43
1.31
0.84
0.77
1.32
1.13
0.78
1.01
0.89
0.67
1.242
0.298
1.541
0.944
24.032
48.064
72.096
0.15
0.16
0.12
0.14
0.18
0.17
0.17
0.16
0.14
0.14
0.16
0.14
0.18
0.18
0.18
0.18
0.14
0.15
0.16
0.17
0.16
0.14
0.14
0.14
0.156
0.017
0.173
0.139
10.901
21.802
32.702
0.19
0.18
0.16
0.18
0.19
0.19
0.20
0.19
0.15
0.13
0.15
0.14
0.13
0.18
0.16
0.13
0.13
0.17
0.14
0.14
0.15
0.20
0.15
0.13
0.161
0.024
0.185
0.137
15.115
30.230
45.345
152
131
148
129
161
123
221
201
157
239
227
224
219
223
226
233
190
148
223
257
175
208
195
133
189.292
40.345
229.637
148.947
21.314
42.627
63.941
161
118
155
123
130
115
159
135
122
144
141
116
118
165
111
116
123
126
115
117
164
164
180
158
136.500
20.831
157.331
115.669
15.261
30.521
45.782
99.8
76.2
76.2
76.2
91.2
115.0
102.0
99.2
84.4
87.3
94.2
76.2
76.2
107.0
112.0
124.0
76.2
98.3
107.0
102.0
76.2
76.2
95.2
97.0
92.717
14.231
106.948
78.485
15.349
30.669
46.048
36.2
36.6
35.9
35.7
36.6
373
37.5
41.2
36.4
38.0
40.0
36.9
35.6
38.2
38.7
35.4
35.5
36.0
37.8
36.7
35.4
34.7
35.7
36.0
36.833
1.513
38.346
35.321
4.107
8.213
12.320
0.10
0.10
0.10
0.10
0.72
0.10
0.10
0.10
0.10
0.75
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.70
0.10
0.10
0.52
0.10
1.01
0.10
0.229
0.272
0.502
-0.043
118.870
237.740
356.610
0.53
0.53
0.53
0.53
0.53
0.53
0.53
1.76
0.53
0.53
1.69
1.79
0.53
1.50
1.36
0.53
0.53
0.53
0.53
1.80
1.50
2.58
0.53
0.53
0.936
0.608
1.544
0.328
64.996
129.992
194.988
-------
TABLE 2
ANALYTICAL DATA PRECISION FOR KO61
N>
Average
StdDev
CALCIUM
7.13
7.08
7.07
7.03
7.13
7.13
7.01
7.05
7.04
7.15
7.08
7.15
7.00
7.08
7.08
0.049
ARSENIC
ppm
0.17
0.16
0.15
0.17
0.15
0.15
0.19
0.15
0.18
0.17
0.17
0.18
0.18
0.17
0.17
0.013
IRON
NICKEL
ppm
42.8
43.0
42.8
42.9
42.8
42.9
43.1
42.9
42.9
43.0
43.0
43.4
43.2
43.5
43.01
0.210
103
88.9
88.9
88.9
107
99.2
122
119
117
93.3
97.3
108
102
111
103.25
10.875
COPPER
0.19
0.17
0.17
0.17
0.15
0.17
0.18
0.19
0.21
0.17
0.15
0.18
0.23
0.16
0.18
0.021
ZINC
9.09
9.15
9.01
9.05
9.08
9.15
9.13
9.13
9.18
9.03
9.07
9.24
9.24
9.15
9.12
OfflB
RSD%:
0.69
7.6
0.48
10.5
1.16
0.74
-------
FIGURE 1A
o
cc
UJ
0.
1.5
1.4
1.3
1.2
1.1
i
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
ELEMENTAL DISTRIBUTION PERCENT
24 K061 SAMPLES GENERATOR A
LEAD
CHROMIUM
ELEMENT
COPPER
0.
a.
200
190
180
170
160
150
140
130
120
110
100
90
80
70
60
SO
40
30
20
10
0
FIGURE IB
ELEMENTAL DISTRIBUTION PPM
24 KO61 SAMPLES GENERATOR A
CADMIUM BARIUM NICKEL SILVER MERCURY SELENIUM
ELEMENT
11-278
-------
FIGURE 2A
ELEMENTAL DISTRIBUTION PERCENT
z
Ld
O
1.5 -
1 .4 -
1.3 -
1.2 -
1.1
1 -
0.9 -
0.8 -
0.7 -
o.e -
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -
sN
\
\.
v\
\
X
\
s/X
X
V,
k \
N^
^
x^
^O
X^
ง^
XX
N V
XX XV
\ \> v x\
LEAD CHROMIUM COPPER
ELEMENT
a
a
FIGURE 2B
ELEMENTAL DISTRIBUTION PPM
24 K061 SAMPLES GENERATOR A
i an
I JW
180 -
170 -
160 -
150 -
140 -
130 -
120 -
1 10 -
100 -
90 -
80 -
70 -
60 -
50 -
40 -i
30 -
20 -
10 -
\
\
N
\
\
\
*r
X
x\
X
x>
X
\Xv
X
\
\X
Xs
\\
s
0 | 1
X
^
^
\
1^
"s
i
XN
X
x^
XN
X
\x
N
\
x^
N
\
I 1 r
X
\
N
X
X
X
X
\
r
\
k
\
x
s.
X
V
\
\
V
\
^.
X
\\
\
ELEMENT
1-279
-------
FIGURE 3A
Ui
2.2
2
1.3
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
ELEMENTAL DISTRIBUTION PERCENT
4 Koei SAMPLES GENERATOR B
LEAD
CHROMIUM
ELEMENT
COPPER
o.
a
600
500 -
400 -
300 -
200 -
100 -
FIGURE 3B
ELEMENTAL DISTRIBUTION PPM
4 K061 SAMPLES GENERATOR B
CADMIUM BARIUM NICKEL SILVER
ELEMENT
MERCURY SELENIUM
11-280
-------
FIGURE 4A
z
UJ
(J
IT
0.3
0.28
0.26
0.24
0.22
0.2
0.18
0.16
0.14
0.12
0.1
O.O8
0.06
0.04
0.02
0
ELEMENTAL DISTRIBUTION PERCENT
7 FQ06 SAMPLES GENERATOR C
\
LEAD
T ^-TT
CHROMIUM
T"
COPPER
ELEMENT
T
BARIUM
NICKEL
Q.
CL
FIGURE 4B
ELEMENTAL DISTRIBUTION PPM
7 F006 SAMPLES GENERATOR C
100 -
90 -
80 -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
0 -
O<
^x
\\
JvN
I r 1 1 1 f
CADMIUM SILVER MERCURY SELENIUM
ELEMENT
1-281
-------
QO SEMI-QUANTITATIVE DETERMINATION OF INORGANIC
CONSTITUENTS IN SPECIFIC AND NON-SPECIFIC CATEGORICAL
WASTESTREAMS USING EDXRF
Kat Fennel1. Senior Chemist
Roger M. Olbrot, Laboratory Manager
Ted G. Howe, Spectroscopy Group Leader
USPCI, A Subsidiary of Union Pacific
Grassy Mountain Facility
8960 North Highway 40
Lakepoint, Utah 84074
ABSTRACT
X-Ray fluorescence spectroscopy was used to determine
several inorganic constituents and associated interferences
in F006 and K061 wastestreams. The primary objective was
the optimization of stabilization reagents and materials by
analyzing incoming waste loads prior to stabilization. By
applying EDXRF techniques to the pre-accepted waste loads,
an overall increase in site stabilization efficiency can be
realized. Total metal disparities between the pre-
acceptance sample from the generator and the actual load
sample are ubiquitous and problematic. Presently, there
exists a paucity of quantitative data concerning the role of
EDXRF as a useful analytical tool applied to the
environmental analysis of categorical solid waste.
Wastestreams that were approved for treatment, stabilization
and disposal were randomly sampled and subjected to salient
preparation methods. Three instrumental calibration
techniques were investigated using the K061 samples:
fundamental parameters (FPT), single similar standard and
simple linear regression calibration. Because of the extreme
variation of the F006 samples, only two instrumental
calibration techniques were investigated on these samples:
FPT, and a sorting program developed for alloy analysis.
Sample characterization using these techniques can be both
semi-quantitative and quantitative depending on several
parameters, notably, the analyte, method of sample
preparation and interferences. The sensitivity inherent to
11-282
-------
this analytical technique was sufficient to meet our
objectives and provided a certain degree of direction for
further study. Moreover, pre-stabilization profiling or
analysis of these types of wastestreams optimizes the use of
proprietary additives while reducing the total volume of
waste placed in the landfill. A complex database can be
structured and implemented to work in concert with our
rigorous stabilization program furthering our expertise in
the field of waste management.
1-283
-------
IDENTIFYING SOURCES OF ENVIRONMENTAL CONTAMINATION
THROUGH LASER SAMPLING ICP-MASS SPECTROMETRY
Kenneth J. Fredeen. Sr. Technical Specialist, ICP Worldwide Marketing Group,
The Perkin-Elmer Corporation, 761 Main Aye, Norwalk, CT 06859-0215; Mark
Broadhead, Vice President, Chemical and Mineralogical Services, Inc., 445 West
2700 South, Salt Lake City, UT, 64115
ABSTRACT
Laser sampling ICP-mass spectrometry (LS-ICP-MS) is shown to be an effective
technique for helping to identify sources of environmental contamination. This
rapid, sensitive elemental analysis technique requires little or no sample preparation
and provides elemental fingerprints of solid samples in a few minutes. In this study,
LS-ICP-MS was used to analyze high-volume air filters and sludge samples, with
emphasis placed on obtaining elemental isotopic fingerprints of all the samples and
semiquantitative analysis results for the air filter samples.
INTRODUCTION
Determining the elemental composition of environmental contaminants or
contaminated materials is one of the many concerns of environmental regulatory
agencies and the analytical chemistry community. For many sample analyses, the
analysis goal is to determine accurately, a specific list of elements. These analyses
usually have rigid, well-defined analysis protocols and specified maximum levels for
the elements to be determined. For some analyses, however, it can instead be more
advantageous to obtain a full elemental profile of the samples, without the need for
a high degree of accuracy. For other samples, it may also be advantageous to know
the isotopic abundances for specific elements. Such elemental profiles and isotopic
information can often be used as a "fingerprint" to help identify a source of
environmental contamination.
As an example, the presence of smog and particulate matter in air is an important
environmental concern. Determining the relative elemental compositions of the
smog and particulates will help regulatory agencies decide what elements need to be
monitored and regulated. Also important, however, is identifying the point sources
of the emissions so that proper controls can be placed on these sources. This
example can likewise be extended to ground and water contamination.
In this study, laser sampling inductively coupled plasma mass spectrometry (LS-ICP-
MS) has been used to analyze sludge samples of domestic and industrial origins and
high-volume air filters from various locations in a metropolitan area. By comparing
and combining the analysis results with other information, these results can then be
used to help establish the sources of environmental contamination.
The well-established ICP-MS technique has the ability to measure rapidly as many
as eighty elements. Besides the vast numbers of determinations that can be made in
1-284
-------
a relatively short time using the technique, ICP-MS is known for its high sensitivity,
with detection limits generally in the parts-per-trillion range for solution sampling.
As a mass spectrometry-based technique, it also has the advantage of being able to
determine isotope ratios for certain elements present in a sample.
Using laser sampling, as opposed to solution sampling, as the sample introduction
technique, has the advantage of eliminating the usually tedious sample dissolution
step required for most atomic spectrometry techniques. This is accomplished by
focussing a high-power laser beam onto the sample, thereby creating a sample vapor
directly from the solid state. This vapor is then transported to and analyzed by the
ICP-mass spectrometer. Sample results can normally be obtained in a few minutes.
Among the other advantages are its ability to provide spatially resolved sample
information in the form of lateral distributions and depth gradients of elements in a
solid. An increasingly important advantage of eliminating the sample dissolution
step is avoiding the need to work with concentrated acids and to dispose of acid
wastes. A thorough review of the LS-ICP-MS technique has been recently published
[1]-
LS-ICP-MS has been shown previously to be a good technique for rapid,
semiquantitative analysis of various geological materials [2] and several materials of
environmental interest [3],.as well as many other sample types. Given proper
calibration standards for the elements of interest, LS-ICP-MS can also be used for
quantitative analysis of many materials. However, like many other solid sampling
techniques, the availability of appropriate solid calibration standards can limit the
quantitative aspect of LS-ICP-MS for some analyses.
The primary interest in this study was to use LS-ICP-MS to obtain elemental
fingerprints of the various samples analyzed. Such fingerprints indicate the relative
elemental compositions of the samples without regard for the absolute element
concentrations. While some interpretation of the mass spectra is required to
compensate for spectral interferences, subsequent quantitation using the element
intensities, and thus calibration of the system, is not required.
Of secondary interest in this study was to quantitate the results from the analysis of
the high-volume air filters. In order to accomplish this task, a calibration method
appropriate for laser sampling of the air filters had to be developed.
EXPERIMENTAL
Instrumentation
Two different LS-ICP-MS systems were used for data collection in this study. The
first system consisted of a Perkin-Elmer SCIEX Model 320 Laser Sampler coupled
to a P-E SCIEX ELAN 500 ICP mass spectrometer. The laser sampler consists
primarily of a pulsed, Q-switchable Nd:YAG laser; an enclosed sample cell
mounted on a three-axis translation stage; and a closed-circuit video monitoring
system. The laser and translation stages are all computer controlled. The ICP-MS
was controlled using an IBM PS/2 Model 70 microcomputer running the P-E
SCIEX ELAN 5000 software under the Xenix operating system. The laser sampler
1-285
-------
was controlled by an 80286-based microcomputer running the P-E SCIEX laser
sampler software under Windows and DOS.
The second LS-ICP-MS system consisted of a Model 320 Laser Sampler coupled to
a P-E SCIEX ELAN 5000 ICP mass spectrometer. Both the ICP-MS and the laser
sampler were controlled from a single IBM PS/2 Model 70 microcomputer. The
software used was the same as for the first system except that the laser sampler
software was run under Windows and VP/ix. A block diagram of the laser
sampler/ICP-MS system is shown in Figure 1.
PE-SCIEX ELAN SOOO ICP Man Spcctrofiwtw
Syซ(ซni Compute
PE-SCIEX Mod.4 32O Uซw Simpter
Figure 1. Diagram of laser sampling ICP-MS system.
Methodology
Standard and Sample Preparation. For the sludge analysis, the certified reference
materials studied were combined with an X-ray fluorescence binding agent,
SpectroBlend (Chemplex Industries, Inc.), at a sample:binder ratio of 3:1. These
mixtures were shaken well and then pressed into 0.5 g pellets using an IR pellet dye
and a 12-ton press. Had the reference materials not been finely powdered, they first
would have been ground to a 350 to 400 mesh before mixing with the binding agent.
Standards were prepared for the high-volume air filters analysis by evenly loading
measured amounts of standard reference materials (SRM) onto 1.5 x 3.0 cm pieces
of blank filters of the same type used for the air sampling. In order to keep the
SRM's from leaving the surface of the filters prematurely when sampled by the
laser, the filter standards were coated with an aerosol-based binding agent. The air
filter samples were cut into 1.5 x 3.0 cm pieces and also coated with the binding
agent before laser sampling. For solution sampling, 3.0 x 3.0 cm pieces of the filter
samples were leached in 20% nitric acid. The leachate for each sample was then
diluted 20x before analysis. Appropriate blanks were prepared in the same manner
as the respective laser and solution sampling filters samples.
Analysis Procedures - General Standard laser-sampling operating conditions for the
ICP-MS instruments were used for all the analyses. For the semiquantitative
11-286
-------
analyses, the LS-ICP-MS system was calibrated by analyzing a prepared standard for
several certified elements using the TotalQuant routine of the ELAN 5000 software.
The TotalQuant routine then used the results for the certified elements to calibrate
the system for the other elements of interest, for a total of up to eighty elements.
The TotalQuant routine was also used to produce the fingerprint spectra resulting
from the analyses.
Sampling Procedures - Sludges. The reference material pellets were sampled with
the laser in the Q-switched mode, with an energy of 50 mJ/pulse and a repetition
rate of 10 Hz. The laser beam was scanned back and forth across the sample on a 5-
mm line for a 20-second pre-measurement period followed by a 60-second data
collection period. Since only relative sample intensities were desired for this part of
the study, no concentration calibration was performed.
Sampling Procedures - Air Filters. The standard and sample air filters were sampled
with the laser in the Q-switched mode, with an energy of approximately 10 mJ/pulse
and a 10-Hz repetition rate. A 6 x 7 mm area of each filter was sampled using a z-
pattern raster with the laser beam. A 30-second pre-measurement sampling time
was used, followed by 3 successive 60-second data collection periods for each
determination.
RESULTS AND DISCUSSION
Sludge Samples
Spectral fingerprints for the two sludge materials studied were obtained easily using
the described technique. Figures 2 and 3 show the single point per dalton "mass
histograms" for the sludge samples. These spectra were interpreted and elemental
intensities were automatically calculated by the TotalQuant software routine. The
element intensities were then normalized to each sample's total element intensity so
that any differences in laser sampling efficiency could be nullified. The resulting
normalized elemental fingerprints for the sludges are shown in Figures 4 and 5.
7.0
160 120 140 160 180 200 220
1 .0
20
Figure 2. LS-ICP-MS spectral fingerprint for domestic sludge sample.
11-287
-------
7.0
Sanple Indusฑ1- IS liidge:
100 120 140
MflSS < ar-iu
220 241
Figure 3. LS-ICP-MS spectral fingerprint for industrial sludge sample.
0.0000001
H B F A] Cl So Mn Cu Aj Rb Nb Pd Sn Xa Ce Eu Ho Lu Ra Au Bi
Figure 4. Normalized elemental fingerprint for domestic sludge sample.
1 -
0.1 -
0.01 -
0.001 -
0.0001 -
0.00001 -
0.000001 -
n nnnnnm -
-4-
1
' t i MM.
tuu.
I
I
l
ii
i
I
I
I
H B F A/ Cl Sc Mn Cu As Rb Nb Pd Sn Xs Ca Eu Ho Lu Ra Au Si
Figure 5. Normalized elemental fingerprint for industrial sludge sample.
11-288
-------
At first glance, the two sludge elemental fingerprints appear quite similar. However,
upon closer examination, some differences can be seen. In order to make these
difference more apparent, the two elemental fingerprints were ratioed to one
another. Figure 6 shows the ratios of the domestic sludge fingerprint to the
industrial sludge fingerprint. From these ratios, it appears that the domestic sludge
has a higher content of N, F, S, and Se, while the industrial sludge is higher in In,
Te, Pt, and Bi.
1000000 -
100000 -
10000 -
1000 -
100 -
10 -
1 -
0.1 -
0.01 -
0.001 -
0.0001 -
0.00001 -
0.000001 -
N
I y. JU
s
F
-1
Lj-i_A_A
Se
1 W ' ' ' "ป '' i-w-i mm" i i ' -m i i i i ||i in i-i III'II1 ' !ป ||.ปป^"i "i-i|||l 1 1 |||'i;|'|l !|l
I Ce Bi
1
Te Pt
In
Figure 6. Ratios of normalized results for sludges, domestic:industrial.
Because the ratio of two small numbers can still turn out to be a large (or very
small) number, however, it was necessary to apply a filter function to the fingerprint
data before the ratios were calculated. Figure 7 shows the domesticrindustrial
sludge ratios after a le-06 filter function was applied to the data. While the
significant elements on the domestic sludge side didn't change much, the significant
elements for the industrial sludge now appear to be Cd, Sb, Ce, Au, and Bi.
The utility of data such as these goes beyond the ability to determine rapidly the
content of a sample. For the case of the domestic sludge, one might expect a high
organic content, and thus look for carbon as a major constituent. While the
domestic sludge did contain a large amount of carbon, Figure 7 shows that it was not
much higher than was found in the industrial sludge. However, while the nitrogen
and sulfur content of the domestic sludge was much lower than the carbon content,
the relative concentrations of these elements were much higher than those in the
industrial sludge. Therefore, it should be possible to use nitrogen and sulfur as
indicators of the organic content of a sample. Likewise, it appears that the heavy
metals, rare earths, and precious metals may be good tracers for industrial waste.
While these data certainly do not comprise a comprehensive study, combining such
data with methods of principle components analysis shows great promise for helping
environmental scientists to categorize waste types and even pinpoint sources of
environmental contamination.
11-289
-------
10000 -
1000 -
100 -
10 -
1 -
0.1
0.01 -
0.001
0.0001 -
0.00001 -
0.000001
N
F
JLJ l_A_J
Se
.111 i ^. i . J. J. t_i t .!!. i t i_. . i i t ti i _i_ i it _* M tiittttttiiiiitiiitiititJtiLi-ti
B m IgnBTT 1 I l-:im ifB" n I I -m I i ' f|i ||| 111"!*' ' "I'l'II " " w 111' ' ' "ll'll' ''
Nb Sb Ce Au Bi
Figure 7. Domestic:industrial sludge ratios after 1e-06 filter function.
High-Volume Air Filters
In this part of the study, several high-volume air filters from a metropolitan area
were analyzed. These filters sampled the air in various locations in the area and
were designated as "downtown," "North," "petroleum plant," and "airport." There
was also a filter that sampled the output from a point emission source.
Obtaining reproducible spectral and elemental fingerprints of the high-volume air
filters was not quite as straightforward initially as is was for the sludge samples.
Because of the fragility of the glass fiber filters used for this application and the
relatively low laser power with which they could be ablated, the laser sampling
conditions for this application had to be carefully controlled. Once the proper
sampling conditions were established, however, good fingerprints could be
produced readily. The spectral fingerprint for the "downtown" filter is shown in
Figure 8.
Quantitating the results from the air filter analysis required producing an
appropriate calibration standard and analysis method. Previous attempts made in
our laboratories, and those of other workers [4], to provide a proper calibration
standard have included soaking filters with standard solutions, using a mylar-based
filter standard, and coating filters with slurries of reference materials.
Using the first two methods have the disadvantage of having the calibration species
entrained throughout the filter material, whereas the sample filters have the analyte
particles mostly on the surface of the filter. Because the filters used in this
application can contain relatively high levels of some of the elements of interest, the
goal of the laser sampling process is to remove particles from the filter surface while
removing as little of the filter as possible. In addition, removing too much filter
material can leave filter particles deposited throughout the sample transport line,
II-290
-------
causing memory effects. Another possible problem with the solution-spiked filters is
that the chemical forms of the analyte species in the standards and sample are likely
to be quite different. This can cause some differences in the sampling efficiencies
for the species, although using a Q-switched laser should help to reduce the
differences. Finally, there has been no proof to date that this calibration method is
valid for the air filters analysis.
7. a
nple Dountoi-in: Laser- Sampling
160 iaa 2ea 220 2-41
Figure 8. LS-ICP-MS spectral fingerprint for "downtown" air filter.
Advantages of the slurry-coating approach are that the species of interest are
deposited onto the surface of the filter and are more likely to be in the same
chemical form as they are on the sample filters. One potential problem with this
approach, however, is that some analyte species could be leached from the
reference material particles and either be left in the container the slurry was made
in or soaked into the bulk of the filter. Another practical problem with sampling
slurry-coated filters with the laser is that a relatively wide area, up to several
millimeters in diameter, of the reference material is removed by the shock wave
produced by the laser-solid interaction. This may seem to be an advantage at first,
since the sampling efficiency would be large. However, much of the material
removed in this manner is not vaporized by the laser, but instead is removed as
particles that are too large to be analyzed well by the ICP-MS or simply fall out of
the carrier stream and are deposited in the sample line.
The approach used in this study was to load various dry, powdered reference
materials directly onto blank filters. The filters, which were cut to a specific size,
were weighed before and after the reference materials were added so that a loading
factor could be calculated for each filter. This loading factor was then used with the
certified concentrations to determine the loading, in ng cm-2, for the elements of
interest on the filter surface.
When filters prepared in this way were sampled with the laser, the problem with the
wide sampling area, described above, was encountered. To counteract this effect,
the filters were coated with an aerosol-based binding agent and dried. Subsequent
11-291
-------
sampling of the filters was then confined to the immediate area of the laser beam.
Another advantage of this method is that less of the filter substrate was sampled by
the laser, also. Analysis of blank filters coated with the binder only showed that the
binder did not contribute to the background levels of the elements of interest.
Figure 9 shows the semiquantitative analysis results from the laser sampling analysis
of several SRM-coated filters. A filter coated with NIST SRM 1648, Urban
Particulate Matter, was used as the calibration standard for this analysis. SRM's on
the analyzed filters included Coal Fly Ash and Estuarine Sediment. Also included
are results from analysis of a filter coated with the Urban Particulate Matter SRM,
but with a different loading factor than the standard filter. While these results
indicate that this approach used for calibration has some validity, it is clear that
more work needs to be done to refine the technique.
100000
0.1
10
100
Reference
1000
10000
100000
Figure 9. Semiquantitative analysis results (ng crrr2) for SRM-coated filters.
All of the sample filters were also analyzed against the SRM 1648 standard filter.
Figure 10 shows the semiquantitative elemental fingerprint for the "downtown"
sample. Of note are the relatively high results for As, Cd, Hg, and Pb. The other
filters showed similar results. While it was not done in this study, these results could
be compared in the same manner as the sludge results to determine what elements
show the most significant differences and could be used as tracer elements to help
track sources of pollution.
Comparison to Solution Sampling
The analysis of air filters using atomic spectrometry techniques is in wide practice.
While X-ray fluorescence can be used to analyze these filters directly, most analyses
are performed using atomic absorption and ICP techniques which require the
sample to be in a solution form. The most widely-used technique for getting the
samples into solution is to use a nitric acid leaching procedure. In fact, it is because
of this leaching procedure that the glass fiber filters are used instead of the more
common cellulose fiber filter papers.
1-292
-------
N
10000 T
1000
100 -
Loading 10 -
1 -
0.1 -
n nt -
i t i i i .
re
S
t-H illi*.
, L. t i L M.M.M,
Cu
Pb
As
ULA M i
c
Jt-l L. .L-J-J*.
;d
|
A i
t i
c
tM
Hg
W
. . Mill t I M l
\ \ \ \ I Mi
I
H B F A) Cl Sc Mn Cu As Rb Nb Pd Sn Xe Ce Eu Ho Lu Re Au Bi
Figure 10. Semiquantitative elemental fingerprint for "downtown" air filter. (Loading in ng crrr2)
The results for laser sampling ICP-MS were compared to results obtained from
solution sampling ICP-MS for the same filter samples. After nitric acid leaching
procedures were performed on the filters, the leachate solutions were diluted to a
factor that would give ICP-MS results in the same intensity range found for laser
sampling. Figure 11 shows the solution sampling spectral fingerprint for the
"downtown" filter sample. Many features of this fingerprint are similar to those
found for the laser sampling fingerprint of this filter (Figure 8).
Sanplg Dountoun: Solution Sanpling
1 .8
III
lea tea
220 2-fl
Figure 11. Solution sampling ICP-MS spectral fingerprint for "downtown" air filter.
In order to make comparison of the solution and laser sampling results easier, the
results were normalized and ratioed in the same manner as the sludge samples.
Figure 12 shows the ratios comparing the solution and laser sampling results for the
1-293
-------
"downtown" filter sample. Of particular note from these results is that the laser
seems to have provided a more complete sampling for N, S, Cl, Br, and Hg. It is not
surprising that these elements were either not leached well from filters using the
nitric acid procedure or did not remain in solution.
10000 i
1000
100
10
1 -
0.1 -
0.01
0.001
0.0001 -
p
c
\
Mo
Ti
II
I
Ce
Csll
1 II
a Br '
Hg
N
Figure 12. Ratios of solution sampling:laser sampling results for "downtown" air filter,
Because of specific interest in determining sulfur for the filters, the laser sampling
results for sulfur were examined more closely. This interest is compounded by the
fact that X-ray fluorescence determinations for sulfur are suspect in their accuracy.
Figure 13 shows an intensity versus sulfur loading plot for the semiquantitative
analysis of the SRM-coated filters. While this is not a perfect calibration curve, it
does show promise for laser sampling ICP-MS (or LS-ICP-AES) as a method for
determining sulfur on the filters.
1000000
Intensity 100000
10000
1000
10000
Sulfur Loading (ng cm-2)
100000
Figure 13. Intensity (counts/sec) versus sulfur loading results for SRM-coated filters.
I-294
-------
Solution and laser sampling results were also compared for the point emission
source filter sample, since this sample contained high levels of several difference
elements. These results are shown in Figure 14. On the solution side, it was found
that Zn and Ba could be determined better by solution sampling than with the laser.
This is because the blank filters used in this application contained high levels of
these elements which, when sampled by the laser, produced very high background
signals that made them difficult to determine.
100000 -
10000 -
1000 -
100 -
10 -
0.1 -
0.01 -
0.001 -
0.0001 -
0.00001 -
I 'ซ,1 1 '
1 l (H 1 !
N
1
Al
1 1 1 1 1 ll
1 I I r t !
J
c
c
p
1
1
a
i
a
Zn
.jUUluUlJL
I1 '
Sc
S
B
Rb
1
I Cs
1 1 1, ซ_il ,,,!,,. -., 1
Pd Sn
1
e
a
AuHg
Figure 14. Ratios of solution sampling:laser sampling results for point source filter.
For laser sampling of the point source filter, once again the laser seems to be more
proficient for N, Hg, and the halogens. Also note that Au and Pd were sampled
better with the laser. Figures 15 and 16 show the spectral fingerprint data from 170
to 240 daltons for solution and laser sampling of the point source filter. The
difference in the Hg isotope intensities in the 196 - 204 range for the two methods
can be seen clearly, while intensities for Pb (204 - 208), Tl (203 & 205), Re (185 &
187), W (180 - 186), Th (232), and U (238) are quite similar. Also, the Au signal at
197 is quite strong for laser sampling while essentially absent from the solution
sampling data.
As indicated in the sludge application, N, S, and the precious metals may be
important elements for helping to identify and track sources of environmental
contamination. The ability of laser sampling ICP-MS to determine these elements,
even semiquantitatively, could be quite important for this type of application. For
toxic elements such as As, Cd, Pb, and Hg, the importance of the more complete
sampling using the laser is even more apparent. Had solution sampling techniques
alone been used for these samples, it is possible that important elemental
information would not have been uncovered.
Another advantage of using laser sampling for the analysis of air filters is that by
eliminating the sample leaching step, the need for using the glass fiber filters, as
11-295
-------
opposed to cellulose fiber filters, is also eliminated. During methods development
for this application with the laser sampler, it was found that cellulose filters coated
with SRM's could be sampled more easily and reproducibly than the glass filters.
The semiquantitative analysis results for laser sampling of an SRM-coated cellulose
filter are shown in Figure 17. Besides the generally good agreement between the
observed and certified values, note that zinc could be determined using these filters,
whereas it could not be determined on the glass filters because of the high
background levels.
7.e
Solution Sanuling
i.e
IBS
ise
19B
28B 21B
Mass < anu >
22B
23B
Figure 15. Spectral fingerprint from 170 - 240 dalton region for solution sampling analysis of point
source filter.
7.a
l.B
163
isa
ise
280 21B
MASS Canu>
220
23B
Figure 16. Spectral fingerprint from 170 - 240 dalton region for laser sampling analysis of point
source filter.
II-296
-------
1000 T
L 100 -
O
a
d
10
n
g
0.1 I'*!
Ba V Cr Mn Co Ni Cu Zn Go As Se Sr Mo Cs T1 Pb
Element
Figure 17. Semiquantitative LS-ICP-MS results (ng crrr2) for an SRM-coated cellulose filter.
SUMMARY
Laser sampling ICP-mass spectrometry can be used to produce full, semiquantitative
elemental profiles for elements present at trace to major levels in a wide variety of
samples. In this study, the LS-ICP-MS technique has been applied to the analysis of
solid sludge samples and high-volume air filters. The elemental fingerprints
produced by the technique can be used to help identify which elements may be
important to monitor in order to identify sources of environmental contamination.
The technique also shows great promise in being able to determine important
elements that are difficult to determine by other techniques.
REFERENCES
[1] E. R. Denoyer, K. J. Fredeen, and J. W. Hager^a/. Chem. 63, 445A (1991).
[2] M. Broadhead, R. Broadhead, and J. W. Hager,^f. Spectrosc. 11, 205 (1990).
[3] E. R. Denoyer and K. J. Fredeen, "Application of Laser Sampling ICP-Mass
Spectrometry to Environmental Analysis," 1991 European Winter Conference on
Plasma Spectrochemistry, Dortmund, Germany.
[4] D. Potter and R. C. Hutton, "Direct Analysis of Airborne Particulate," Paper No.
1075, 1991 Pittsburgh Conference on Analytical Chemistry and Applied
Spectroscopy, Chicago, IL.
11-297
-------
ICP/MS ANALYSIS OF
TOXIC CHARACTERISTIC LEACHING PROCEDURE (TCLP) EXTRACT
ADVANTAGES AND DISADVANTAGES
Michael G. Goergen. Violetta F. Murshak,
Paul Roettger, Isaiah Murshak and Dan Edelman
ABSTRACT
The use of the ICP/MS analytical technique is growing for
various analytical applications. The authors of this work
describe the advantages and disadvantages of the ICP/MS
analysis of toxic characteristic leaching procedure (TCLP)
extract. This aqueous extract buffered with acetic acid may
cause interferences with the ICP/MS qualitative and
quantitative analyses.
This study presents the matrix spike recoveries for all
of the RCRA metals except mercury (Hg). These metals
include: arsenic (As), barium (Ba), cadmium (Cd), lead (Pb),
selenium (Se), and silver (Ag), as well as zinc (Zn) and
copper (Cu). In addition, these results are compared to
ICP/AES recoveries. Based on these recoveries and other
observations in our laboratory, we conclude that the ICP/MS is
appropriate and convenient for the analysis of TCLP extract.
INTRODUCTION
The recent change in toxic characteristic analysis from
EP Toxicity to the toxic characteristic leaching procedure
(TCLP) provides analytical environmental chemists with another
matrix for analysis. Traditionally environmental laboratories
have performed analyses for metals in the TCLP leachate with
atomic absorption spectrometry and inductively coupled plasma
Atomic Emission Spectrometry (ICP/AES).
Our laboratory has used ICP/AES since 1988 to perform
analyses on the TCLP extract. In 1989 we installed an ICP/MS
to perform analyses of metals. Our incentive for installing
the ICP/MS was the low detection limits the method provides
using traditional aspiration for sample introduction. In
addition, the ICP/MS performs the simultaneous qualitative and
quantitative analyses of up to 70 or so elements. This allows
for rapid sample through put. However, we soon discovered
many other advantages of ICP/MS and it became the analytical
method of preference for metal analyses in our laboratory.
11-298
-------
This work discusses observations made in our laboratory
regarding the advantages and disadvantages of analyzing TCLP
extract for metals with an ICP/MS. They are based on real
life samples submitted to our laboratory by clients. Thus,
this ICP/MS work did not involve a predesigned experiment as
would be done in academic research.
ANALYTICAL METHOD
The inductively coupled plasma spectrometer (ICP/MS) in
use at FECL is the SCIEX ELAN model 250, converted to a model
500 capability (Perkin Elmer, Norwalk, Connecticut). FECL
also has a Perkin Elmer Plasma 40 Spectrometer (ICP). The
ICP/MS system components include Xenic System V Software, and
IBM Personal System 2 computer and an Epson LQ-850 printer.
The various features and conditions of the system are shown in
Table 1.
For standard preparation equal volumes of 10 ppm
multielement stock solution and 10 ppm silver stock solution
are added together resulting in a 5 ppm solution for all
metals. The stock solutions are obtained from PlasmaChem
Associates, Bradley Beach, New Jersey.
Under FECL's Standard Operating Procedures, samples of 50
ml each are spiked with 1 ml of 5 ppm solution resulting in a
0.1 ppm spike concentration for the various elements. To each
of the samples, 2.5 ml of 70% nitric acid (HNOs) is added.
Spiked and unspiked samples are subsequently digested in a
microwave oven for 30 minutes at medium power. After the
samples have cooled down following digestion, they all
receive 1 ml of 25 ppm internal standard. The samples are
diluted to their final volume of 50 ml yielding a 0.5 ppm
concentration. The samples are then ready for internal
standard ICP/MS analysis.
The internal standard is made up by combining 10 ml of
the 1000 ppm solutions for each of the five standards, i.e.
In, Se, Y, Rh and Re. The total volume is diluted to 400 ml
of 25 ppm concentration. The internal standards are supplied
by Inorganic Ventures, Inc. of Toms River, New Jersey.
For instrument calibration, two calibration solutions of
0.50 ppm and 0.10 ppm, respectively, are prepared weekly from
the multielement stock solution and internal standard. For
the 0.5 ppm calibration solution 5 ml of 10 ppm multielement
stock solution are mixed with 2 ml of 25 ppm internal standard
and bringing the mixture to a final volume of 100 ml. The 0.1
ppm calibration solution combines 10 ml of 1.00 ppm
1-299
-------
TART.B 1
ICP/MS Operating Conditions
ICP torch
Forward power
Argon flowrate (L/min)
Outer
Auxiliary
Aerosol gas
Sampling position
Sampler
Skimmer
Ion lens settings
Bessel box barrel
Bessel box plate lens
Einzel lenses 1 & 3
Einzel lens 2
Bessel box stop
Operating pressure
Interface
Quadrupole chamber
Data acquisition
Isotopes monitored
Ames laboratory design (28); outer
tube extended 30 mm from inner
tubes
1.25 kW
12
0.5
1.0 - 1.2
22 mm above load coil, on center
Nickel, 1.2 mm orifice
Platinum, 0.90 mm orifice
Upgraded ion optical system
+30 ฑ 5 V
-13 ฑ 3 V
-70 V
-130 V
-30 ฑ 5 V
1 torr
2.3 x 10-6 torr
Multi-element monitor ing mode, normal
resolution setting; three
measurements per peak spaced 0.1
m/z units about peak top; dwell
time at each position is 20 ms,
with total measurement time of
0.17 s allows detection of three
analytes per injection without
missing tops of peaks.
B2cr, escu, eezn, ^B^S, ^BSe,
107-Ag, niCd, is^Ba, 208Pb
11-300
-------
multielement solution and 2 ml of 25 ppm internal standard and
diluting to a volume of 100 ml. A calibration blank and check
standard are also prepared weekly. The blank is made up of 5
ml of 1.00 ppm multielement solution and 2 ml of 25 ppm
internal standard. The mixture is added to 100 ml of
deionized water to result in a 0.05 ppm concentration. The
check standard of 0.2 ppm is prepared by diluting 2 ml of 10
ppm multielement solution and 2 ml of 25 ppm internal standard
to 100 ml. The check standard is used each time after five to
seven samples are run as an independent calibration solution.
Operation of the ICP Atomic Emission Spectrometer
(ICP/AES) is similar to that of the ICP/MS except the spiking
level of samples is higher. The ICP samples are spiked with
5 ml of 10 ppm multielement stock solution to an element
concentration of 1 ppm.
We use EPA method 200.7 for ICP/AES analysis and EPA
method 200.8 for ICP/MS analysis. Furthermore, we use EPA
method 1311 for the TCLP extraction.
The data presented herein are grouped by sample type and
type of extraction fluid. Method 1311 requires Fluid ttl which
is more buffered for samples with low pH (pH <5). The method
requires Fluid #2 when a special HC1 test yields a pH of
greater than five. Figure 1 shows a flow chart of the TCLP
procedure.
11-301
-------
Figure 1
TCLP EXTRACTION PROCEDURE
Determine
Sample pH
See Start
Procedure
Below
filtrate <0.5*
of total mass
Determine
Sample Type
solid
Filter and
Veigh
Filtrate
solid portion
Place Solid
Portion into
Borosilicate jar
u
pH > 5
JLI
Ho Extraction
Necessary; Filter
and Continue as is
Store
Filtered
Liquid
\.pH
Add (2L x KFiltrate)
of Fluid tt2
Add (2L x XFiltrate)
of Fluid ttl
Tumble for
18 Hours
Remove from
Tumbler
Measure pH
of Extracted Fluid
if solid
Re-add Filtered
Portion to Jar
Let Settle
(as Becessary)
Filter
Store 250ml
of Filtrate in
Plastic Bottle
Proceed to
Digestion
NOTES:
1. Fluid #1: 5.7ml HOAc/L
+ 64.3ml NaOH/L.
2. Fluid *2: 5.7ml HOAc/L
Start Procedure:
Add 5g sample to 96.5ml
DI Water. If pH > 5,
add 3.5ml HC1. Heat to
60 C and recheck pH.
Use this pH value to
determine extraction
fluid needed.
11-302
-------
ANALYTICAL RESULTS
Table 2 summarizes the detection limits established
according to CFR 136 Appendix B for ICP/MS and ICP/AES
analysis of the TCLP analytes. It also provides detection
limits from the analytical methods published in EPA method
200.7 and 200.8 for comparison. Note, for most of the metals
presented there is a significantly lower detection limit for
the ICP/MS. Indeed in many cases, it is one hundred times
lower, than ICP/AES detection limits. But, in every case, the
ICP/MS detection limit is at least a factor of ten lower.
These lower detection limits offer one of the significant
advantages of the ICP/MS. That is, the ICP/MS has the ability
to analyze a sample diluted to the point where matrix effects
are minimized and still yields an acceptable detection limit.
Tables 3, 4, and 5 summarize the performance of the
ICP/MS for known and blank spike samples. The knowns are
standards obtained from outside our laboratory certified for
the concentrations shown in the table. The blank spikes were
dilutions of our standard solutions. The diluted
concentrations are shown in Table 4.
Tables 3 and 4 show the accuracy obtained with the
ICP/MS. Standard deviations of less than nine percent are
obtained in every case. Table 5 shows the precision of the
ICP/MS from the analysis of duplicate blank spikes. These
tables summarize the accuracy and precision that we routinely
observe with the ICP/MS on these standards.
Tables 6, 7 and 8 show the percent recoveries for the
metals using the ICP/MS and ICP/AES. These tables are grouped
by sample type and the type of extraction fluid used in the
TCLP. In Tables 6 and 8 the standard deviations for all of
the metals, except for arsenic, copper and zinc in Table 6 and
for cadmium and selenium in Table 8 are lower with the ICP/MS.
For the soil matrix in Table 7 ICP-40 provided lower standard
detections for most of the metals. However, the ICP/MS
samples were spiked with 0.1 ppm of analyte compared to 5 ppm
for the ICP/AES analysis.
The ICP/AES, of course, is an established method for
toxic characteristic analyses. So, these observations
indicate that the ICP/MS is also a useful method for these
analyses.
Tables 9, 10, 11 and 12 show the spike recoveries from
the ICP/MS analysis of TCLP extract for four of our most
common sample types. Again, these tables group similar
extraction fluid and sample types. Table 9 presents the
1-303
-------
As
Bo
Cd
Cr
Cu
Pb
Se
Ag
Zn
TABLE 2
Method Detection Limits
ICP-MS(mg/l)
Calculated1 EPA Estimated 2
0.0015 0.0009
0.0015 0.0005
0.0009 0.0001
0.0005 0.00007
0.0012 0.00003
0.0007 0.00008
0.0047 0.0050
0.0032 0.00005
0.0032 0.0002
ICP-40(mg/l)
Calculated1 EPA Estimated 3
0.20
0.053
0.02
0.01
0.04
0.01
O.OB
0.50
0.05
0.02
0.002
0.004
0.007
0.006
0.042
0.075
0.007
0.002
1. Using CFR136 Appendix B. Based on 10 Blank Spikes of O.OIppm
2. From EPA Method 200.8.
3. From EPA Method 200.7.
11-304
-------
TABLE 3
Quality Control Knovns by ICP/MS X Recovery
ICP/MS
QC-ICP InowoB
g/1 SET II SET 12 SIT 13 SET 14 SET 15 Mean S*
As
Ba
Cd
Cr
Cn
Pb
Se
Ag
Zn
0
0
0
0
0
0
0
0
0
.2
.2
.2
.2
.2
.2
.2
.2
.2
97
99
98
97
99
101
101
103
100
104
106
103
104
102
103
104
100
96
96
98
95
97
98
101
98
94
97
96
105
98
98
96
96
113
111
95
101
98
102
99
99
101
104
114
107
98.
102.
98.
99.
98.
100.
104.
102.
97.
80
00
50
00
75
25
00
00
00
3.
4.
3.
2.
2.
2.
5.
8.
5.
56
06
36
92
17
61
61
57
34
TABLE 4
Blank Spikes by ICP/MS X Recovery
As
Ba
Cd
Cr
Cn
Pb
Se
Ag
Zn
g/1
SET fl SET 12 SET t3 SET 14 SET 15
0.05
0.
0.
0.
0.
0.
0.
0.
0.
05
05
05
05
05
05
05
05
104
104
110
100
108
108
108
110
108
100
102
100
98
99
100
114
100
111
96
104
100
94
100
104
98
107
95
108
105
96
95
98
98
98
100
99
94
102
99
100
99
100
97
96
100
Mean
100.
103.
101.
96.
101.
102.
104.
104.
103.
40
75
50
75
25
50
50
25
25
5
1
5
2
4
4
7
6
6
S
.73
.40
.32
.89
.12
.04
.80
.02
.70
TABLES
Blank Spike Difference in Duplicates by ICP/MS
As 0.05
Ba 0.05
Cd 0.05
Cr 0.05
Cu 0.05
Pb 0.05
Se 0.05
Ag 0.05
Zn 0.05 0 4.8 1.6 0 6.8 1.60 3.26
*S = Standard Deviation
1 SET 12 SET 13 SET 14 SET 15
0
0
0
0
0
0
0
0
0
3.4
0
0
0
0
0
0
0
0
0
0
1.9
0
0
0
0
4.1
0
0
0
0
0
0
1
5
1
0
.2
0
0
.9
.0
0
0
Mean
0
1
0
0
0
0
0
0
.00
.88
.00
.00
.47
.00
.00
.00
S
0.00
1
0
0
2
0
0
0
.92
.00
.00
.83
.50
.00
.00
11-305
-------
ICP-40 & ICP/MS
TABLE 6
Spike Becoveries in % for TCLP Analysis of Industrial Faint Sludge
Comparison Between ICP/MS and ICP-40
SET 14
ICP/MS
IHDUSTilAL PAIHT SLUDGE
Em.l 1 1
6711-E1 6711-E2
Mean
S*
Saiple
As
Ba
Cd
Cr
Cu
Pb
Se
Ag
Zn
1
180
80
100
110
80
90
175
82
170
2
168
92
100
98
98
84
160
80
90
174.00
86.00
100.00
104.00
89.00
87.00
167.50
81.00
130.00
8.49
8.49
0.00
8.49
12.73
4.24
10.61
1.41
56.57
ICP-40
INDUSTRIAL PAINT SLUDGE
EXTLI 1 1
6711-E1 6711-E2
Mean
S*
Saiple
As
Ba
Cd
Cr
Cu
Pb
Se
Ag
Zn
1
85
114
86
98
93
90
130
10
94
2
88
88
75
83
78
69
107
7
80
86.50
101.00
80.50
90.50
85.50
79.50
118.50
8.50
87.00
2.12
18.38
7.78
10.61
10.61
14.85
16.26
2.12
9.90
TABLE?
Spike Becoveries in X for TCLP Analysis of Soils
Comparison Between ICP/MS and ICP-40
ICP/MS
SOILS
EITE.I
Saiple
As
Ba
Cd
Cr
Cu
Pb
Se
Ag
Zn
2 2 2
6704-E1 6704-E2 6734-E1
1
149
R!J
106
110
94
99
159
81
190
2
136
100
100
94
85
93
131
80
70
118
90
80
76
73
75
123
38
80
Mean
134.33
95.00
95.33
93.33
84.00
89.00
137.67
66.33
113.33
S*
15.57
7.07
13.61
17.01
10.54
12.49
18.90
24.54
66.58
ICP-40
SOILS
EITB.t 2 2 2
6704-E1 6704-E2 6734-E1
S*
Saiple
As
Ba
Cd
Cr
Cu
Pb
Se
Ag
Zn
1
99
169
90
94
90
88
87
64
107
2
110
108
91
100
93
80
105
13
91
87
71
78
74
77
73
116
107
74
98.67
116.00
86.33
89.33
86.67
80.33
102.67
61.33
90.67
11.50
49.49
7.23
13.61
8.50
7.51
14.64
47.06
16.50
*S : Standard Deviation
11-306
-------
TABLE 8
Spike Recoveries in X for TCLP Analysis of WTP Sludge
Comparison Between ICP/MS and ICP-40
SIT 15
ICP/MS ICP-40
WTP SLUDGE WTP SLODGE
im.S 1111 EXTR.I 1111
6613-11 6613-E2 6613-13 6613-E4 Mean S* 6613-E1 6613-E2 6613-E3 6613-E4 Mean S*
Saiple 1 2 Saiple 1 2
As 110 105 100 104 104.75 4.11 As 104 94 103 76 94.25 12.97
Ba 100 100 100 100 100.00 0.00 Ba 58 74 89 59 70.00 14.63
Cd 78 74 80 80 78.00 2.83 Cd 86 85 85 81 84.25 2.22
Cr 141 141 141 135 139.50 3.00 Cr 100 96 89 87 93.00 6.06
Cn 90 84 77 75 81.50 6.86 Cu 104 85 89 85 90.75 9.03
Pb 77 75 78 77 76.75 1.26 Pb 86 87 75 78 81.50 5.92
Se 89 81 128 130 107.00 25.63 Se 100 100 86 90 94.00 7.12
Ag 68 69 70 70 69.25 0.96 Ag 47 15 20 4 21.50 18.27
Zn 100 100 100 100 100.00 0.00 Zn 85 62 70 67 71.00 9.90
*S = Standard Deviation
1-307
-------
TABLE 9
Spike Recoveries in X on TCLP Extract of soils by ICP/MS
with Mean and Standard Deviation(s)
SOILS
ERR* 2222 22222222222222 Mean Sซ
As 112 117 119 130 179 104 124 146 105 121 116 106 94 112 97 118 136 149 121.39 20.24
Ba 140 220 270 100 117 100 100 100 100 100 100 72 100 100 100 90 100 149 118.18 48.78
Cd 100 96 103 102 116 104 94 112 95 109 103 99 101 98 103 80 100 106 101.17 7.54
Cr 135 145 130 118 133 126 115 145 128 128 126 109 103 114 107 76 94 110 119.00 17.13
Cn 95 103 100 82 89 100 82 89 76 83 92 78 76 79 78 73 85 94 86.33 9.02
Pb 100 60 73 76 79 95 74 84 74 81 83 79 82 77 82 75 93 99 81.44 9.79
Se 147 142 153 157 199 130 127 116 96 116 106 103 98 100 99 123 131 159 127.89 26.88
Ag 79 70 77 71 72 90 86 96 91 86 90 79 91 76 89 38 80 81 80.11 12.67
Zn 100 80 95 80 105 100 59 72 50 86 140 100 71 54 80 80 70 190 89.56 31.98
TABLE 10
Spike Recoveries on TCLP Extract of WTP sludge
with Mean and Standard Deviation!s)
WTP SLUDGE
EM*
As
Ba
Cd
Cr
Co
Pb
Se
Ag
Zn
1
94
106
103
94
91
79
118
114
185
1
97
35
102
100
88
82
122
109
85
1
102
99
100
101
86
80
135
105
80
1
137
108
100
117
92
84
154
115
88
1
117
100
92
111
76
55
148
84
160
1
135
100
99
128
85
70
141
101
100
1
130
94
106
104
86
52
86
86
100
1
80
62
82
106
80
82
72
74
100
1
110
100
78
141
90
77
89
68
100
1
105
100
74
141
84
75
81
69
100
1
100
100
80
141
77
78
128
70
100
1
104
100
80
135
75
77
130
70
100
Mean S*ซ
109 16.7
94 18.7
91 11.2
118 17.2
84 5.7
74 9.9
117 26.8
89 18.1
108 30.0
- TCLP Extraction Fluid f (i.e. Fluid II or Fluid 12)
**S = Standard Deviation
11-308
-------
TABLE 11
Spike Recoveries on TCLP Extract of Industrial sludge
with Mean and Standard Deviation(s)
INDUSTRIAL SLUDGE
EXTR* 1 1 1111 Mean S**
As 135 116 104 120 180 168 137 27.8
Ba 110 100 98 100 80 92 97 9.1
Cd 95 104 104 91 100 100 99 4.7
Cr 130 100 106 104 110 98 108 10.6
Cu 89 92 90 96 80 98 91 5.8
Pb 74 120 72 72 90 84 85 16.9
Se 147 96 114 125 175 160 136 27.2
Ag 91 100 92 84 82 80 88 6.9
Zn 100 96 100 34 170 90 98 39.5
TABLE 12
Spike Recoveries on TCLP Extract of Industrial sludge
with Mean and Standard Deviation(s)
INDUSTRIAL SLUDGE
EXTR* 2 2 Mean S**
As 138 91 115 23.5
Ba
Cd
Cr
Cu
Pb
Se
Ag
Zn
*EXTR = TCLP Extraction Fluid # (i.e. Fluid #1 or Fluid
**S = Standard Deviation
100
102
86
100
86
145
100
100
100
100
89
80
77
89
109
100
100
101
88
90
82
117
105
100
0.0
1.0
1.5
10.0
4.5
28.0
4.5
0.0
1-309
-------
percent spike recoveries for soils extracted with Fluid #2.
The highest standard deviation of all the metals shown is for
barium. It is 1.5 times the next higher standard deviation
(44 vs. 32). However, in Tables 10, 11 and 12 the standard
deviations for Barium are more than two times lower- We
cannot explain this.
The standard deviation for wastewater treatment plant
sludge (Table 10) and industrial sludge extracted with Fluid
#1 (Table 11) and Fluid #2 (Table 12) show standard deviations
less than 30 percent for all metals. This is at sample spike
levels of 0.1 ppm analyte concentration.
The percent spike recoveries presented here are typical
of those we observe in our TCLP analysis. Please note that
none of these samples have been tested for matrix
interferences. Indeed, the samples themselves have a
significant impact on the recovery of the metals. The ICP/MS
performs about the same for each of the four sample types.
OQHCTJJSIOMS
Based on the data summarized in this work and on other
observations made in our laboratory we find the ICP/MS to be
an appropriate and convenient method for analyzing TCLP
extracts. We find the advantages of the method out-weigh the
disadvantages. The major advantages and disadvantages of the
ICP/MS analysis which we've identified are summarized in Table
13.
f 1-310
-------
TABT.K ia
ICP/MS Analysis of TCLP Extracts
Advantages and Disadvantages
Advantages:
Fast - 3 readings per sample for 10 metals in less than
20 seconds
Able to verify calibrations at any place in the run
and then continue
Recalibration to compensate for drift by internal
standards takes only a few seconds
High sensitivity allows diluting to minimize matrix
effects
Avoids atomic emission interferences
Uses only 2 ml of sample for analysis
Able to screen samples semi-quantitatively
High sensitivity requires significant diluting of
higher analyte concentrations to be within the
calibration range (or switch to high concentration
mode)
Operator must be knowledgeable in recognizing and
correcting interferences
11-311
-------
oc SEVENTH ANNUAL WASTE TESTING
ฐ^ AND QUALITY ASSURANCE SYMPOSIUM
ABSTRACT
Submitted by:
Larry B. Lobring, Chief
Inorganic Chemistry Branch, Chemistry Research Division
Environmental Monitoring Systems Laboratory - Cincinnati
Office of Research and Development
U. S. Environmental Protection Agency
(513) 569-7372, FTS 684-7372
Title:
Chromium VI; An Overview of Its Relevant Environmental Occurrence, Analytical
Methods of Quantitation, and Report on Recent Ion Chromatography Methods
Development and Validation Activities.
This presentation covers the various forms of chromium found in nature and
those that are significant in environmental samples and to human and ecosystem
health. The interconversion of trivalent and hexavalent chromium in the
environment and related problems associated with sample collection,
preservation and quantitation of the various species is discussed. Topics
covered include the current analytical methodology that utilizes
chelation/extraction or coprecipitation with iron or lead. These approaches
have several potential chemical interferences or deficiencies that are
discussed.
A description of recent methods development studies, utilizing ion
Chromatography and inductively coupled plasma mass spectroscopy, for the
determination of total and hexavalent chromium in incinerator particulate
emissions is presented. The ion Chromatography method developed in this study
was adapted for use in aqueous environmental samples and is now available for
use in the Environmental Protection Agency's compliance monitoring programs.
The water method is identified as Method 218.6, " Determination of Dissolved
11-312
-------
Hexavalent Chromium in Drinking Water, Ground Water and Industrial Effluents
by Ion Chromatography". Results of a recently completed multi-laboratory
method validation study conducted in cooperation with ASTM are presented.
Additional efforts needed in the area of sample processing to extend
application of the technique to a wider variety of sample types will also be
presented.
1-313
-------
og Rapid High Performance Microwave Digestion
Ron Rubin. Michael Moses, Questron Corporation, PO Box 2387,
Princeton, KJ 08543-2387
Microwave Digestion Techniques have reduced the time required to
place a sample in solution. However, there are still limitations in
sample handling, cooling and recoveries of elements* In this paper
we will present several different types of Digestion Systems and show
how each of them addresses the above mentioned problems. To be
covered in the study are: Closed Vessel Microwave Ovens; Open Vessel
Microwave; High Pressure conventional Digestion, and High Pressure
Microwave Digestion. Comparisons are made based upon what we
consider the two most important operating factors: throughput and
recovery. Throughput encompasses all of the cost factors such as
speed of digestion, speed of cooling, number of samples per batch,
amount of reagents and operator time* Recovery, especially its
reproducibility, defines the success or failure of the procedure.
1-314
-------
87
THE PERFORMANCE OF A LOW COST ICP-MS
FOR THE ROUTINE ANALYSIS OF ENVIRONMENTAL SAMPLES
R. Craig Seeley. Thomas M. Rettberg, Peter D. Blair, Fisons Instruments,
24 Commerce Center, Cherry Hill Drive, Danvers, Massachusetts 01923
ABSTRACT
Inductively Coupled Plasma Mass Spectrometry (ICP-MS) is rapidly being
recognized as the choice of instrumentation for trace element analysis of
environmental samples. It possesses a number of distinct advantages when
compared with other established techniques: sensitivity, speed and
versatility.
This presentation will demonstrate the performance and cost effectiveness
of the new Fisons PQe ICP-MS. The system was designed to meet the
specific and varying needs of laboratories concerned with analyzing
environmental samples. Data will be presented on water and soil samples
according to USEPA Methodology 6020 showing the PQe's multi-element
capability and determination of all elements in a single acquisition.
Detection limit studies will be presented along with information on the
profit potential of the PQe when used to perform contract analysis of
environmental samples.
INTRODUCTION
For a laboratory to enter the Environmental Protection Agency's Contract
Laboratory Program, it must have the capability to determine 22 metals in
water and soils. Until recently, most laboratories have been using a
combination of ICP-OES and GFAA instrumentation. The choice of
instrumentation is made on the basis of the CLP contract required
detection limits (CRDL's). The ICP-MS technique is particularly suitable
for environmental analysis due to its exceptional multi-element
sensitivity. ICP-MS can meet or exceed the CRDL's for all 22 elements in
Method 6020 (with the exception of Se in soil) and perform the analysis in
one sample cycle. Two instruments are now combined in one.
PROFIT POTENTIAL AND DESIGN
Laboratories performing high through-put routine environmental analysis
have been reluctant to invest in such technology because of the high
capital cost of current ICP-MS instrumentation. However, using a
completely new approach to ICP-MS, Fisons has introduced the new PQe. A
customized, low-cost ICP-MS instrument specifically aimed at the
environmental market. The instrument is based on a radical new design
which emphasizes robustness of hardware and simplicity of operation. A
completely new mass spectrometer and detector system is employed and the
benefits to the analyst are detection limits and a dynamic range more than
sufficient to meet EPA legislative requirements1. Profitability
projections will be presented in this poster/paper to help illustrate the
cost effectiveness of the new ICP-MS design.
ANALYTICAL REQUIREMENTS FOR WATER ANALYSIS
The term "water analysis" covers a wide variety of sample types and
matrices. In particular, environmental water samples may vary
significantly in terms of inorganic and organic dissolved solid content
1-315
-------
and suspended material. The PQe has a range of features which enable it
to deal routinely with this wide range of sample types, for example matrix
independent calibration and wide linear dynamic range.
Matrix independent calibration
The PQe requires only a single set of calibration standards, even for the
analysis of different sample matrix types, such as rain water, riverine
waters and effluent. Matrix independent calibration obviates the need to
run different standards for each type of matrix, or perform standard
additions on each sample, thus saving valuable analysis time.
Wide Linear Dynamic Range
Environmental water samples may include analytes at high concentrations
e.g. Na, Mg, K and Ca as well as the trace and ultra-trace components e.g.
Cr, Cd, Tl and U. For efficient sample analysis, it is essential that all
the elements of interest should be determined in the same solution,
without the need for preconcentration, separation or dilution. The wide
linear dynamic range of the PQe allows the determination of major, minor
and trace elements in a single acquisition, without the need for operator
input. Furthermore, there is no necessity to match the concentration of
each analyte in the calibration standard to the expected sample
concentration, thus simplifying calibration procedures and further
improving sample through-put2.
ANALYTICAL PERFORMANCE
A series of experiments were carried out to assess the performance of the
PQe in terms of accuracy, precision, spike recovery, stability, dynamic
range and detection limits. Data will be presented on certified materials
as well as routine water and soil digested samples.
SUMMARY
The EPA Contract Laboratory Program protocol for inorganics is a
complicated program to enter successfully. With the help of low cost,
simplified, and high sample through-put instrumentation, it becomes a
straight forward and profitable task.
References:
^e, C.T., et al, 1991 Pittsburgh Conference, March 3-8, 1991
2PQe Technical Note 2, VG Elemental, Winsford, Cheshire, UK, 1990
11-316
-------
Robotics For Automated Digestion Of Environmental Samples
C.Balas7 Questron Svsteiaa. A. Grfrllo,Questron Corporation, P.O. Box
2387, Princeton, NJ 08543-2387
Microwave Ovens are ideal for preparation of Environmental samples
for metals analysis. However, the oven does present problems of
vessel handling and storage. A new robotic system, utilizing
several microwave stations, has been configured to digest samples, at
a sufficiently high rate, to enable the digestion to keep up with the
pace of a simultaneous ICP system. Protocols for many different
types of samples can be stored, recalled/ and implemented, in order
to allow the robotics to accept many different samples of various
sizes and consistencies. In our paper we will describe the software
and protocols and show how they can be utilised to accommodate the
day to day changes in the types and quantities of samples encountered
in the typical environmental laboratory.
1-317
-------
QQ APPLICATION OF LASER SAMPLING ICP-MASS SPECTROMETRY TO ENVI-
0 RONMENTAL ANALYSIS
K. J. Fredeen, R. J. Thomas
Perkin-Elmer Corporation
Norwalk, Connecticut
Laser Sampling ICP-Mass Spectrometry is increasingly becoming recognized as an
analytical tool for the direct analysis of solid samples. Early work with LS-ICP-MS
focused mainly on geological and metallurgical type applications mainly because of
the ability to bypass the lengthy sample dissolution stage.
However as the technique progresses, other application areas for LS 1C? MS nre
becoming more and more attractive. One such area is in the analysis of environ-
mental type applications. The ability to bind and/or press samples into a small
pellet allows LS-ICP-MS to be used for the analysis of samples such as urban
particles or river sediments.
This work will discuss some of the capabilities and limitations of LS-ICP-MS for
the analysis of these type of environmental samples. In addition, approaches to
the difficult problem of sampling some of these materials will be discussed.
RT:td.329
11-318
-------
REGULATORY
COMPLIANCE
-------
QQ Status of Developing Land Disposal Restrictions
for Superfund Soils
Richard Troast
Carolyn Offutt
U. S. Environmental Protection Agency
Office of Solid Waste and Emergency Response
Washington, DC
Joan O'Neill Knapp
CDM Federal Programs Corporation
Fairfax, Virginia
ABSTRACT
RCRA Land Disposal Restrictions (LDRs) for contaminated soil and
debris at Superfund sites are currently being developed. This
paper will present the current status of the EPA sponsored
testing and the design of an integrated data base for both
technology transfer and the development of the LDRs.
/
The unique physical and chemical characteristics of Superfund
soil and debris make these wastes more difficult to treat than
more homogeneous industrial process wastes. The National
Contingency Plan acknowledges that Best Demonstrated Available
Technology (BOAT) standards are generally inappropriate for
Superfund soils. In response to this, EPA is in the process of
developing separate LDR standards for contaminated soil and
debris (CSD). LDRs for CSD are being developed under section
3004 of the Hazardous and Solid Waste Amendments of 1984 to RCRA.
1
11-321
-------
Until the final CSD standards are in place, treatability variance
levels, also based on the actual treatment of soil, will be
used. In addition, the paper will discuss some preliminary
findings on the treatment of debris, and the analytical methods
used for determining the BDAT for CSD.
1.0 INTRODUCTION
Section 3004(m) of the Resource Conservation and Recovery Act
(RCRA) mandates that the U. S. Environmental Protection Agency
(EPA) require treatment of hazardous wastes prior to land
disposal. Known as the "land disposal restrictions" (LDRs),
these regulations were designed for industrial process wastes
defined to be hazardous under RCRA. They apply as well to
contaminated soil/ sludge and debris from RCRA facilities and
Superfund sites. RCRA requirements for treatment are mandatory
and self-implementing at all RCRA regulated facilities, but apply
at a CERCLA site only if a) the waste is a RCRA listed or
characteristic waste; b) the CERCLA activity constitutes
treatment of RCRA hazardous waste, as defined under RCRA; and c)
the treatment activity constitutes "placement."
The Office of Solid Waste (OSW) is responsible under EPA's Office
of Solid Waste and Emergency Response (OSWER), for responding to
directives under RCRA, and therefore, prepares and presents the
LDR standards to the regulated community.
2
11-322
-------
The Office of Emergency and Remedial Response (OERR) is
responsible under OSWER for responding to directives under the
Comprehensive Environmental Response, Compensation, and Liability
Act (CERCLA), or Superfund activities. As the majority of soil
and debris contaminated with hazardous wastes are found on
Superfund sites, LDRs have a profound potential effect on the
government's efforts at site remediation.
OSWER has recognized that contaminated soil is more difficult to
treat than RCRA industrial process wastes, and that it is not
likely that these wastes can be treated to meet the LDRs
developed for RCRA listed wastes because of the physical and
chemical complexity of contaminated soils. In response, OSWER
initiated a program to develop Treatability Variances, which are
alternate treatment standards based on actual treatment of
Superfund and RCRA soil and debris. Data was collected, and in
1989, treatability variance levels were established for soils
utilizing 67 data sets (Superfund LDR Guides #6A and #6B).
OSWER developed a strategy for calculating variance levels from a
quantity-limited data base. The data are categorized into 13
"contaminant groups" which are groups of contaminants having
similar chemical and physical characteristics. Examples of
contaminant groups include non-polar halogenated aromatics, and
PCBs/dioxins/furans and their precursors. The variance levels
3
11-323
-------
that were developed quantified the effectiveness of various
available technologies on the contaminant groups.
EPA OSWER determined that the existing soil treatment data base
was not comprehensive enough to support a formal set of LDRs for
CSD. Several available technologies had insufficient performance
data to develop regulations. EPA therefore implemented a
research program to obtain all of the necessary data to support
the development of LDRs for CSD. In 1988, OSWER's OERR, OSW, and
Technology Innovation Office (TIO), and the Office of Research
and Development's (ORD) Risk Reduction Engineering Laboratory
(ORD-RREL) in Cincinnati, Ohio established a work group to
develop BOAT standards for CSD. The work group objectives
include a review of the current data base, recommendations for
additional studies on treatment performance, implementation of
treatability studies, collection of new available data, and
development of BOAT regulations based upon new and available
data. There has been significant progress with these efforts.
2.0 DATA COLLECTION/DATA BASE DESIGN AND OPERATIONS
OSWER, in its initial data collection effort, collected and
examined over 500 studies conducted by the EPA, federal agencies,
industries and universities. These studies formed the basis for
the development of treatability variances. Of these studies, 67
met the criteria established for the development of variance
4
11-324
-------
levels for contaminated soils. The established criteria required
that the: (1) data be of sufficient quality; and (2) untreated
and treated soil contamination be measured. The current criteria
for setting final LDR treatment standards are more rigorous than
the criteria for variance levels. They require more
documentation of quality assurance/quality control (QA/QC)
procedures as well as bench, pilot and full-scale testing data.
A formal data summary form (DSF) has now been developed by OSWER
to extract pertinent data from all studies reviewed for inclusion
into the data base.
EPA utilizes a four-tiered project category approach in its QA
program in order to more effectively focus QA. Category I
involves the most stringent QA approach, whereas Category IV
represents the least stringent. Category II projects are those
producing results that complement other inputs and are designed
for use in rulemaking, regulation making, or policy making.
Therefore, all data used to support the CSD LDRs should have a
Category II objective designed into the QA project plan (QAPjP).
After a thorough QA review using the established criteria, only
13 of the 67 studies used for variance levels were determined to
be adequate for consideration in the development of LDR treatment
standards. However, all studies reporting data are accepted as
Category IV data and included in the data base for technology
transfer purposes.
5
11-325
-------
Lack of soil treatment data prompted an aggressive data
collection effort by OSWER and ORD. Figure 1 shows the system
for data collection and treatment research in the CSD program.
Additional data will be collected from recent remedial/removal
actions/ including DOD and DOE actions/ SITE program
demonstrations, and treatability tests conducted by the CSD
program. Currently the new data base is planned to contain not
only the original data base/ but studies that have been collected
since the variance levels were published as well. OSWER will
also use the data base to manage technology transfer information
collected during this project.
This new EPA data base, the Superfund Soil Data Management System
(DMS) is an important tool for fostering technology transfer
involving contaminated soil/ debris and sludge and relating the
information to applicable LDRs of HSWA which are applicable or
relevant and appropriate requirements (ARARs) to Superfund
actions. The Superfund Soil DMS will allow maximum utility of
the data obtained from any source. Data meeting a minimal
criterial will be included in the data base.
The data base construction allows for easy user access and
tailoring of reports to individuals' needs. Sorting will allow
questions concerning technology, waste characteristics/ soil
matrix and other parameters to be addressed.
6
11-326
-------
CO
ro
ORIGINAL SUPERFUND
SOIL TREATMENT
DATA BASE
ALTERNATE
TREATABILITY
VARIANCE
LEVELS
GUIDANCE
DOCUMENT &
OSWER DIRECTIVE
TREATMENT
TESTS
EXISTING
DATA
REVISED
DATA BASE
WEATHERED
SLUDGE
VARIABILITY
CSD
LDRs
TECHNOLOGY
TRANSFER
Figure 1. EPA OSWER Data Collection and Research Approach
-------
Outside access to the Superfund Soil DMS will be through a
central EPA system. At this time this is envisioned to be the
Agency's ATTIC System which is being managed at the Agency's
Edison NJ laboratory facility.
3.0 SOIL TREATMENT TESTS
The CSD Program reviewed existing data and identified
technologies that lacked treatment performance data, but would be
available technologies for treating CSD (Table 1). Twelve
treatment tests are planned. The technologies that will be
tested include slurry bioremediation, low temperature thermal
desorption, chemical extraction, soil washing, and stabilization
(Table 2). The technologies are applied to different types of
soils and wastes. For example, the biotreatment tests will be
conducted on three soil types. The soil classifications range
from sandy to clay type soils. In addition, different types of
wastes, including wastes high in PNAs, PCBs and metals, will be
tested. The stabilization technology will be tested as both a
primary technology and as a residual treatment.
The treatability tests will be conducted according to the OSW
Quality Assurance Project Plan for Characterization Sampling and
Treatment Tests Conducted for the Contaminated Soil and Debris
Program (QAPP) and site specific Sampling and Analysis Plans.
The individual sampling plans specify holding times, analytical
7
11-328
-------
CO
JHORKMEHATION
IMMOBILIZATION
BBCHLORJNATTON
TREATABIL1TY
GROUP
TECHNOLOGIES
NON-POLAR HALOGENATED
AROMATICS
(W01)
fCB*. HALOGENATED
D1OJONS, FVRANS, AND
THEIR PRECURSORS
(W02)
HALOGENATED PHENOLS.
CRBSOLS, AMINES, THIOLS,
AND OTHER POLAR
AROMATICS (W03)
HALOGENATED
ALIPHATIC COMPOUNDS
(W04)
HALOGENATED CYCLIC
AUPHATICS, ETHERS,
ESTERS, AND KETONES
(Wป5)
NITRATED COMPOUNDS
(W0ซ)
HETEROCYCUCSAND
SIMPLE NON-HALOGENATED
AROMATICS
(WOT)
POLYNUCLEAR
AROMATICS
(W0ป)
OTHERPOLAR
NON-HALOGENATED
ORGANIC COMPOUNDS
(W0ป)
NON-VOLATILE
METALS
(W10)
VOLATILE
METALS
(Wll)
HZ)
EXISTING DATA EXPECTED
CSftDTESTDATA
X
INDICATES TECHNOLOGY IS NOT EXPECTED TO BE EFFECTIVE
Table 1: Available Soil Treatment Technologies
-------
SITE
SOIL TYPE
CONTAMINANTS
TECHNOLOGY
Jennison-Wright
Jennison-Wright
Jennison-Wright
Bayou Bonfouca
Bayou Bonfouca
New Hampshire
Brown's Battery
Burlington Northern
Burlington Northern
Ninth Ave.
MIDCO
C&R Battery
Clayey
Clayey
Clayey
Silty
Silty
Silty
Silty
Silty, Sandy
Silty, Sandy
Sandy
Sandy
Sandy
Organics
Organics
Organics
Organics
Organics
Metals
Metals
Organics, Metals
Organics, Metals
Organics, Metals
Organics, Metals
Metals
Bioremediation
LTTD
Solvent Extraction
LTTD
Solvent Extraction
Soil Washing
Stabilization
Bioremediation
LTTD
Bioremediation
Sandy
Soil Washing
Table 2. Planned Treatment Tests
-------
methods, chain of custody, and quality control measures, such as
blanks and spikes. The tests will include measurements of
contaminant concentrations before and after treatment, and
measurements of the waste characteristics that affect the
performance of soil treatment technologies. Examples of waste
characteristics that affect treatment performance such as
moisture content, oxidation/reduction potential, and particle
size distribution are listed in the QAPjP.
4.0 DEBRIS
OSWER collected existing data on debris treatment in their data
collection program. The study determined that debris could
constitute as much as fifty percent of the contaminated media,
such as at a wood preserving site. The study also found that the
sampling procedures were not well documented. Recognizing the
importance of debris, the CSD Program has implemented a
comprehensive review of debris sampling, analysis and treatment.
The characteristics of debris that have been determined to affect
treatment include permeability and destructibility. The
potential treatment technologies have been generalized into three
categories for debris: 1) destruction, 2) extraction and removal,
and 3) sealing/solidification (Table 3). The Agency will discuss
the use of specified-technology standards for debris remediation
in an upcoming Advanced Notice of Proposed Rulemaking (ANPRM).
8
1-331
-------
"^^^^CONTAMINANT
DEBRK^^GROUPS
MATRICES ^^
PERMEABLE
DESTRUCTIBLE
PERMEABLE
NON-DESTRUCTIBLE
NON-PERMEABLE
DESTRUCTIBLE
NON-PERMEABLE
NON-DESTRUCTIBLE
ORGANICS *
(EXCEPT NITRATED
ORGANICS)
Destruction
Chemical Extraction,
Physical Removal,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Destruction,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Sealing/Solidification
NITRATED
COMPOUNDS
Destruction
Chemical Extraction,
Physical Removal
Chemical Extraction,
Physical Removal,
Destruction
Chemical Extraction,
Physical Removal
METALS
Chemical Extraction,
Physical Removal,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Sealing/Solidification
CYANIDE
Destruction
Chemical Extraction,
Physical Removal,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Destruction,
Sealing/Solidification
Chemical Extraction,
Physical Removal,
Sealing/Solidification
ft
ro
* Organics include volatile, acid extractable, and base neutral organics, pesticides, dioxins and PCBs
Table 3. Potential Management Practices for Debris Decontamination
-------
5.0 SLUDGE
The previous OSWER survey of Superfund sludge data found that
sludges are not consistently defined in the literature. However,
sludges, when identified, had higher concentrations of
contaminants than soils, and as a result, did not meet variance
level standards as frequently as soil. Of the OSWER survey data,
55% of the sludges treated met variance levels, while 78% of the
soils treated met variance levels. OSWER believes that to fully
characterize the treatment of sludge much additional work will be
required. To this end, OSWER, in conjunction with ORD, is fully
characterizing sludges from several hazardous waste sites on
sludge later this year. In addition, EPA is holding a symposium
during the summer of 1991 to broaden the background information
and share collective views on this topic. Additional information
on the symposium will be made available to any interested parties
by contacting the authors of this paper.
6.0 VARIABILITY
The OSWER study of Superfund soil treatability has found an order
of magnitude difference in treatability between remedy selection
testing and full scale treatment. As a result, treatability
tests must achieve an order of magnitude better treatment than
the standards in order to achieve compliance with the full scale
process. The factors that affect treatment effectiveness include
9
11-333
-------
mixing effectiveness, homogeneity of the soil matrix, feed
specifications, and contaminant concentrations. Variability of
the treatment results for the relatively homogeneous RCRA waste
streams has been accounted for by using classical statistics
which assume a less variable data set than Superfund soils.
EPA has begun a study to determine whether the soil matrix
presents unique problems in specific treatment methods and types
of wastes. EPA's study will use "clean" soils of similar
characteristics as the contaminated soil and artificially "mark"
the soil with a non-hazardous contaminant. Soils will then be
mixed and analyzed to determine the efficiency of mixing as a
treatment condition. The results of the study are expected to
show whether variability mixing effectiveness exists as a
function of soil type, equipment scale or moisture content, which
is representative of different treatment technologies. The
results of the study are not expected to conclusively show what
the variability function is or to allow for a direct correlation
into the LDR. Additional experimentation will be required to
assess the magnitude of the variability as it impacts on the
treatment standards for contaminated soil.
7.0 CONCLUSIONS
The current schedule provides for completion of data collection
and data analysis in the fall of 1991. We are soliciting
10
11-334
-------
existing treatment data, and new tests which may meet these
needs. We welcome comments on the above described program to
advance this study effort on soils, sludges, debris and
variability. If you have data, comments or questions regarding
the LDRs for contaminated Superfund soils and debris please
contact:
Richard Troast
Project Manager, CSD Program
703-308-8323
Carolyn K. Offutt
Chief, Special Projects
and Support Staff
703-308-8330
Hazardous Site Control Division (OS 220W)
U.S. Environmental Protection Agency
2800 Crystal Drive
Arlington, Virginia 22207
11
1-335
-------
8.0 References
1. Superfund LDR Guide #6A: Obtaining a Soil and Debris
Treatability Variance for Remedial Actions. 1989, Revised 1990.
Office of Solid Waste and Emergency Response, USEPA. Directive:
9347.3-06FS.
2. Superfund LDR Guide #6B: Obtaining a Soil and Debris
Treatability Variance for Removal Actions. September, 1990.
Office of Solid Waste and Emergency Response, USEPA. Directive:
9347.3-06BFS.
3. Quality Assurance Project Plan for Characterization Sampling
and Treatment Tests Conducted for the Contaminated Soil and
Debris (CSD) Program. November, 1990. Office of Solid Waste and
Emergency Response, USEPA.
12
11-336
-------
91 Certification Protocol for Meeting Organic Treatment Standards
for Incineration Ash
SEVENTH ANNUAL WASTE TESTING & QUALITY ASSURANCE
SYMPOSIUM
American Chemical Society
Washington, D. C.
July 8-12, 1991
William R. Schofield, PhD, PE, Schofield Environmental Associates, 1500 Marina Bay Drive, Suite
1612, Kemah, Texas, 77565( formerly Technical Manager, Chemical Waste Management, Inc., Texas
Facilities); John W. Kolopanis, Director, Technical Services, Chemical Waste Management Inc., 150
W. 137th Street, Riverdale, Illinois, 60627; Teresa S. Johnson, Area General Mgr., Chemical Waste
Management, Inc., 2700 N. S. 48th Street, Pompano Beach, Florida, 33073
11-337
-------
Certification Protocol for Meeting Organic Treatment Standards
for Incineration Ash
William R. Schofield. Technical Manager, Chemical Waste Management, Inc., P. O.
Box 2563, Port Arthur, Texas 77643-2563; John W. Kolopanis, Director, Technical
Services, Chemical Waste Management Inc., 150 W. 137th Street, Riverdale, Illinois,
60627; Teresa S. Johnson, Area General Mgr., Chemical Waste Management, Inc.,
2700 N. S. 48th Street, Pompano Beach, Florida, 33073
ABSTRACT
The Hazardous and Solid Waste Amendment of 1984 (HSWA) of the Resource
Conservation and Recovery Act (RCRA) requires the treatment of hazardous waste
to a specified treatment standard prior to land disposal. Testing to verify that
treatment residuals (i.e., incinerator ash and scrubber sludge/filter cake) meet
treatment standards is an expensive and time consuming process, especially for
commercial incinerators in which each batch of residuals has a different set of EPA
waste codes and consequently different treatment standards.
The challenge is to develop a testing approach or protocol which will simultaneously
provide: (1) a high level of assurance that treatment standards are being
consistently met while (2) holding testing and residual storage costs and testing
turnaround time at reasonable levels and (3) insuring that permitted residual storage
limits are met.
Chemical Waste Management has developed a practical testing protocol based on
EPA developed or supported concepts which is sufficiently flexible to fit the widely
varying incinerator facilities within our system. In concept, the universe of EPA
waste codes is divided into "treatability groups" based on chemical and physical
similarities. Each treatability group is then represented by one or more "indicator
waste code(s)" selected on the basis of: (1) treatment standard chemical species and
acceptance levels, (2) volume of waste with that code needing treatment, (3)
volatility and thermal stability of the chemical species present in the waste, (4)
matrices effects and (5) related issues.
A demonstration or "trial burn" is conducted in which the incinerator is operated
within an "operating envelope" and under a "quality assurance/quality control system"
1
11-338
-------
which will assure operation within the envelope.
If the incineration residuals meet treatment standards for the carefully selected
indicator codes, then compliance with treatment standards for the subject chemical
species are assumed for other codes within the treatability group which have
treatment standards ar or above the level detected on the demonstration test
residuals.
Spot tests are conducted on at least a quarterly basis to confirm continued
compliance. New demo burns are initiated whenever the operating envelope is
changed in a manner in which would be detrimental to the destruction or removal
of organics from residuals.
This approach has been successfully used with all three CWM incinerator facilities.
Other aspects of the protocol will be discussed in the presentation.
INTRODUCTION - DEFINITION OF THE PROBLEM
With the implementation of the HSWA (Hazardous and Solid Waste Amendment
of 1984) of the RCRA (Resource Conservation and Recovery Act) landbans and the
gradual elimination of the remaining variances, a large fraction of the hazardous
waste generated in the U. S. must be treated to meet stringent BDAT (Best
Demonstrated Available Technology) treatment standards before it can be land
disposed. For hazardous waste containing organics, sufficient treatment frequently
requires incineration to destroy the organic and cyanide compounds. The resulting
solid residuals (ash and scrubber cake) are'then stabilized to chemically immobilize
any regulated metals present prior to landfilling.
A great deal of effort has been invested in the hazardous waste management industry
to develop a practical and reliable method to verify that each treatment step has met
the HSWA treatment standards. In the case of organics contaminated waste, this is
frequently a two stage process: (1) verification that the incinerator ash and filter
cake meet organic and cyanide standards for all EPA waste codes present followed
by (2) verification that metal mobility or leachability has been sufficiently reduced
in the stabilization process to meet TCLP (Toxic Constituent Leaching Procedure)
limits for metals.
The focus of this paper is the certification protocol for organics in residuals from a
commercial hazardous waste incinerator; a simplified version of this protocol would
apply to captive incinerators. The protocol is described in toto; however, no attempt
will be made to cover every possible contingency which can occur when attempting
to satisfy, with total regulatory compliance, a program as exacting and complex as the
EPA landban regulations.
11-339
-------
The challenge in developing a practical and reliable protocol can be seen by
considering the following facts:
1. The EPA landban program presently contains a total of 1,467 variations of waste
codes, categories, and subcategories. There are typically 20 to 40 EPA waste codes
associated with a given batch of solid residuals from a commercial incinerator; these
codes can change dramatically with time. Consequently, the required treatment
standards which the residuals must meet also change with time.
2. EPA has established treatment standards for one or more organic/cyanide
compounds (target compounds) for each waste code which requires incineration.
(There is typically one target organic compound for U, P, and D codes, 5 to 12 for
each F and K code and from dozens of target compounds for F001-5 codes to
hundreds for F039 codes.)
3. Frequently the same target compound will appear associated with two or more
waste codes present in a batch of residuals and typically the treatment standard level
will vary from code to code even for the same target compound. Thus, the lowest
treatment standard present must be simultaneously met for every target compound
present before a batch of residuals can be certified as having met BDAT.
4. There is no on-line method of testing incineration residual for organics and
cyanides - one or more different extraction protocols must be completed on each
sample, typically, followed by multiple GC scans, GC/MS volatile and semi-volatile
scans and other testing depending on which EPA waste codes are present.
This an analogous situation to the use of EPA Modified Method 5 sampling of an
incinerator stack in a trial bum and subsequent extractions and analyses as a means
of verifying that the incinerator met the required minimum destruction and removal
efficiency (DRE) during the trial burn; thus, CWM as the permittee is authorized
to infer that the unit is meeting DRE requirements during subsequent commercial
operations as long as the unit is operating within the permitted operating envelope.
5. The sampling and analysis turnaround time for HSWA residual testing is
extremely slow, very complicated and disruptive to the residual management process
(e.g., one week, under ideal circumstances, to a more typical 30 to 60 days). In
addition, the cost is quite expensive (e.g., $3,000 - 10,000 per event).
6. HSWA requires the treater (in this case the incinerator owner/operator) to
"certify under penalty of law" to the land disposal facility where the ash will be
landfilled that the waste has met applicable organics treatment standards. Thus,
11-340
-------
boxes of ash can not be shipped until the certification can be completed.
7. A commercial incinerator typically generates 2-6 boxes of ash and other solid
residuals during a normal operating day and with larger volumes being frequently
generated during maintenance turnarounds. As a point of interest, the CWM Port
Arthur incinerator has the physical capacity and permitted authority to process
sufficient waste to produce up to 20 boxes of residuals per day.
A quick review of these facts indicates that if every box of residuals must be sampled
and analyzed for the applicable target compounds, this would result in ongoing
inventories of at least 50 - 100 boxes of uncertified residuals and would disrupt the
ability to manage residual in a timely and environmentally sound manner. In
addition, analysis of each box would result in analytical cost in the order of hundreds
of thousands to millions of dollars a year, diverting resources from areas that would
afford more protection to the environment.
DEVELOPMENT OF A CERTIFICATION PROTOCOL - CONCEPTS AND RATIONALE
The factors which affect the degree to which organics are destroyed on or vaporized
from solid residuals in a given hazardous waste (typically rotary kiln) incinerator
include (see Figure 1 for a schematic diagram of an incinerator process):
1. Residence time of the solid in the hot zone (i.e., rotary kiln length, waste loading
and RPM).
2. Temperature in the hot zone (i.e., kiln temperature).
3. Oxygen concentration (i.e., % excess air).
4. Degree of agitation of the organic contaminated solid (i.e., kiln RPM)
5. Solids loading (i.e., feed rates of solids bearing waste).
6. Volatility of organic compound (i.e., vapor pressure of target compounds).
7. Thermal stability (i.e., thermodynamic stability of target compounds).
8. Nature of solid substrate or matrix factor (i.e., Is the waste liquid, sludge or solid?
Is contamination a surface or depth phenomenon?).
1-341
-------
WASTE
WASTE,
WASTE,
ROTARY
KILN
AFTER
BURNER
ASH
STACK
RECYCLE LIQUID
QUENCH
JJ
ABSORBER
I
I
IONIZING
WET
SCRUBBER
SUSPENDED
SOLIDS
REMOVAL
I
FAN
FILTER CAKE
FIGURE 1 A schematic diagram of a typical hazardous
waste incinerator with a wet, acid gas absorber.
-------
The process variables (i.e, temperature, oxygen concentration, solids residence time,
loading, and agitation) are controlled directly or indirectly by the permit limits
and/or the incinerator design. The matrix factor is waste stream specific and
volatility and thermal stability is determined by the molecular structure of the target
compound. Thus, a protocol could be based on: (1) direct measurement and
control of the pertinent process variables while controlling (2) the waste streams
selected to provide a range of matrix types (i.e., solid organics, inorganic solids
contaminated by organics, particle size, liquids, etc., and (3) waste codes selected to
provide target compounds covering a range of volatilities and thermal stabilities.
A calibration test (or demonstration burn) would then be conducted which would
establish that the treatment (incinerator) system meets the treatment standards for
the codes present while operating within permit/design limits. The idea being that
an incinerator destroys organic compounds without regard to the waste codes
associated with those compounds.
This approach is quite analogous to the EPA trial burn approach to demonstrating
that a given incinerator will meet DRE requirements as long as the incinerator
operates within its permitted operating envelope.
In developing the RCRA/HSWA landban treatment standards EPA drew on several
concepts which are equally useful here. These concepts are: "treatability groups"
based on chemical and physical similarities among wastes with certain codes,
"transference of data" on treatment efficiency in an incinerator from waste code to
another code in the same treatability group, i.e., the use of one waste code as an
"indicator code" for other codes within the same treatability group. CWM has used
these concepts to develop a certification protocol.
The universe of all EPA waste codes was divided into 16 treatability groups along the
same chemical similarity lines EPA used in the RCRA/HSWA third third regulations
(see Table I for a listing of waste codes in each treatability group).
From each treatability group one (or more) code(s) were selected as indicator codes
based on the following criteria:
1. The number and type of target compounds for that code.
2. Treatment standard levels for that code.
3. The thermal stability and incinerability of the target compound(s) of that code
compared to the stability and incinerability of the target compounds of other codes
11-343
-------
TABLE 1 EPA waste codes1 assigned to each treatability group.
GROUP 1 - Solvents and Dioxin: P001-05, F020-23, P026-28
GROUP 2 - Halogenated Pesticides: D012-17, K032-34, K041, K097-98, P004, P037, P050-51, P059-60,
P123, U036, U060-61, U128-130, U142, U240, U247
GROUP 3 - Chlorobenzene: K042, K085, K105, UQ37, U070-72, U127, U183, U185, U207
GROUP 4 - Halogenated Phenolics: U039, U048, U081-82
GROUP 5 - Brominated Organics: U029-30, U066-68, U225
GROUP 6 - Miscellaneous Halogenated Organics: P024, U024-25, U027, U043,
U045, U047, U075, U121, U138, U158, U192
GROUP 7 - Aromatic & Other Hydrocarbons: U019, U220, U239
GROUP 8 - Polymiclear Aromatic Hydrocarbons: K001, K015, K022, K035, K048-52, K060, K087,
U005, U018, U022, U050-51, U063, U120, U137, U157, U165
GROUP 9 -Phenolics: P020, P047-48, U052, U101, U170, U188
GROUP 10 - Oxygenated Hydrocarbons & Heterocyclic: K023-24, K086, KO93-94, U002, U004, U028,
U031, U069, U088, U102, U107-08, U112, U117-18, U140, U159, U161-62, U190
GROUP 11 - Organo-Nitrogen Compounds: K011, K013-14, K083, K101-04, P069, P077, P101, U003,
U007, U009, U012, U105-06, Ulll, U152, U169, U172, U174, U179-81, U196
GROUP 12 - Halogenated Aliphatic: F025, K009-10.X016-21, K028-30, K073, K095-96, U043-44, U076-
80, U083-84, U131, U184-85, U208-11, U226-28, U243
GROUP 13 - Other Chlorinated Organics: P024, K043, K099
GROUP 14 - Organo-Sulfur Compounds: K036-38, K040, P039, P071, P089, P094, P097, U235
GROUP 15 - Pharmaceuticals: U141, U155, U187, U203
GROUP 16 - Cyanide: F006-12, F019, P013, P021, P029-30, P063, P074, P098-99, P104, P106, P121
1 Waste codes with BDAT specified technology of incineration (INCIN) are not listed.
1-344
-------
within that group.
4. The volatility of the target compounds of this code compared to the other codes
in the group.
5. The volume of waste with this code in the market place, i.e., the commercial
significance of the code compared to other codes in the group. In this and numerous
other ways CWM has attempted to conduct these tests in a manner which would
simulate routine, day to day operating conditions.
Once this list of indicator codes is developed one or more waste streams are selected
for each code based on commercial significance and matrix effects. Waste
inventories are collected and used in a trial burn type demonstration in which the
waste is burned in the treatment process under normal operating envelop conditions
and with a defensible QA/QC program in effect.
In the case of the CWM Port Arthur incinerator the incinerator is controlled by a
computer which continuously monitors all permit limited parameters. If a single
operating parameter moves outside of its permitted range or operating envelope, the
computer automatically discontinues waste feeds.
In planning the demonstration test, we generally will want wide ranges of thermal
stabilities, volatilities, matrix effects and treatment standard levels represented;
however, for the sake of minimizing uncertainty or risk, we have tended to choose
waste codes with less volatile, highly stable target compounds in solid substrates with
quite low treatment standard levels. (See Table II for the selected indicator codes
for each treatability group.)
The concept is that waste with the indicator codes will incinerate in a like manner
to other wastes within the same treatability group, i.e., we can transfer data
concerning how well the treatment process incinerated one waste code to the other
codes with chemical similarities.
Once the demonstration test is completed and the resulting residuals are carefully
sampled and analyzed (in triplicate), then waste codes from a represented treatability
group with treatment standards at or above the level of target compound(s) detected
in the demonstration test residuals can be certified as long as the process is operated
within the operating envelope and the QA/QC system is maintained. Conversely,
codes with treatment standards which are lower than the residual concentrations
found in the demonstration burn could not be certified without process adjustments
as needed followed by a successful new demonstration test.
8
11-345
-------
TABLE II Indicator code(s) selected for each treatability group.
TREATABILITY GROUP INDICATOR CODE(S)
1. SOLVENT & DIOXIN F001-5
2. HALOGENATED PESTICIDES U129, P123
3. CHLOROBENZENE U070, UO72
4. HALOGENATED PHENOLICS
5. BROMINATED ORGANICS U029
6. MISC. HALOGENATED ORGANICS (K019, F001, F002)1
7. AROMATIC & OTHER HYDROCARBONS U220
8. PNA HYDROCARBONS U165
9. PHENOLICS P020
10. OXYGENATED HC & HETEROCYCLIC U002, U069, U190
11. ORGANO-NTTROGEN K011, K013, U012
12. HALOGENATED ALIPHATICS K019, K020
13. OTHER CHLORINATED ORGANICS
14. ORGANO-SULFUR P039, P071, P089, P094
15. PHARMACEUTICALS
16. CYANIDES F0007, F008
11-346
-------
TABLE III Permit and other limitations which the 1990 operating envelope for the
CWM Port Arthur incinerator
HOURLY TOTALS
CHLORINE
SULFUR
ORGANIC HALOGEN
ASH (SCC)
TOTAL ORGANIC CONTENT
POLYCHLORINATED BIPHENYLS
WASTE AND FUEL FEEDS
ENERGETIC LIQUIDS (KILN)
ENERGETIC LIQUIDS (SCC)
ENERGETIC SLUDGE
NON-ENERGETIC SLUDGE
ENERGETIC SOLIDS
NON-ENERGETIC SOLIDS
AQUEOUS WASTE
MINIMUM KILN HEAT VALUE
MAXIMUM KILN HEAT VALUE
MINIMUM SCC HEAT VALUE
MAXIMUM SCC HEAT VALUE
MINIMUM TOTAL HEAT VALUE
MAXIMUM TOTAL HEAT VALUE
ACRYLAMIDE
CHLOROMETHYLMETHYLETHER
1,2-DIBROMO-S-CHLOROPROPANE
SYM-DICHLOROMETHYLETHER
DICTROTOPHOS
DIMETHYL CARBAMOYL CHLORIDE
DIPHENYLMETHANE DIISOCYANATE
ISOPROPYL MERCAPTAN
ISOPHORONE DIISOCYANATE
N-NITROSODIETHANOLAMINE
N-NITROSODIETHYLAMINE
PHOSPHINE
LEAD (FEED RATE LIMITS)
CADMIUM
VANADIUM
MERCURY
ARSENIC
BERYLLIUM
CHROMIUM
NICKEL
PERMIT LIMIT
1,690
250
1,352
240
20,000
3,172
50,270
3,000
8,900
5,300
10,000
3,000
41,475
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
(determined by DCS)
47.9 MM BTU/HR
79.5 MM BTU/HR
35.0 MM BTU/HR
77.2 MM BTU/HR
79.0 MM BTU/HR
150.0 MM BTU/HR
900
660
79
79
1,330
660
1,330
240
1,330
130
130
1,660
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
CWM OPERATIONAL LIMITATIONS
1,350.0
240.0
240.0
240.0
48.0
17.5
900.0
135.0
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
LBS/HR
1 Although K019, FD01, and PD02 are EPA waste codes which do not appear in treatability Group 6, these codes have
components in common with codes from Group 6. Since K019, F001, and F002 were fed during the demonstration test and the
residual values for the target compounds were found to be lower than all the treatment standard levels specified for these three
codes and for Group 6 codes, the common compounds could and have been used to certify Group 6.
10
1-347
-------
A SPECIFIC EXAMPLE - PORT ARTHUR INCINERATOR, 1990
Shortly after the third landbans became effective in 1990 a broad based
demonstration burn was conducted on the CWM Port Arthur (Texas) incinerator;
the results are summarized here as an example.
Appropriate streams could not be found for groups 4, 13, and 15. The test took
place over approximately 24 hours. Table III defines the operating envelope in effect
at the time of the test. Three samples were carefully collected to represent three
eight hour time periods, independently extracted and analyzed per SW846.
Organic concentrations were below the practical quantitation levels (QL) in almost
all cases; exceptions were:
MIN TREAT1
PQL STANDARD
TEST RESULTS
0.010
0.005
0.005
0.005
0.100
0.010
0.005
0.010
0.330
0.100
1.000
0.590
28.000
5.600
65.000
170.000
360.000
33.000
0.100
3.600
0.100
57.000
0.010
7.865
2
0.055
2
2
0.070
0.020
2
0.100
1.800
0.043
6.100
0.105
0.115
1.145
2
0.200
0.020
2
2
0.011
0.955
2
0.108
2
0.560
0.115
2
0.363
2
1.300
TARGET COMPOUND
ACETONE
TRICHLOROTRIFLUOROETHAN
CARBON TETRACHLORIDE
IODOMETHANE
ISOBUTYL ALCOHOL
ETHYL CYANIDE
TRICHLOROFLUOROMETHANE
DISULFOTON
DIS-N-BUTYL PHTHALATE
FAMPHUR
TOTAL CYANIDE
All numbers are expressed as TCA (rag/Kg) except Acetone which is expressed as TCLP (mg/L).
1 These are the lowest treatment standards within the landban program for these target compounds
regardless of waste code.
2 The results were below the practical quantitation level (PQL) of the analytical instruments used.
Inspection of these data indicates that this test was successful on all compounds of
the codes/treatability groups tested. Thus, based on these results the Port Arthur
facility was able to certify that ash generated with codes from the tested treatability
groups meets all applicable organic and cyanide treatment standards as long as the
unit is within the operating envelope and the QA/QC program is maintained.
An additional demonstration burn would be required to qualify codes/treatability
groups not covered in this initial effort or to requalify the treatment system if one or
more conditions were changed in a manner which could reduce the incinerator
system's ability to produce clean treatment residuals.
11
11-348
-------
92 ABSTRACT
Factors Affecting the Admissibility
and Weight of Environmental Data as Evidence
Jeffrey C. Worthington. Director of Quality Assurance; Kerri G. Luka, Audit
Programs Manager, TechLaw, Inc. 12600 West Colfax Avenue, Suite C-310,
Lakewood, Colorado 80215.
Many factors may affect the potential admissibiiity of environmental data in litigation.
These factors include but are not limited to:
0 Integrity of the sampling method
0 Integrity of the analytical method
0 Comparability to other sets of environmental data
0 Documented sample custody
0 Documented quality control results
0 Authenticity of the data
These same factors may also enter into the weight of the data as evidence. For example,
some data may include rigorous quality control including the use of performance
evaluation samples with each batch of samples from the field; other data may include less
rigorous quality control. The first set of data may be given greater weight by the trier of
fact.
The admissibiiity and weight of environmental data evidence may figure prominently into
pre-trial settlement discussions. Data is not often accepted at "face value". Litigants
usually need to address all the issues concerning the environmental data before proceeding
to other litigation matters.
The authors present a discussion of these factors and summarize several cases where the
admissibiiity and weight of the data as evidence were items of concern.
1-349
-------
REVIEW OF GROUNDWATER MONITORING REQUIREMENTS AT RCRA SITES
William G. Stelz. CPG. RCRA Enforcement Division, Office Of Waste
Programs Enforcement, U.S. Environmental Protection Agency, 401 M
Street S.W., Washington, D. C. 20460
ABSTRACT
This paper summarizes the groundwater monitoring requirements for
RCRA facilities under both interim status and operating permit
conditions. In addition, it highlights the major differences in
the regulations for facilities subject to both interim status as
well as permit requirements. Along with this overview, this
paper addresses how these regulations are enforced and what
mechanisms are set up to ensure that facilities are in compliance
with the groundwater requirements under the RCRA program.
INTRODUCTION
Subtitle C of the Resource Conservation and Recovery Act of 1976
(RCRA) regulates hazardous waste treatment, storage, and disposal
facilities (TSDFs) . Section 3004 of RCRA requires owners and
operators of hazardous waste TSDFs to comply with standards
established by EPA. Section 3005 provides for implementation of
these standards under permits issued to owners and operators by
EPA or authorized States. Section 3005 also provides that owners
and operators of existing facilities that comply with applicable
notice requirements may operate as " interim status" facilities
until a permit is issued or denied. Owners and operators of
interim status facilities also must comply with standards set
under Section 3004.
EPA promulgated regulations for permitted facilities in 1982 (47
FR 32274, July 26, 1982), codified in 40 CFR part 264, Subpart F
and 40 CFR part 270, Subpart B. These regulations establish
programs for protecting groundwater from releases of hazardous
wastes or constituents from treatment, storage, and disposal
units .
BASIC GROUNDWATER MONITORING REQUIREMENTS
The basic groundwater monitoring program under RCRA consists of
three main components: Interim status requirements, permit
application requirements and operating permit requirements (see
figure 1) . Each of these components contains specific
requirements and is designed to follow a sequence of applications
as a facility moves into different segments of the regulatory
process .
1-350
-------
INTERIM STATUS GROUNDWATER MONITORING REQUIREMENTS
On May 19, 1980, EPA promulgated comprehensive standards under 40
CFR part 265 for owners and operators of hazardous waste
treatment, storage, and disposal facilities (TSDF's) that qualify
for interim status. A facility owner or operator who has fully
complied with the requirements for interim status specified in
Section 3005(e) of RCRA and 40 CFR 270.70 may comply with the
part 265 regulations in lieu of part 264 pending final
disposition of the permit application. Part 265 Subpart F
contains groundwater monitoring requirements applicable to owners
and operators of interim status landfills, surface impoundments,
and land treatment facilities. The goal of the interim status
groundwater monitoring program is to evaluate the impact that the
facility may have on the uppermost aquifer underlying the site.
The regulations establish a two-stage groundwater monitoring
program designed to detect and characterize the migration of any
wastes that may have contaminated the groundwater. Stage I
consists of a detection monitoring phase where the objective is
to determine if hazardous wastes have leached into the uppermost
aquifer in quantities sufficient to cause a significant change in
groundwater quality. Stage II is an assessment monitoring phase
that is initiated when a significant change in water quality has
been detected at a hazardous waste facility and contamination is
suspected. The assessment monitoring program is directed at
characterizing the rate and extent of contaminant migration.
Assessment monitoring under Section 265 entails a determination
of both the vertical and horizontal concentration profiles of all
hazardous waste constituents in the plume(s) that escape from the
hazardous waste management areas. Figures 2, 3 and 4 outline the
major features of interim status groundwater monitoring.
PART 270 - PERMIT REQUIREMENTS
Part 270.14(c) establishes permit application requirements (Part
B), that an owner/operator must submit in order for EPA to
determine if the facility is in compliance with the part 264
standards. Part 270.14(c) requires the applicant to establish
the nature of the facility's impact on the groundwater, as well
as the hydrogeologic characteristics of the site's subsurface and
the extent of the waste management area.
OPERATING PERMIT REQUIREMENTS
The part 264 Subpart F groundwater monitoring requirements apply
to owner/operators that treat, store, and or dispose of hazardous
waste in surface impoundments, waste piles, land treatment units,
or landfills that receive waste after July 26, 1982. Such units
are referred to as "regulated units." These requirements are
11-351
-------
effective only at facilities that have failed to qualify for
interim status (see Part 270). An interim status land TSD
facility would not be subject to part 264 standards until a
permit is issued to that facility and unless waste was received
after July 26, 1982.
Groundwater Monitoring at permitted facilities has three phases:
1. Detection - of indicator parameters, waste constituents, or
reaction by-products in the uppermost aquifer.
2. Compliance Monitoring - to better define the extent of aquifer
contamination by identifying which hazardous waste constituents
are present in the groundwater and by describing the shape and
concentration of the contaminant plume
3. Corrective Action - to remove hazardous waste constituents
from the groundwater or to treat them in place.
Typically, a facility employs a detection monitoring program
until there's a statistically significant increase in that
program's parameters or constituents, after which a compliance
monitoring program begins. If there's a statistically
significant increase in the concentrations established in the
compliance monitoring program, i.e., if the groundwater
protection standards have been exceeded, the facility must enter
a corrective action program. However, a facility need not begin
with detection monitoring - if there is existing evidence of
groundwater contamination (such as from an interim status
monitoring program), the facility can be put directly into a
compliance or corrective action program when the facility's
permit is issued. Figures 5 and 6 illustrate the main components
of groundwater monitoring for facilities with operating permits.
SUMMARY
The part 264 groundwater monitoring standards differ from those
in part 265 in that the part 264 standards are more flexible and
go beyond just contaminant assessment and allow for corrective
action to be directly incorporated; whereas under interim status,
corrective action has to be achieved via another mechanism such
as from an enforcement order (e.g., a 3008(h)).
Instead of testing for specific parameters as in part 265, part
264 requires the Regional Administrator to specify parameters and
hazardous waste constituents to be monitored on a site-by-site
basis. In each phase of the groundwater monitoring program under
part 264, the number, depth, and location of wells must yield
representative samples of groundwater. In addition, under part
264, a groundwater protection standard is set up for each
constituent found in the groundwater, and if exceeded, corrective
action is initiated. Figure 7 summarizes the various options for
groundwater monitoring for land disposal facilities.
1-352
-------
REFERENCES
Code of Regulations, Protection of Environment, 40, parts 190 to
399, revised July 1, 1990.
Solid Waste Disposal Act, as amended by the Resource Conservation
and Recovery Act of 1976, as amended by the Hazardous and Solid
Waste Amendments of 1984, (42 U.S.C. 6905, 6912(a), 6921, 6924,
6925, and 6935).
U.S. Environmental Protection Agency. 1986. RCRA Ground-Water
Monitoring Technical Enforcement Guidance Document. OSWER-9950.1
11-353
-------
Basic Groundwater Monitoring Requirements
CO
Groundwater monitoring requirements include:
Interim status requirements
(40 CFR Part 265 Subpart F)
Permit application requirements for groundwater monitoring
(40 CFR Part 270.14 (c) Part B)
Operating permit requirements
(40 CFR Part 264 Subpart F)
-------
Interim Status Groundwater Monitoring
CO
en
01
Groundwater monitoring program includes:
Monitoring wells
1 hydraulically upgradient
3 downgradient
Sampling/analysis plan
Statistical comparison test
Outline of groundwater quality
assessment program
(Based on regulations at 40 CFR 265.91)
-------
Interim Status
Groundwater Monitoring (cont'd)
Is Contamination Present During Interim Status?
NO
en
Continue monitoring
program until
closure or permit
application
issuance.
YES
Implement groundwater
assessment program as
stipulated by submitted
outline.
Continue to make
assessment quarterly
until closure or permit
issuance.
(Based on regulations at 40 CFR 265.90 through 265.94)
-------
Monitoring Over Time
Quarterly monitoring for:
Drinking water standards
Groundwater quality parameters
Indicator parameters
Water elevations
Semiannual monitoring for:
Indicator parameters
Water elevations
Annual monitoring for:
Groundwater quality parameters
(Based on regulations at 40 CFR 265.90 through 265.94)
-------
Groundwater Monitoring During Permit
Groundwater monitoring program requirements:
Specify the point of compliance
Sufficient wells properly located to yield both
- background groundwater quality and
- water quality passing the point of compliance
Consistent sampling/analysis procedures
Determination of groundwater elevations during all
sampling periods
Background groundwater quality
Statistical comparison procedure
-------
Potential Results of Monitoring
Groundwater monitoring program requirements
CO
en
CD
No contamination
Hazardous
constituents
detected at point
of compliance
Hazardous
constituents
detected at point
of compliance
and downgradient
of facility boundary -
exceeding the
groundwater
protection standard.
Detection
monitoring
(40 CFR 264.98)
Compliance
monitoring
(40 CFR 264.99)
Corrective
Action
(40 CFR 264.100)
-------
Options for LDFs
INTERIM
STATUS
FACILITY
Operating Permit
No release to
groundwater
Detection
Monitoring
GROUND-WATER MONITORING
Closure
Release In excess of the groundwater
protection standard at or beyond the
point of compliance
Detection monitoring
(I.e., with no groundwater
contamination)
Assessment monitoring
(l.e.,wlth groundwater
contamination)
Corrective
Action Monitoring
Clean Closure
Release in excess of background at
the point of compliance
NO MONITORING
Closure With Waste
In Place:
Post-Closure
Requirements
30 YEARS OF MONITORING
1
Compliance
Monitoring
i
Post-Closure
Permit
-------
94 ABSTRACT
The Paperless Environmental Laboratory:
A Plan for Realization
Jeffrey C. Worthinyton. Director of Quality Assurance; George A, Duba,
PhD, Vice President; TechLaw, Inc., 12600 West Coifax, Suite C-310,
Lakcwoo<:>- Colorado 8021S.
Many laboratories are buying, installing, or modifying their current Laboratory Information
Management Systems (LfMS) to produce all the documents necessary to effect the smooth
flow of samples through the laboratory. Laboratory managers and analysts most
comfortable with keyboards hope to make all paper disappear on the work bench by using
direct data entry.
Users of data from environmental laboratories often include attorneys who may need to
demonstrate sample custody and integrity of the sample data in order to admit the
information into court. These data users are less than comfortable seeing hand-written
documents disappear from the laboratory to be replaced by electronic records.
% author presents guidelines for the development of a paperless laboratory system. The
ฃ. ""nes include consideration for laboratory management issues and litigation related
1-361
-------
95 DATA MANAGEMENT ISSUES
IN THE
HAZARDOUS WASTE INDUSTRY
Gerald Austiff. CRM, Technical Records Archivist, Marty Cahill, Manager-Waste
Analysis Plans Group, John Krecisz, Technical Manager-Incineration, Chemical
Waste Management, 150 West 137th Street, Riverdale, IL 60627
ABSTRACT
One of the most direct and yet unaddressed consequences of increased Federal,
State, and Local Government regulation of U.S. Industry in the later half of the
twentieth century has been the added responsibility of creating those documents and
data necessary to verify compliance with these regulations. For an industry such as
hazardous waste management, the responsibilities of mandatory records creation
have proven to be especially great.
What has not necessarily followed, however, is the development of records and data
management systems proportional to the importance that information serves in the
operation of a hazardous waste facility. In today's business climate, however, the
opposite is equally true-the lack of management attention to the records and data
that is routinely produced by the organization can cost plenty, both in terms of
dollars and in reduced productivity.
This paper will address the data management issues facing every company in the
hazardous waste industry and outline a records management strategy that such
companies must consider not only to avoid costly fines/penalties, but to turn their
records and data into a positive asset.
INTRODUCTION
As one of the most heavily-regulated sectors of the world economy, the hazardous
waste industry has many specific and long-term records/datamanagement
requirements which must be met in order to be allowed to continue to conduct
business in its operating jurisdiction. The ability to create those documents and data
required for waste profiling, analyses of waste samples, and facility operation has
been greatly enhanced by sophisticated laboratory equipment and the
computerization of manual recordkeeping practices in general. This development
has not, however, resulted in an equal ability to provide long-term protection and
11-362
-------
retrievability of this information. To the contrary, the burden to keep analytical
records and data for the time periods required by facility permits and government
rules has become even more challenging and elevates the task of records
management to a higher management priority that it may have been raised to in
the past.
RETENTION REQUIREMENTS
All hazardous waste facilities are subject to a number of Federal and State
Regulations which impact recordkeeping and data management. Some of the most
important recordkeeping regulations are:
RCRA
40CFR, Parts 264.16 and 265.16
- Requires that Training Records on current personnel must be kept until
264.16&265.16 closure of the facility.
40CFR, Parts 164.73 and 165.73
- Requires the owner or operator to keep the written operating record at his
facility until closure. Monitoring data at disposal facilities must be kept
throughout the post-closure period.
40CFR, Part 265.94
- The owner or operator (for ground water monitoring purposes) must keep
records of required analyses throughout the active life of the facility, and,
for disposal facilities, throughout the post-closure care period as well.
TSCA
40CFR, Part 761.180 (Subpart J)
- Documents that include the; dates, ID of facility & owner of Facility from
whom whom PCBs were received, Dates of PCB disposal or transfer,
summary of total weight of PCBs, and total number of PCB articles
received or transferred for 5 years after the facility is no longer used for the
storage or disposal of PCBs (Chemical landfills must keep this documentation
at least 20 years after the landfill is no longer used for the disposal of PCBs.
Incineration facilities must collect and maintain data on PCB incineration
rates & quantities, combustion temperatures, stack emissions, monitoring
data for 5 years from the date of collection.
11-363
-------
Page 3
40CFR, Part 761.180(a)
- Requires that owners or operators of chemical waste landfills collect and
maintain water analyses & operating records for at least 20 years after the
landfill is no longer used for the disposal of PCBs.
OSHA
29CFR, Part 1904.6
- Requires that "...logs and summaries of occupational injuries, supplemental
records of each occupational injury, and annual summaries of injuries ... be
maintained for 5 years following the end of the year to which they relate."
Failure to maintain these records shall be punished by a fine of not more
than $10,000, or by imprisonment, for not more than 6 months, or both. (29
CFR Part 1904.9)
ACCOUNTABILITY
For the hazardous waste facility, recordkeeping and data management is clearly a
long-term responsibility that, if neglected, will result in substantial fines. An analysis
of administrative actions initiated against regulated facilities by the U.S.
Environmental Protection Agency reveals that from the period 1972-1989 there were
12,250 actions which resulted in over $105 million in penalties where inadequate
recordkeeping was cited as one of the major violations.1
To illustrate the degree to which the recordkeeping practices of a hazardous waste
facility can be held accountable by Federal Regulators, consider the activities of the
National Enforcement Investigations Center (NEIC) which provides the U.S.
Environmental Protection Agency's Office of Legal and Enforcement Council with
technical information and evidence in support of potential enforcement actions on
a site's violations of permit conditions or federal regulation.
The scope of an NEIC investigation will generally involve the request to have access
to all records maintained at the facility. The NEIC project team will gain consent
to enter the facility from the owner or operator and schedule a date for the
FY 1989 Enforcement Accomplishments Report. U.S. Environmental Protection Agency, Office of
Enforcement, Compliance Evaluation Branch.
11-364
-------
Page 4
investigation to begin. NEIC actions can be of 2-3 weeks in duration and usually
require the assistance of facility management staff to obtain requested records and
provide additional information. Records/data requested generally originate with
randomly selected Waste Manifest Files. The NEIC team will request all related
paperwork and data associated with select Manifests. Objectives of this request are
to:
1) Track the movement of waste streams through the facility
2) Verify that all documentation is traceable to original manifest, and is
logically filed and retrievable.
Related paperwork that must be produced for the NEIC investigators includes the
operational records (weight tickets, time & date stamps, logbook pages, records
which detail the movement of the waste, charts from emission monitoring for
incinerators, location of drums), the laboratory data (analytical raw data, instrument
readouts, result summaries, log books, QA/QC checks, and QC tests),
and residue management records (for incinerators).
In order for a facility to successfully met the demands of an NEIC investigation and
provide timely retrieval of requested records, the facility's recordkeeping system
must be in order to demonstrate to regulators that it is in compliance with the
recordkeeping requirements of the Code of Federal Regulations and their operating
permit. It would be extremely damaging for the facility to be unable to produce a
complete tracking record for a Waste Manifest- with the result being additional fines
and disruption of normal activities. Beyond this specific example, try to imagine the
impacts to a company's operation if a body of records and data were lost due to
fire, flood, theft, or slow deterioration in poor storage conditions. It is for these
reasons that a systematic- proactive approach has been developed by many
companies to provide protection to critical records and data. The need for such an
approach for a hazardous waste TSD facility is no less important.
DATA MANAGEMENT ISSUES FOR HAZARDOUS WASTE FACILITIES
The objectives of a records/data management program are simply stated to:
1) Furnish accurate and complete information when it is required
11-365
-------
PageS
to manage and operate the organization efficiently.
2) Process recorded information as efficiently as possible.
3) Render maximum service to the customer(user of the records)2
In addition to these goals, the organization must also ensure that the records and
data they have on hand will be admissible in court to defend their actions/decisions
which may have occurred much earlier in time. The existence of a record
management program satisfies the Uniform Rules of Evidence requirement that a
process be in place to produce an accurate result and that the records created by
the organization are trustworthy.3 The organization's records management program
must have written procedures, training, and regular audits in order to demonstrate
that the organization carefully developed its records program, that staff was fully
aware of the recordkeeping requirements, and that the procedures were actually
followed by organization's staff.
The components of a records management program for a hazardous waste facility
must take into account the following requirements:
1) Long term retention of data (over 30 years),
2) Ability to retrieve analytical records and data based on waste manifest
numbers, customer profile IDs, Dates of tests,
3) Timely responses for customer requests for information to decision waste
streams, and to recertify wastes for final disposition.
The requirement to maintain facility operating records and analytical data for such
long periods of time makes it difficult to rely exclusively on a paper-based system
of recordkeeping to stay in compliance with federal regulation. First, paper simply
will not last as long as the law requires. Secondly, paper-based recordkeeping
2
Information and Records Management, Robek, Brown^nd Maedke, Glencoe Publishing Co. 1987
Donald S. Skupsky, Legal Requirements For Microfilm, Computer and Optical Disk Records, Information
Requirements Clearinghouse 1991.
11-366
-------
Page 6
systems take up a large amount of space-space that is limited or unavailable in a
laboratory environment. Lastly, paper-based systems provide no information
protection to the organization in instances of fires, floods or misfiling of data. It is
for these reasons that a forward-thinking records management program should plan
and provide for the long-term preservation and security of analytical data by
replacing (or reducing) the organization's dependance on paper-based systems with
other media.
In those instances where it is impossible to incorporate microfilm or image scanning
technology, the organization must provide for controlled climate, secure storage that
meets federal guidelines for fire protection.
LABORATORIES
Records management responsibilities for any organization presents a significant
challenge to management. In the hazardous waste industry, however, the presence
of laboratories dictates an even higher degree of data complexity that the program
must address. The sophistication of modern-day laboratory instruments and their
ability to produce data means that the records program will have to take into
account many forms of output that become part of the analytical record of the waste
disposal decision. Examples of different media produced by laboratories are:
1) Perkin Elmer 5000 writes data on 5 1/4" floppy disks
2) Jarell Ash ICAP writes data on RLO1K-DC disk packs or 158 mb tape
cassettes
3) Leeman ICAP writes data on paper tapes
4) Hg CV Instrument produces data compilations on thermal paper.
5) INCOS Mass Spectrometry Instruments backup data onto 45 mb tape
cassettes w/ IDOS as the system's operating system.
6) Parr Bomb Calorimeter Instrument produces data in the form of 3 and 1/2"
paper rolls.
7) Logbooks which provide indexes to the computer media noted above.
11-367
-------
Page 7
The presence of such varied forms of information media dictates two key
components of data management strategy for a laboratory environment. One, the
nature of electronic media enables the creator to routinely back-up or duplicate data
from the instrument's hard disk onto a portable media such as floppy disks and
transfer the back-up media to an off-site storage facility or on-site vault designed
for computer media. Secondly, although it is relatively easy to create back-ups, the
media itself has never been considered an archival storage media and is subject to
data loss over a period of time. It is for this reason that a sound data management
policy must provide for periodic conversions of data from old media to new media
to arrest any possible loss of information due to the age of the original magnetic
storage device. The greatest threat to the retrievability of electronic data, however,
is neither a physical calamity or human error. The greatest challenge to maintaining
control of electronic media is the continual hardware/software technology migration
that the computer industry is subject to. New hardware means different size tape
drives and new operating systems. New software releases are not automatically
compatible with earlier versions. For information that must be maintained and
made available for periods in excess of 30 years, the organization can certainly
expect to have a significantly different computer hardware configuration and new
software requirements than originally where put into place.
In order to ensure that data remains accessible to the organization, the persons
responsible for data management must rigorously review the impacts of new
hardware and software purchases on data recovery and make necessary conversions
before the old equipment leaves the site.
Another major issue facing any organization which desires to develop a data
management program is in the selection of media to provide long-term protection
for their information. Storage space reductions, time to access files, admissability
of media in legal actions, cost, longevity of media, and the type of information
being recorded are all factors with varying degrees of priority for different
organizations. For the hazardous waste industry, however, primary consideration
should be given to the media type which satisfies its need to keep information
secure for the required periods of retention. As previously noted, paper-based
systems are vulnerable to natural disasters, take an increasingly larger amount of
space away from staff and equipment, are subject to misfiles, and, perhaps worst
of all, as soon as the file leaves the desk of the user, represents a loss of staff time
to retrieve the file for reference purposes.
1-368
-------
PageS
comparative analysis of information media types follows:
Media Type
Disadvantages
Advantages
Paper -Traditionally preferred in
legal actions
-Requires no special viewing
device
-Comfort level of users is high
-Requires large storage area
-Deteriorates over time
-Easy to misfile
-Can only be indexed one way
Microfilm -All courts & governmental -The equipment needed to read
agencies will accept as evidence film is bulky and must be in
-Recognized by ANSI as an
archival media (silver based
film will last 200 years)
-Reduces storage space by 95%
a common access area
-Film (without computer aided
computer aided software) is
slow to load and retrieve the
the desired image.
-The hard copy record must be
sent off site to be photographed
and processed before the film is
available. Information is not
available for this period of time.
Optical Disk -Provides the fastest, most flexible -Admissibility in court not
Imaging access to documents (files can be
Technology indexed numerous ways)
-With the existence of a PC or
terminal on a desk the information
is sent to the user in seconds.
-As with traditional computer media,
optical disks can be easily dupli-
cated for offsite security
-If indexed properly, it is
impossible to misfile or lose
a file
established
-No industry standards
1-369
-------
Page 9
-One 12" Optical Disk can hold up to
260,000 documents (118 cu. ft. or
15 legal sized file cabinets)
The media that the organization selects to establish a records management program
will have to be evaluated against these characteristics-with the ultimate decision
based upon its own particular concerns. The rate at which companies that are
currently receiving and maintaining great amount of information are installing
imaging systems, however, demonstrates that its ability to send information
directly to users in very short periods of time establishes it as the office technology
of the future. The U.S. Environmental Protection Agency has issued a position
relative to this technology-stating that it is permissible to maintain compliance
information on electronic imaging systems, but due to the lack of industry standards
on optical disk technology, recommends that the original paper records also be
maintained. The legal admissibility and industry standards concerns are being
addressed at the present time and will soon not be obstacles in evaluating the
suitability of this technology for an organization in the hazardous waste field.
ESTABLISHING THE RECORDS MANAGEMENT PROGRAM
A records & data management program for an organization in the hazardous waste
industry must have as its principle objectives the following:
1) The protection and security of analytical data created in support of
waste treatment and disposal decisions.
2) The maintenance of the facility operating record and the ability to
track waste streams & verify that all permit requirements have been
satisfied in treatment and final disposition.
3) The ability for staff to quickly access records and data to respond to
customer or regulator inquiries.
The facility must establish a written procedure or program for the management of
its official records and data. The first place to start is by conducting a records
survey (inventory of facility records). The survey will identify the number of
different record groups, the volume(# of file drawers or storage boxes), the current
1-370
-------
Page 10
media(paper, magnetic tape, etc.), the present location(s) of records/data, the
interrelationships between different facility groups in creating records, and how the
records and data are used by the facility. After completion of the initial survey, the
person responsible must research and establish the legal retention requirements and
any existing company policies which will determine how long the records must be
maintained.
The survey when in final form should be reviewed by the company's upper
management for approval and designation as official company policy in regard to
recordkeeping requirements. The survey is now the facility's official records
retention schedule and is a key component of the requirement to have a written
plan or program in place to insure that the facility's records are deemed
"trustworthy." The facility's recordkeeping program will also require the
development of a corporate-wide directive or procedure which establishes the
standards that must be followed in maintaining those records related to disposal
decisions, waste receiving, processing, disbursement of waste product, supporting
analytical data, and quality assurance.
The recordkeeping policy should address the following topics:
1) Permit Requirements-the policy must include a statement that all
record and data will be maintained in accordance with facility permit
conditions.
2) Retrieval Requirements-the policy must make clear the requirement
to keep records in sufficient detail in order to be able to retrieve
analytical data and quality assurance records for individual waste
samples and manifests for the duration of the records retention
period.
3) Records Storage Area-the policy should state that all records be
maintained in secure storage under conditions that will prevent
deterioration of the information for the duration of the retention
period4. The conditions for storage could be those established by the
4 Good Automated Laboratory Practices. Office of Information Resource Management, U.S. Environmental
Protection Agency.
1-371
-------
Page 11
National Archives and Records Service5, for electronic media the
storage facility standards outlined by the National Bureau of
Standards in Special Publication 500-101 "Care and Handling of
Computer Magnetic Storage Media," and the standards set by the
National Fire Protection Association publications NFPA
232AM "Archives and Records Centers" and NFPA 232 "Protection
of Records."
4) Analytical Records-the policy must state that analytical data must be
logically filed, manual entries be made in permanent, reproducible
ink, must be dated and signed or initialed by the technician. Any
changes to the data must be crossed out with a single line, dated and
initialed, with no obscuring of the corrected data.
5) Logbooks-must be used whenever information cannot be recorded on
the analytical data, loose paper must be permanently affixed to the
logbook & dated/initiated.
6) OA Records-must be retained as outlined by the facility operating
permit and company policy.
Once the records survey and the Standard Operating Procedure have been
completed, the person responsible for the program must investigate the use of
microfilm or optical disk technologies and the suitability of each for converting the
facility's data/records into a media that will provide for longevity and security of
analytical information. Either technology will reduce storage space requirements,
allow the duplication and off-site storage of back-up documentation, and insure the
integrity of information for the terms of the retention periods. The ability of optical
disk technology to provide rapid access to detailed inquiries, however, has
established this information media as the preferred method of managing large
amounts of data for those organizations which wish to remain responsive to their
clients and have a significant recordkeeping burden.
A model of a typical imaging system follows:
Center Operations Division, Office of Federal Records Centers, National Archives and Records Service,
General Services Administration(August 1976)
11-372
-------
Page 12
Manage/Store
Recommended Configuration
for
Chemical Waste Management, Inc.
1-373
-------
Page 13
For a waste disposal decision process, an organization could electronically scan the
Generator's Waste Profile Sheet, all analytical test results, and the Disposal
Decision. An electronic folder would be created and indexed by; Customer Profile
Number, Customer Name, Laboratory Sample ID Number, and storage box # for
the original paper documents. The electronic file could now be retrieved by any of
the index keys in a matter of seconds. The original paper file could be transferred
to archival storage as the electronic data should satisfy all subsequent information
needs. The data written on the imaging system's optical disk is routinely duplicated
on a separate disk for security purposes and stored in an off-site location. Customer
service is enhanced when a customer requests a copy of an entire file or only a
specific document, the imaging system has the capability to transmit a facsimile
directly from the provider's PC to the requestor's facsimile transmission device. If
Waste Decision Files are indexed by date of decision, the user group would have the
ability to retrieve all files prior to the expiration of the original decision in order to
recertify waste streams for disposal.
The effect of changes in governmental regulation might dictate the reconsideration
of a number of Waste Decision Files and imaging technology has the capability to
retrieve electronic files by "key word" searches. For disposal sites, having imaging
technology would enable the facility to index all required documents to the original
waste manifest number. Raw Data if it is maintained separately from the rest of
the Decision File could be indexed to the Customer's name and the Testing
Laboratory's sample control number system. With this technology, an organization
could truly have control of their files and put their information to work for them.
CONCLUSION
Proper data management techniques for a hazardous waste facility must be based
on the realization that required recordkeeping is not only the obligation to create
certain forms and data, but that this information must be retrievable for the
duration of legal periods of retention. This requirement dictates that the facility
apply a systematic approach to record/data creation, active use, and long-term
storage. The belief that the filing of a record in a file cabinet or storage box has
provided adequate protection to the company's interests has been the source of
much later grief and unnecessary expense. As is the case with any regulated
industry, the data that is maintained for compliance purposes is (and will be for
long periods of time) an extremely valuable asset to a company and requires the
11-374
-------
Page 14
implementation of a program to guarantee the longevity of the records/data not
only for compliance reasons, but also to become a positive asset in company
operations.
11-375
-------
AIR/GROUNDWATER
-------
95 NEW DIRECTIONS IN RCRA GROUND-WATER
MONITORING REGULATIONS
James R. Brown. Environmental Scientist, Vernon B. Myers, Chief, Monitoring and
Technology Section, Office of Solid Waste (OS-341), U.S. Environmental Protection
Agency, Washington, D.C., 20460; Ann E. Johnson, Hydrogeologist, Regulatory Analysis
and Support Division, Science Applications International Corporation, McLean, Virginia
22102
ABSTRACT
EPA soon expects to issue a Notice of Proposed Rulemaking (NPRM) in the Federal Register concerning
amendments to the ground-water monitoring requirements for land-based hazardous waste treatment, storage,
and disposal facilities (TSDFs) that are regulated under Subtitle C of the Resource Conservation and Recovery
Act (RCRA). The notice will propose amendments to the list of ground-water monitoring constituents for
TSDFs, Appendix IX to Title 40, Code of Federal Regulations. Part 264, ("Appendix IX"), and require that
certain procedures be used in the design, installation, and operation of ground-water monitoring systems at
TSDFs. The proposed changes to Appendix IX include the addition of a required list of detection monitoring
analytes (Appendix IX-A), the deletion or addition of several Appendix IX compounds due to analytical
considerations, and a site-specific variance from the annual Appendix IX analysis requirement during compliance
monitoring. The proposed standards for ground-water monitoring procedures will be specified in revisions to
Chapter Eleven of the U.S. EPA document SW-846, Test Methods for Evaluating Solid Waste,
Physical/Chemical Methods," (Third Edition), or more generally referred to as "Chapter Eleven of SW-846."
Chapter Eleven of SW-846 specifies requirements concerning the characterization of site hydrogeology, placement
of detection monitoring wells, monitoring well design and construction, and ground-water sampling and analysis
programs. All hydrogeologic investigations and monitoring activities must comply with the methods and
procedures required in Chapter Eleven of SW-846.
The proposed requirements in Chapter Eleven of SW-846 represent EPA's establishment of qualitative data
quality objectives (DQOs) for the RCRA ground-water monitoring program. At the same time, EPA's
Environmental Monitoring Systems Laboratory in Las Vegas, Nevada (EMSL-LV) is in the process of evaluating
the efficacy of establishing quantitative DQOs for ground-water monitoring system performance. If appropriate,
quantitative DQOs would allow EPA to determine the minimum number and location of monitoring wells
required to achieve a specified probability of leak detection. This paper will summarize some of the methods
that have been investigated to establish quantitative DQOs for RCRA ground-water monitoring.
INTRODUCTION
Subtitle C of RCRA creates a comprehensive program for the safe management of
hazardous waste. Owners and operators of facilities that treat, store or dispose of hazardous
waste must comply with standards established by EPA that are "necessary to protect human
health and the environment." Implementation of these standards occurs through permits
issued to owners and operators by EPA or authorized States.
Standards for protecting ground water from releases of hazardous wastes from permitted
TSDFs were promulgated by EPA in 1982 (47 FR 32274; July 26, 1982), and are codified
1-379
-------
at 40 CFR Part 264, Subpart F ("Subpart F1). Subpart F requires TSDF owners and
operators to characterize their site's hydrogeology, install and maintain a ground-water
monitoring system, and to sample and analyze ground water at specific intervals to
determine whether hazardous wastes or hazardous waste constituents from the facility are
contaminating ground water.
The Subpart F requirements consist of a three-phase ground-water monitoring program:
detection monitoring, compliance monitoring, and corrective action. The first phase,
detection monitoring, involves at least semi-annual monitoring of "indicator" parameters,
waste constituents, or reaction products specified in the facility permit that provide a reliable
indication of the presence of hazardous constituents in ground water. Owners and operators
employ detection monitoring at new land disposal facilities and at land disposal facilities not
believed to be releasing contaminants to ground water. If monitoring indicates that the
concentration of a monitored constituent has shown a statistically significant increase over
background concentrations, then EPA requires analysis for all Appendix IX constituents and
the facility enters compliance monitoring.
Compliance monitoring, the second phase of ground-water monitoring, requires at least
semi-annual monitoring for constituents identified in the facility permit, including those
constituents detected in ground-water during the detection monitoring program. A facility
in compliance monitoring must also monitor ground water for all Appendix IX constituents
at least annually and report the concentration of any new constituent detected to the
Regional Administrator. All detected Appendix IX constituents are then monitored at least
semi-annually during compliance monitoring. The concentrations of all compliance
monitoring constituents are compared to concentration limits specified in the facility's
permit. Concentration limits are an element of the facility's ground-water protection
standard used to determine if ground-water contamination has occurred.
If any compliance monitoring constituent shows a statistically significant increase in
concentration above the concentration limits set forth in the facility's ground-water
protection standard, the facility enters the third phase of ground-water monitoring,
corrective action. In corrective action, the facility owner or operator is required to "remove
or treat hi place" all constituents that exceed the allowed concentration limits specified in
the facility permit. The monitoring associated with corrective action must demonstrate the
effectiveness of the clean-up and must be able to determine whether any other constituents
are entering the ground water at concentrations above the concentration limits.
The 1982 regulations required that contaminated ground water be analyzed for all
constituents contained in Appendix VDI to Part 261 ("Appendix VIII"). While appropriate
for hazardous waste listing purposes, the Appendix VIII list presents a number of difficulties
when used for purposes of ground-water monitoring (RMAL, 1984; U.S. EPA, 1987c).
These difficulties include practical and analytical problems such as monitoring for large
categories of chemicals, lack of availability of some analytical standards, and the lack of
reliable analytical methods for many constituents. Other problems relate to the dissociation
11-380
-------
or actual decomposition of many Appendix VQI constituents in water, rendering monitoring
for these constituents impractical. To address these analytical problems, EPA proposed on
July 24,1986 to replace the requirement to monitor for all Appendix VIII constituents with
a requirement to monitor for a new series of ground-water monitoring analytes listed in
Appendix IX. Appendix IX was promulgated as a final rule on July 9, 1987 (52 FR 25942),
and included those constituents in Appendix VIII that had available analytical methods for
the ground-water matrix, plus 17 constituents routinely monitored in the Superfund program.
DISCUSSION OF PROPOSED CHANGES TO SUBPART F
Detection Monitoring Analytes (Appendix IX-A)
The Agency expects to propose amendments to the Subpart F regulations to change the
provisions governing the selection of detection monitoring analytes. The regulations
currently require an owner or operator of a facility in detection monitoring to monitor for
indicator parameters, waste constituents, or reaction products that provide a reliable
indication of the presence of hazardous constituents in ground water. Studies have shown
that volatile organic compounds (VOCs) serve as reliable leak indicators at hazardous waste
TSDFs because they frequently occur in leachate and contaminated ground water (Eckel et
al., 1985; Plumb, 1987; Lawless, 1987; Rosenfeld, 1990). Inorganic constituents (e.g., metals)
have also been reported to occur in leachate from hazardous waste TSDFs (Bramlett et al.,
1987; WMI, 1990 and 1991), and in ground water in the vicinity of TSDFs (Lawless, 1987).
In consideration of these data, the Agency expects to propose a list of detection monitoring
constituents known as Appendix IX-A.
The Appendix IX-A constituents are a subset of the Appendix IX constituents, and consist
of 48 VOCs and 16 metals that the Agency believes would serve as good "release indicators"
for hazardous waste disposal sites that receive a variety of wastes. The specific VOCs
contained in Appendix IX-A were chosen primarily by determining which VOCs could be
identified by gas chromatography/mass spectroscopy (GC/MS) with a reasonable degree of
precision and accuracy (Lawless, 1990). Other considerations regarding the selection of
VOCs was their reported frequency of occurrence in leachate and ground water (discussed
above). The GC/MS method recommended minimizes the number of separate analyses
required to determine the concentration of many VOCs in a ground-water sample (e.g., all
VOCs in Appendix IX-A can be determined in a single GC/MS analysis). Thus, the use of
GC/MS procedures for ground-water analyses provides reliable results and conserves
analytical resources. Because of these advantages, EPA assumed that the newly proposed
SW-846 GC/MS Method 8260, a modification of Method 8240 (54 FR 3213; January 1989)
in SW-846 would be the standard method used for this analysis. All but two (i.e., barium
and vanadium) of the 16 metals on Appendix DC-A have been on EPA's Priority Pollutant
List since 1979 (U.S. EPA, 1979), and can be analyzed by inductively coupled plasma
emission spectroscopy (ICP) or atomic absorption spectroscopy (AA). The analytes that
comprise Appendix IX-A are listed in Table 1.
11-381
-------
Although Appendk IX-A may be appropriate for the majority of TSDFs, situations may
arise that would warrant further tailoring of the list to meet site-specific concerns. For
example, a facility that manages only smelter ash may not normally need to monitor for
VOCs, because VOCs would not likely be present in the ash, and therefore, would not
provide a reliable indication that the regulated unit was releasing hazardous wastes to
ground water. In order to retain flexibility in the current regulations, alternative provisions
will allow the Regional Administrator to add or delete constituents from Appendix IX-A
after considering the waste managed hi the regulated unit.
Proposed Revisions to Appendix IX
Under the Subpart F ground-water monitoring program, Appendk IX is the "master" list of
ground-water monitoring analytes. Appendix IX constituents are monitored at facilities that
are in compliance monitoring or corrective action. Appendix IX contains 222 constituents
that in 1987 had analytical methods that were verified to a sufficient degree, and that were
amenable to ground water monitoring on a routine basis (U.S. EPA, 1987c). The 222
constituents on Appendix IX consist of 17 metals and metalloids, 2 inorganic ions, 6 classes
of organic compounds (i.e., chlordanes, toxaphenes, PCBs, PCDDs, PCDFs, and xylenes),
and 197 specific organic chemicals.
EPA expects to propose removing eleven analytes from the current Appendix IX. The
eleven constituents proposed for deletion were chosen on the basis of new analytical data
that EPA has generated or received since the Appendix IX rule was first promulgated in
1987. These new data indicate that, for the eleven compounds proposed for deletion, the
analytical procedures described in SW-846 do not provide consistently acceptable results in
terms of method performance for determining their concentration in ground water (Lawless,
1990). In addition, 4-nitroquinoline 1-oxide is being proposed for deletion from Appendix
IX because SW-846 does not provide QC criteria or accuracy and precision data for its
analysis. Further, since 4-nitroquinoline 1- oxide is an experimental pharmaceutical, it was
not produced in commercial chemical quantities and therefore has a low frequency of
occurrence hi ground water near TSDFs (Plumb, 1991). The chemical compounds proposed
for deletion from Appendix IX are listed in Table 2.
The Agency expects to propose the addition of six constituents to Appendix IX. All six
constituents are members of the volatile organic class of compounds, and, as discussed
earlier, VOCs have been shown to serve as good release indicators. In addition, all six
VOCs are amenable to analysis by GC/MS (Method 8260 in SW-846), and are included in
Appendix VIE to Part 261 as part of small classes of hazardous constituents. Furthermore,
five of the six compounds are halogenated alkanes, many of which are suspected
carcinogens. The six constituents suggested for addition to Appendix IX are listed in Table
3.
11-382
-------
TABLE 1
LIST OF CONSTITUENTS FOR DETECTION MONITORING rAPPENDIX IX-A1
Common Name
1. Acetone
2. Acrylonitrile
3. Benzene
4. Bromochloromethane
5. Bromodichlorornethane
6. Bromoform; Tribromomethane
7. Carbon disulfide
8. Carbon tetrachloride
9. Chlorobenzene
10. Chloroethane; Ethyl chloride
11. Chloroform; Trichloromethane
12. Dibromochloromethane
Chlorodibromomethane
13. l,2-Dibromo-3-chloropropane
14. 1,2-Dibromoethane;
Ethylene dibromide
15. 1,2-Dichlorobenzene
16. 1,4-Dichlorobenzene
17. trans-l,4-Dichloro-2-butene
18. 1,1-Dichloroethane; Ethyldidene
chloride
19. 1,2-Dichloroethane;
Ethylene dichloride
20. 1,1-Dichloroethylene;
1,1-Dichloroethene;
Vinylidene chloride
21. cis-l,2-Dichloroethylene;
cis-l,2-Dichloroethene
22. trans-l,2-Dichloroethylene;
trans-1,2-Dichloroethene
23. 1,2-Dichloropropane;
Propylene dichloride
24. tis-l,3-Dichloropropene
25. trans-13-Dkhloropropene
26. Ethylbenzene
27. 2-Hexanone; Methyl butyl ketone
28. Methyl bromide; Bromomethane
29. Methyl chloride; Chloromethane
30. Methylene bromide;
Dibromomethane
31. Methylene chloride;
Dichloromethane
32. Methyl ethyl ketone; MEK;
2-Butanone
33. Methyl iodide; lodomethane
34. 4-Methyl-2-pentanone
Methyl isobutyl ketone
CASRN2
67-64-1
107-13-1
71-43-2
74-97-5
75-27-4
75-25-2
75-15-0
56-23-5
108-90-7
75-00-3
67-66-3
124-48-1
96-12-8
106-93-4
95-50-1
106-46-7
110-57-6
75-34-3
107-06-2
75-35-4
156-59-2
156-60-5
78-87-5
10061-01-5
10061-02-6
100-41-4
591-78-6
74-83-9
74-87-3
74-95-3
75-09-2
78-93-3
74-88-4
108-10-1
Common Name1 CASRN2
35. Styrene 100-42-5
36. 1,1,1,2-Tetrachloroethane 630-20-6
37. 1,1,2,2-Tetrachloroethane 79-34-5
38. Tetrachloroethylene; 127-1&45
Tetrachloroethene;
Perchloroethylene
39. Toluene 108-88-3
40. 1,2,3-Trichlorobenzene 87-61-6
41. 1,1,1-Trichloroethane; 71-55-6
Methylchloroform
42. 1,1,2-Trichloroethane 79-00-5
43. Trichloroethylene; 79-01-6
Trichloroethene
44. Trichlorofluoromethane; 75-69-4
CFC-11
45. 1,2,3-Trichloropropane 96-18-4
46. Vinyl Acetate 108-05-4
47. Vinyl Chloride 75-01-4
48. Xylene (Total) 1330-20-7
49. Antimony (Total)
50. Arsenic (Total)
51. Barium (Total)
52. Beryllium (Total)
53. Cadmium (Total)
54. Chromium (Total)
55. Cobalt (Total)
56. Copper (Total)
57. Lead (Total)
58. Mercury (Total)
59. Nickel (Total)
60. Selenium (Total)
61. Silver (Total)
62. Thallium (Total)
63. Vanadium (Total)
64. Zinc (Total)
1-383
5
-------
TABLE 2
Chemical Compounds Proposed for Deletion from Appendix IX
Common Name1 CAS RN2
1. Aniline 62-53-3
2. Aramite 140-57-8
3. alpha, alpha-Dimethylphenethylamine 122-09-8
4. 1,4-Dioxane 123-91-1
5. Hexachlorophene 70-30-4
6. 4-Nitroquinoline 1-oxide 56-57-5
7. N-Nitrosomorpholine 59-89-2
8. Pentachloroethane 76-01-7
9. 2-Picoline 109-06-8
10. Pyridine 110-86-1
11. Tetraethyl dithiopyrophosphate 3689-24-5
TABLES
Chemical Compounds Proposed for Addition to Appendix IX
Common Name1
1. Bromochloromethane
2. cis-l,2-Dichloroethylene
3. 1,3-Dicbloropropane
4. 2,2-Dichloropropane
5. 1,1-Dichloropropene
6. 1,2,3-Trichlorobenzene
CASRN2 APPENDIX WI REFERENCE
74-97-5 Halomethane, N.O.S.
156-59-2 1,2-Dichoroethylene
142-28-9 Dichloropropane, N.O.S.
594-20-7 Dichloropropane, N.O.S.
563-58-6 Dichloropropene, N.O.S.
87-61-6 Chlorobenzene, N.O.S.
1 Common names are those widely used in government regulations, scientific
publications, and commerce; synonyms exist for many chemicals.
2 Chemical Abstracts Service registry number. Where "Total" is entered, all analytes
in the ground water that contain this constituent are included.
11-384
-------
Variance From the Annual Appendix IX Analysis During Compliance Monitoring
During compliance monitoring, the owner or operator is required to monitor for parameters
identified in the facility permit at specified frequencies. In addition, ง 264.99(g) requires
the owner or operator to analyze samples from all wells located at the point of compliance
for all Appendix IX constituents at least annually. If additional constituents are found, then
the owner or operator must report their concentrations and list these constituents in the
facility permit. All additional constituents that are subsequently listed in the facility permit
as a result of the annual Appendix IX monitoring, also form the basis for compliance
monitoring, and are sampled and analyzed at least semiannually.
Experience has shown that for some hazardous waste TSDFs, annual monitoring for all
Appendix IX constituents may not be necessary. Certain analytes such as EPA's Priority
Pollutants have been shown in studies to have a higher frequency of occurrence in leachate
and contaminated ground water than do other constituents in Appendix IX (WMI, 1990 and
1991; Plumb, 1991). Furthermore, in each of these studies, "non-priority pollutant" Appendix
IX constituents were not detected in the absence of priority pollutants. This suggests that
routine monitoring for non-priority pollutants may not be necessary at every TSDF. In light
of this new information, EPA expects to propose a site-specific variance to the annual
Appendix IX monitoring requirements under certain circumstances. Such a variance could
involve performing an abbreviated Appendix IX analysis on an annual basis. To exclude a
constituent from the annual Appendix IX analysis, the owner or operator would be required
to demonstrate that the constituent could not be present in the waste managed by the
facility (either as a constituent of the waste, or as a reaction product), and is not present in
the facility's soil and ground water. The benefits of the variance would be realized primarily
for those constituents that require special analytical methods (e.g., TCDD) rather than for
those that are amenable to analytical "scan" techniques such as SW-846 method 8260.
EPA expects that any variance from the annual monitoring requirements for Appendix IX
constituents would not relieve the owner or operator from ever monitoring for the excluded
constituent(s). The initial, full Appendix IX analysis would still be required in detection
monitoring. Retention of this requirement is necessary to characterize the nature and extent
of a release and could be used to demonstrate that the excluded constituents are not present
in ground water at the facility. In addition, if a successful demonstration is made and the
Regional Administrator excludes constituents from the annual Appendix IX compliance
monitoring requirements, the owner or operator would be required to monitor for all
Appendix IX constituents (including the excluded constituents) at least once every five years,
when the permit is usually reviewed or renewed. If at any time a facility that has a
monitoring exclusion began to receive or generate wastes that contain any excluded
Appendix IX constituents, the facility would be required to resume annual monitoring for
the appropriate analytes. Likewise, these steps would need to be followed if a treatment
process was modified, or a new one begun, that resulted in the production of the excluded
constituents.
11-385
-------
CHAPTER ELEVEN of SW-846: GROUND-WATER MONITORING
One of the major topics that arose during EPA's recent review of over forty TSDF operating
permits (Permit Quality Review) throughout the nation was the importance that each of the
numerous steps taken by an owner or operator to comply with RCRA ground-water
monitoring requirements has in determining the quality of the data produced by monitoring
networks (U.S. EPA, 199 la). One example of this is the value of an adequate
characterization of site hydrogeology. Improper site characterization can lead to incorrect
placement of monitoring wells, or a failure to recognize ground-water flow paths and
contaminant migration pathways.
In addition, inspections conducted by EPA's Hazardous Waste Ground-Water Task Force
(HWGWTF) during the years 1984 to 1987 identified deficiencies in existing ground-water
monitoring systems and determined that many of the deficiencies resulted from owners and
operators collecting poor quality hydrogeologic data, collecting inadequate quantities of
hydrogeologic data (or misinterpreting such data), and using improper sampling and analysis
techniques (U.S. EPA, 1988). The deficiencies almost always involved technical areas for
which die RCRA regulations provided the least specificity, but that were covered extensively
in non-binding EPA guidance documents (e.g., subjects such as hydrogeologic
characterization, well construction and location, and ground-water sample collection). The
HWGWTF and Permit Quality Review experience highlighted the need to develop
nationally consistent regulatory requirements addressing the process an owner/operator must
follow to characterize site hydrogeology, to design and construct a ground-water monitoring
system, and to collect and analyze ground-water samples.
As a result, EPA expects to propose to require owners and operators to use the methods
described in proposed revisions to Chapter Eleven of SW-846 when conducting
hydrogeologic investigations, designing and constructing monitoring systems, and performing
ground-water sampling and analyses. Chapter Eleven of SW-846 embodies the Agency's
best judgment and current understanding regarding ground-water monitoring techniques, and
addresses a variety of ground-water monitoring techniques and procedures including:
hydrogeologic characterization, well placement, well design, well drilling, well completion,
well casing materials, well development, well purging, sampling equipment and methods, and
sample handling. EPA does not expect that any new burdens will be placed on the vast
majority of owners and operators by requiring them to conform to the methods discussed
in SW-846 because these techniques and methods are based on widely accepted practices
of most geologists and ground-water professionals. Furthermore, for each phase of ground-
water monitoring system design and operation, Chapter Eleven of SW-846 generally offers
several methods that are acceptable depending on the specific hydrogeologic setting of a
facility, the waste management practices, and the waste characteristics. Where specific
techniques or procedures are not provided because of the complexity and site-specific nature
of ground-water monitoring programs, Chapter Eleven of SW-846 provides discussion and
technical guidance on the available alternatives. In these cases, there is a significant amount
of flexibility allowed in the choice of methods used for ground-water monitoring system
design and operation.
11-386
-------
Data Quality Objectives for Ground-water Monitoring
The Agency has begun using DQOs for evaluating remedial response activities associated
with Superfund sites to define the type and quality of data required to support specific
regulatory decisions (EPA, 1987a and 1987b). DQOs include both qualitative and
quantitative specifications. Many of the current ground-water monitoring requirements that
are specified in Subpart F are broad-based performance standards that allow the Regional
Administrator, RCRA Part B permit writers, and owners and operators of hazardous waste
TSDF's to account for site-specific factors when designing a ground-water monitoring system
that meets the requirements of the RCRA regulations. The wide variations in waste
management practices coupled with diverse hydrogeologic settings and geochemical
environments across the United States make it difficult to promulgate a regulation specifying
a minimum number of monitoring wells and their location that would be applicable to all
facilities. EPA instead has relied on technical guidance documents (e.g., U.S. EPA 1986;
U.S. EPA, 1989) and the experience of permit writers to implement these types of general
performance standards on a site-specific scale. Presently, given the current "state-of-the-art"
of ground-water monitoring practices, a qualitative approach to defining the adequacy of
ground-water monitoring systems is the norm. However, significant efforts are underway at
EPA to develop quantitative approaches for designing ground-water monitoring systems.
The Agency continues to focus on efforts that will improve both the type and quality of
RCRA ground-water monitoring data. Changes to Appendix IX and the creation of
Appendix IX-A will improve the type of data collected, by changing the constituents for
which owners/operators must monitor. A variance to the annual Appendix IX compliance
monitoring requirement will ensure that meaningful data are collected. The incorporation
of Chapter Eleven of SW-846 into the Part 264 and Part 270 ground-water monitoring
requirements will offer more prescriptive directions on what methods and procedures should
be used in the design and operation of ground-water monitoring systems. These are part
of the Agency's efforts to establish qualitative data quality objectives for the RCRA ground-
water monitoring program.
Data quality for ground-water sampling and analysis activities is also addressed in Chapter
One of SW-846 titled, "Quality Control." Chapter One of SW-846 identifies the minimum
quality control (QC) components to be used when performing all RCRA sampling and
analysis activities, and includes the QC information which must be documented. Chapter
One of SW-846 provides guidance on the development of quality assurance project plans for
field and laboratory work that is conducted in support of the RCRA program. Chapter One
was part of the first update package to SW-846, third edition, and is mandatory for
compliance with RCRA sampling and analysis requirements.
Quantitative Data Quality Objectives for RCRA Ground-Water Monitoring
The Agency is assessing the feasibility of establishing quantitative DQOs for ground-water
monitoring under 40 CFR Part 264, Subpart F. Quantitative DQOs would be developed for
11-387
-------
each phase of monitoring and would establish numeric standards that specify the level of
performance for RCRA ground-water monitoring systems. Thus, quantitative DQOs for
detection monitoring could, for example, require that ground-water monitoring networks
achieve a specified probability of detecting contamination. Quantitative DQOs for
compliance monitoring could require that ground-water monitoring networks achieve a
specified probability of characterizing the extent of ground-water contamination.
Quantitative DQOs for corrective action could require that ground-water remediation efforts
achieve clean-up standards within a specified probability. A similar approach has been used
to support decisions concerning the design of remedial actions for contaminated soils at
Superfund sites (Neptune, et al., 1990). In all phases of RCRA ground-water monitoring,
quantitative DQOs would allow the Agency to specify the exact number and location of
monitoring wells, and number of ground-water samples, required to achieve a desired level
of performance.
The Agency's Office of Research and Development is investigating the efficacy of
establishing quantitatively-based DQOs for ground-water monitoring (U.S. EPA, 199 Ib).
Research plans are oriented toward developing a process aimed at defining, with a specified
probability, that a monitoring well system will detect a release from a TSDF. This process
will still involve the collection of detailed site-specific hydrogeologic data to support the
development of a conceptual model. This data may then be integrated with a conditional
simulation model and/or a contaminant fate and transport model that would predict
preferential flow paths of contaminant migration and estimate the probability of leak
detection based on monitoring well network configuration.
Relatively early research performed by Massmann and Freeze (1987), calculated the
probability of contaminant plume detection by monitoring networks. As noted by Meyer and
Brill (1988), however, these investigations stopped short of optimizing ground-water
monitoring network performance (in terms of the probability of detecting a contaminant
plume) by failing to generate alternative networks that are more efficient with respect to
contaminant plume detection. Meyer and Brill utilized Monte Carlo simulations of plume
releases to develop a method for optimizing the location of monitoring wells.
Quantitative monitoring network design methods offer intriguing advantages over their
qualitative alternatives, and are beginning to find applications at hazardous waste sites. A
two-dimensional deterministic model based on the work of Meyer and Brill has been used
to predict low density, aqueous-phase contaminant plume detection in unconfined aquifers
(Wilson, et al., 1991). This model offers a quantitative yet user-friendly approach to
monitoring network design. Other applications of quantitative monitoring network design
will likely continue to surface in the literature.
A more recent development of a procedure to estimate the probability of contaminant
plume detection uses geostatistical conditional simulation and parameter estimation
sequentially to generate contaminant migration pathways (Weber, et al., 1991). Recognizing
that aquifer heterogeneities and the high cost of hydraulic conductivity measurements often
inhibit adequate site characterization, these researchers utilized hydraulic head and available
11-388
-------
hydraulic conductivity measurements to estimate the distribution of flow paths (Figure 1),
and developed a relationship between the probability of plume detection and monitoring
system cost (Figure 2). The procedure is also amenable to conditional simulation of
hydraulic conductivity if sufficient measurements are available to perform geostatistical
analysis (Weber, et at., 1991).
The results of research efforts like those described above could provide EPA with a
quantitative means for specifying DQOs for Subpart F ground-water monitoring networks.
Current limitations of ground-water monitoring, subsurface characterization, and modeling
techniques, however, make it difficult to develop quantitative DQOs (most of the current
applications utilize two-dimensional models). For example, it may not be possible or
practical to design a monitoring system that will detect releases at a desired probability of
contaminant plume detection. Before a probability statement can be made, population
characteristics should be known (or assumed to be known). In the context of ground-water
monitoring at TSDFs, the population consists of all possible contaminant migration pathways
in the subsurface. To characterize this population, very detailed site characterization
methods and analyses are required. Consequently, a central issue involves the level of detail
that a site characterization must include to define all of the population characteristics. As
discussed above however, surrogate parameters (i.e., hydraulic head measurements) for
hydraulic conductivity have been used successfully to define the spatial distribution of
contaminant migration pathways and evaluate monitoring well performance where data is
sparse and collection methods are expensive (Weber, et al. 1991).
EPA will continue to support the development of quantitative DQOs for ground-water
monitoring under Subpart F as technical advances allow. EPA will use such information to
assess the feasibility of developing quantitative DQOs for ground-water monitoring. If an
acceptable procedure is developed for establishing quantitative DQOs, it will be proposed
in the Federal Register and formally opened to .public comment.
1-389
-------
FIGURES
Estimated Probability
attribution of Flow Piths
Monitoring Network Cos
Thousands of Dollars
60
50
40
30
20
10
Figure 1 (After Weber, et al., 1991)
40
60
80
100
Estimated Percent Probability of Contaminant Plume Detection
(Based on cost of $5,000 per well, 300 ft spacing)
Figure 2 (After Weber, et al., 1991)
11-390
-------
SUMMARY
EPA expects to propose amendments to the Subpart F requirements for ground-water
monitoring at TSDFs. The proposal will amend the list of ground-water monitoring
constituents (Appendix IX) based on analytical considerations; create a subset Detection
Monitoring list of constituents that have been shown to provide a reliable indication of
releases from TSDFs; issue an exclusion variance from the annual Appendix IX Compliance
Monitoring requirement; and require owners and operators of TSDFs to comply with
ground-water monitoring methods and procedures contained in a revision to Chapter Eleven
of SW-846.
EPA is also conducting research on quantifying ground-water monitoring network design
efficiency. Research efforts are investigating the efficacy of optimizing the probability of
contaminant plume detection for a given monitoring well configuration. The desired
outcome of this research would allow for the establishment of quantitative DQOs for
ground-water monitoring.
ACKNOWLEDGEMENTS
This paper was written, in part, by members of U.S. EPA's Office of Solid Waste,
Washington, D.C. It has not been reviewed by the Agency and the contents do not
necessarily reflect the views and policies of EPA. Mention of trade names, commercial
products, or publications does not constitute endorsement or recommendation for use.
REFERENCES
Bramlett, J., Furman, C, Johnson, A., Ellis, W.D., Nelson, H., Vick, W.H., 1987.
Composition of Leachate from Actual Hazardous Waste Sites. Project Report for U.S.
EPA (EPA/600.S2-87/043).
Code of Federal Regulations, Title 40, Part 261, Section 261.20. Appendix VIII, Hazardous
Constituents.
Eckel, W.P., D.P. Trees, and S.P. Kovel, 1985. Distribution of Chemicals and Toxic
Materials Found at Hazardous Waste Dump Sites. Proceedings, National Conference on
Hazardous Wastes and Environmental Emergencies. Control Research Institute. Silver
Spring, Maryland. May. pp. 250-257.
Garman, J., Freund, T., and Lawless, E., 1987. Testing for Groundwater Contamination at
Hazardous Waste Sites. Journal of Chromatographic Science, vol. 25, pp. 328-337.
Lawless, E., 1987. Preliminary Report on the Analysis of Groundwater Monitoring Data.
Letter Report to J. Garman, Work Assignment 12, Contract No. 68-01-7310.
11-391
-------
Lawless, E., 1990. Technical Background Document. Ground-Water Monitoring
Constituents: Appendix IX-A and Appendix IX. Work Assignment 68, Contract No. 68-01-
7287.
Massmann, J., and R.A. Freeze, 1987. Groundwater Contamination from Waste
Management Sites: The Interaction Between Risk-Based Engineering Design and
Regulatory Policy, 1, Methodology, Water Resour. Res., 23(2), pp. 351-367.
Meyer, P.D., and E.D. Brill, 1988. A Method for Locating Wells in a Groundwater
Monitoring Network Under Conditions of Uncertainty. Water Resour. Res., 24(8), pp. 1277-
1282.
Neptune, D., Brantly, E.P., Messner, M J., Michael, D.I., 1990. Quantitative Decision Making
in Superfund: A Data Quality Objectives Case Study. Hazardous Materials Control, v. 3,
no. 3, pp. 19-27.
Plumb R.H., and A.M. Pitchford, 1985. Volatile Organic Scans: Implications for Ground-
Water Monitoring. Conference on Petroleum Hydrocarbons and Organic Chemicals in
Ground-Water-Prevention, Detection, and Restoration. Houston, Texas. American
Petroleum Institute and National Water Well Association, pp. 207-222, November 13-15.
Plumb, R.H., 1987. A Comparison of Ground-Water Monitoring Data from RCRA and
CERCLA Sites. Ground Water Monitoring Review, v. 8, pp. 94-100.
Plumb, R.H., 1991. The Occurrence of Appendix IX Organic Constituents in Disposal Site
Ground Water. Ground Water Monitoring Review, v. XI, no. 2, pp. 157-164.
Rocky Mountain Analytical Laboratory, 1984. "Evaluation of the Applicability of SW-846
Manual to Support all RCRA Subtitle C Testing." A report to American Petroleum
Institute on Hazardous Waste Management System; Ground Water Testing and
Monitoring Activities. Rocky Mountain Analytical Laboratory, Arvada, Colorado, 148 pp.
December 20.
Rosenfeld, J. K., 1990. Ground-Water Contamination at Hazardous Waste Disposal
Facilities. Proceedings of the National Water Well Association Conference on Ground
Water Geochemistry. February, 1990. Kansas City, Missouri. Ground Water Management,
v. 1, pp. 237-250.
Test Methods for Evaluating Solid Waste, Physical/Chemical Methods, SW-846. U.S.
Government Printing Office, Washington, D.C. Order No. 955-001 00000-1.
U.S. Environmental Protection Agency, 1979. Water-Related Environmental Fate of 129
Priority Pollutants. Volume I. EPA-440/4-79-029a.
11-392
-------
U.S. Environmental Protection Agency, 1982. Hazardous Waste Management System;
Standards Applicable to Owners and Operators of Hazardous Waste Treatment, Storage,
and Disposal Facilities; and EPA Administered Permit Programs. Federal Register, vol. 47,
no. 143, July 26, 1982. pp. 32274-32388.
U.S. Environmental Protection Agency, 1986. RCRA Ground Water Monitoring Technical
Enforcement Guidance Document. U.S. EPA Office of Waste Programs Enforcement,
Washington, D.C., 208 pp. and App.
U.S. Environmental Protection Agency, 1987a. Data Quality Objectives for Remedial
Response Activities (Development Process). EPA/540/G-87/003.
U.S. Environmental Protection Agency, 1987b. Data Quality Objectives for Remedial
Response Activities (Example Scenario: RI/FS Activities at a Site with Contaminated
Soils and Ground Water). EPA/540/G-87/004.
U.S. Environmental Protection Agency, 1987c. List (Phase 1) of Hazardous Constituents for
Ground-Water Monitoring. Federal Register, vol. 52, no. 131, July 9,1987. pp. 25942-25953.
U.S. Environmental Protection Agency, 1988. Hazardous Waste Ground Water Task Force:
1987 Status Report and 1988/1989 Program Recommendations. 34pp.
U.S. Environmental Protection Agency, 1989. Handbook of Suggested Practices for the
Design and Installation of Ground-Water Monitoring Wells. Cooperative Agreement No.
CR-812350-01. (EPA/600/4-89/034).
U.S. Environmental Protection Agency, 199 la. Permit Quality Review Revised Draft Report
Office of Solid Waste, Permits and State Programs Division, (in preparation).
U.S. Environmental Protection Agency, 1991b. Subsurface Monitoring Research Activities:
EPA/600/9-91/003. 45pp.
Waste Management of North America, Inc., 1990. Leachate Characterization Study.
Wastewater Group. Project No. 307CO. 7 pp. and Tables, App.
Waste Management of North America, Inc., 1991. Leachate Characterization Study.
Wastewater Group. Project No. 307CO. 30 pp. and App.
Weber, D., Easley, D., and Englund, E., 1991. Probability of Plume Interception Using
Conditional Simulation of Hydraulic Head and Inverse Modeling. Mathematical Geology,
Vol. 23, No. 2, p. 219-239.
Wilson, C.R., Eichenberger, C.M., Kindred, J.S., Jackson, R.M., and Mercer, R.B.,
1991. "Efficiency Based Monitoring System Design." Presentation at the American Society
of Civil Engineers' Energy in the 90's Conference. March. Pittsburgh, Pennsylvania.
11-393
-------
Q7 DE-MYSTIFYING THE PROBLEM OF FILTERED VS UNFILTERED SAMPLES
Richard D. Brown. Lead Scientist, Hazardous Waste Systems Department, Energy, Resource and
Environmental Systems Division, Center For Civil Systems, The MITRE Corporation, McLean,
Virginia 22102-3481
ABSTRACT
A controversy has persisted for at least a decade concerning the filtration versus non-filtration
of ground water samples, particularly with respect to samples used for metals analyses.
Renewed emphasis on the collection of data of high quality and an ever increasing need to
better understand the subsurface environment have resulted in a resurgence of attention on the
representativeness of ground water samples to adequately reflect the level of threat to public
health and the environment Often, it is difficult to differentiate the contribution of metals from
natural sources, incomplete purging or disturbance of sediments during sampling, or releases
from an abandoned or uncontrolled hazardous waste site. This paper examines the various
facets of the problem, discusses options for filtration versus non-filtration when collecting
samples for different purposes, and clarifies the relative importance of various fractions of a
sample (i.e., suspended solids, colloids, dissolved solids, and colloids and dissolved solids
adsorbed on suspended solids) in understanding the subsurface environment The paper also
discusses the benefits and drawbacks associated with the related issues of acidification, transport
and storage temperatures, use of filters of varying porosity, field versus laboratory filtration, and
the development of a well to a turbidity standard. The above issues also are discussed within
the context of comparing the sample data to health and environmental benchmarks, both for
ground water and for surface water samples. Possible solutions to the problem are suggested.
INTRODUCTION
A topic of fervent debate when discussing ground water sampling plans often focuses on
whether "to filter or not to filter" collected samples. One viewpoint is that filtration results in
a substantial physical and chemical modification of the sample. Another perspective is that
filtration allows data users to concentrate only on those contaminants which are actually
dissolved, excluding any substances which may be adsorbed on, or conveyed by, paniculate
matter in suspension. Both positions have merits and the collection of filtered or unfiltered
samples (or both) may be suitable dependent on the questions which need to be resolved
(Nielsen 1991).
The reasons for filtration of ground water samples include:
Removal of suspended solids to permit analyses only of the dissolved fraction
of substances in the sample, reflecting drinking water quality as delivered
Removal of any interference caused by suspended particles (e.g., when ultraviolet
spectrophotometric screening techniques are used
Analysis of "clear" samples, required when using delicate instrumentation easily
clogged by sediment-laden samples
Separate analyses of constituents associated with suspended solids
Determination of the percent of suspended solids
The disadvantages associated with filtration include:
11-394
-------
Chemical changes in the sample due to changes in partial pressure of dissolved
gases during filtration under positive pressure or a vacuum
Volatile organic substances may be lost to the atmosphere during filtration
Aeration of sample during filtration can cause precipitation of metals
Possible inadvertent removal of substances, both organic and inorganic, that tend
to adsorb on suspended particles
Increased opportunities for sample contamination, especially if filtration is
conducted in the field
Practical difficulties in the field when filtering during sub-zero temperatures and
when filtering sediment-laden water
Generally, the problem of deciding "to filter or not to filter" is associated with the analysis of
metals. Most ground water samples collected for the analysis of organic compounds are not
filtered because:
Many organic hazardous substances are not natural components of ground water,
therefore, the analyst is interested in the total sample concentration
Most volatile organic compounds can easily be lost during filtration
Since water solubility and partition coefficients vary among most organic
substances, there is no compelling reason to differentiate between the suspended
solid and dissolved paniculate fractions of a sample collected for the routine
analysis of organic substances.
Except for variations for some pesticides and PCBs, concentrations of organic
hazardous substances in ground water do not vary as markedly as metals in proportion
to the amount of sediment in a sample.
Thus, the problem of "to filter or not to filter" relates primarily to a perceived need to filter
samples of ground water to be used for the analysis of metals. The problem manifests itself
in the form of artifacts in ground water monitoring data which cannot easily be explained within
the context of having intentionally collected representative samples. For example, very high
metal concentrations have been observed in samples collected to determine contamination from
a waste site when the metals could not be attributable to that specific source. Sometimes,
background concentrations would be highly elevated, but levels near a source would be at trace
levels. Very high concentrations (e.g., 640,000 ugA aluminum, 1,000 ugA nickel, 500 ugA
chromium) of metals commonly found in soil have been observed in ground water, when
normally such concentrations are low (e.g., 200 ug/1, 40 ug/l, and 10 ugA, respectively) in clear
ground water.
Generally, there appears to be a direct relationship between high levels of metals and high
levels of suspended solids in the samples, independent of a sample being representative of
background or site contamination. High levels of suspended solids are suspected to be the
source of the high concentrations of metals. The presence of high levels of suspended solids
in ground water samples complicates efforts to establish representativeness of samples and
attribution to sources of contamination. In the evaluation of data from such samples, it is
11-395
-------
difficult to differentiate the contribution of metals from natural sources, the incomplete purging
or disturbance of sediments during sampling, or releases from a site.
BACKGROUND
There are several terms relevant to a discussion of contaminant particle fractions in water.
Under current "Standard Methods" (American Public Health Association, et al. 1989), waterbome
solids are divided into two components. One component, "suspended solids," is retained during
nitration of water through a 0.45 micron (a micron is one millionth of a meter) filter. The
second component, "dissolved solids," passes through the filter. In addition to suspended solids
and dissolved solids, there is a fraction of suspended solids termed "colloids."
Suspended Solids
Suspended solids are waterbome particles which do not pass through a filter used to produce
a filtrate containing only dissolved solids. In static water, large suspended solids will settle to
form sediments. When sediments are disturbed, such as during the purging of a well, they will
form suspended solids.
Generally, large-sized suspended solids (e.g., greater than 10 microns in diameter) are not found
in ground water. The exception to this norm is the ground water of Karst areas where surface
debris and soil particles can enter the system through sink holes. A rapid discharge rate
through caverns and crevices can entrain more large particles through erosion of soft limestone.
Naturally occurring solids, such as clay particles and quartz silicates, move as suspensates in
ground water. At some locations and at certain times, naturally occurring metallic hazardous
substance(s) of concern can be found at relatively high concentrations in ground water. This
is particularly true for metals found in surface water and ground waters of mineralized areas.
Examples of these areas include locations of ultra-basic rocks rich in nickel and chromium,
basaltic and some sedimentary rocks high in zinc and copper, and galena-bearing rocks rich in
lead. A major fraction of the metals in the ground water of these areas is the suspended solids
present as eroded components of the parent material (rock and overburden). Because eroded
particles, in the form of sediments, can become suspended in wells during sampling, they are
a major focus of concern.
Colloids
Colloids are extremely small solid particles which will not readily settle out of a solution.
Colloids dispersed in water scatter light even though they are too small to be seen by the naked
eye. They are intermediate in size between true dissolved solids and large suspended solids
which are visible to the naked eye.
Colloids vary in size. They are classified according to size, but there is not a uniform
definition with respect to their lower or upper limit The scale used by the U.S. Department
of Agriculture and the Soil Science Society of America defines colloids as clay particles with
diameters less than two microns, but which will not pass through a 0.45 micron filter used to
extricate dissolved solids from a water sample. This classification is equivalent to particles
smaller than fines described by the U.S. Army Corps of Engineers. The Wentworth scale used
for sediments, which is a logarithmic scale in that each grade limit is twice as large as the next
smaller grade limit, defines clay particles to be smaller than 3.9 microns (Blatt et al. 1972).
Since colloids are retained by a 0.45 micron filter, they are considered to be a fraction of
suspended solids.
11-396
-------
In surface waters, where water movement inhibits settling, colloids have been considered to be
somewhat larger, encompassing small microorganisms such as bacteria, protozoans, and small
unicellular diatoms. In this context, colloids are considered to be particles smaller than
10 microns in diameter (Stumm and Morgan 1981).
Where mixtures of pure chemicals are studied under laboratory conditions, colloids may be
viewed as particles smaller that one micron in diameter. Such very fine particles may take up
to one year to settle from suspension.
Studies have shown that colloids can facilitate the transport of contaminants in ground water.
There is evidence that colloids in excess of 1 micron may not only be mobile in ground water
but also may move faster than the average ground water flow in porous media as the result of
such effects as size exclusion from smaller spaces (Puls 1990).
Colloids have demonstrated strong binding and sorption capacities for inorganic contaminants.
As much as 42 percent plutonium in a release has been found to be mobilized as colloids
sorbed on suspended solids (Champ et al. 1982). High metal concentrations, as much as
200 parts per billion of copper, lead, and cadmium, were found to be associated with colloidal
particles CTillekeratne et al. 1986). Other studies have shown a strong affinity for metal
sorption onto colloidal particles in ground water (Gschwend and Reynolds 1987, Enfield and
Bengtsson 1988, Puls and Bonn 1988, Puls 1990).
Dissolved Solids
Dissolved solids are extremely fine particles that pass through a filter with a pore size of 0.45
microns. Such particles will not settle from a water sample, but will remain in a vessel after
evaporation of a sample and its subsequent drying in an oven (American Public Health
Association, et al. 1989). However, for the purposes of this paper the term dissolved solids
will include the "volatile solids" which are ignited and some mineral salts which are volatilized
during a dissolved solids determination.
Strictly speaking, dissolved solids include only chemical species in solution. However, the use
of a 0.45 micron filter to remove suspended solids means that colloidal particles less than 0.45
microns in diameter are usually characterized as dissolved solids. This convention was adopted
as a consensus standard representing a compromise between complete removal of all paniculate
material and the speed with which filtration may be completed. Thus, some colloidal metal
particles have been shown to pass through a 0.45 micron filter, leading to an order of
magnitude or more error in using 0.45 micron filtration as an operational definition for
"dissolved" (Puls and Barcelona 1989a).
Dissolved solids represent the aqueous phase of transport of substances in ground water. It
should be kept in mind that there is a dynamic solid-solution equilibrium in water, wherein
elements move from solution to colloids and larger solids and back again depending on
physical, chemical, and microbiological factors. Thus, metals may exist at one location in an
aquifer in the dissolved state and at another location, or at the same location at a later point
in time, as colloidal metal oxides, metal hydroxides, metal carbonates, or chelated metals bound
in organic matrices. In fact, they all can be present at the same place and time, all in
equilibrium with one another.
Interaction of Fractions
In ground water, the three fractions of particles (suspended solids, colloidal fraction of
suspended solids, and dissolved solids) can exist simultaneously. Also, colloids and dissolved
1-397
-------
substances may be found adsorbed on suspended solids. In addition, dissolved substances can
be adsorbed on colloidal particles.
The relative states (adsorption, de-sorption, solution) of the metals can change abruptly due to
the actions of physical, chemical, and microbiological factors. For example, an acidic
environment may lead to the decomposition of metal laden suspended solids (e.g., natural clay
and silicate particles), thereby releasing natural metals into solution.
Fractions of Interest in Ground Water Assessment
Current filtration procedures (using a 0.45 micron filter) exclude most colloids and suspended
solids from ground water samples, leaving for analysis the aqueous phase containing the
dissolved fraction of the hazardous substances of concern. Filtration is useful because some
suspended solids, such as well sediments inadvertently collected during sampling, may not be
desired and require removal through filtration. However, the removal of colloids may not be
desired because of their reported capacity to adsorb and transport contaminants in the subsurface
environment.
With respect to colloids, a recent article by Puls (1990) summarized the importance of
delimiting their fraction with regard to hazardous waste site assessment activities:
"Inherent in these discussions [concerning colloids] is the concept of 'dissolved' vs.
'paniculate' and the rather arbitrary separation technique of using a 0.45 micron filter,
commonly used in data collection activities in the laboratory and in the field. If colloids as
large as 1 to 2 microns are mobile and capable of transporting contaminants for large
distances, then our sampling protocols must make allowances for this component of transport."
Thus, in summary, the desired ground water fractions of primary interest for evaluating
contamination of ground water are:
Natural, large-sized suspended solids such as found in Karst environments
Dissolved solids and colloids
Dissolved solids and colloids adsorbed on suspended solids.
Not desired are large suspended sediments artificially introduced into the sample during
collection activities.
SAMPLE PROCESSING AND FILTRATION PRACTICES
Ground water samples are collected from active drinking water wells, standby wells, and
monitoring wells. Commonly used sampling devices include electrical submersible pumps,
positive-displacement bladder pumps, bailers, and suction-lift pumps. The type of sampling
device used is based on the rate of well purging possible in view of available well yield, well
diameter, limitations in the lift capability of the device, and the sensitivity of selected chemical
species to the method of sample collection and delivery to a sampling container (Keith 1988).
Metals samples usually are acidified with nitric acid in the field to pH<2. The purpose of the
acidification is to inhibit dissolved and colloidal particles from adsorbing onto solids and the
surface of the sample container and forming precipitates (e.g., hydroxides or hydrated oxides).
11-398
-------
Under the Contract Laboratory Program (CLP) of the U.S. Environmental Protection Agency
(EPA), metals samples are acid digested in the laboratory prior to analysis. This entails
treatment with acid (and also with hydrogen peroxide if analyzed by furnace atomic absorption)
and heat (95ฐC) to oxidize organic materials. The sample is then filtered (or alternatively
centrifuged or allowed to settle by gravity) to remove insoluble material (EPA 1987a).
For the initial screening of hazardous waste sites, the EPA has recommended that the Total
Recoverable Metals Method, a method performed on an unfiltered sample, be the standard
technique in determining metal concentrations in ground water. This technique presumably
releases the loosely bound metals from the paniculate fraction but does not totally destroy the
matrix. This is viewed as preferable to a dissolved metals analysis on filtered samples, which,
by contrast, does not account for those metals that are adsorbed to the soil matrix and which
may move back and forth in equilibrium with the ground water, resulting in an underestimate
of chemical concentrations in ground water from an unfiltered tap (EPA 1989).
However, the Agency has recognized the need for filtering when a sample is highly turbid.
For example, if silt persistently appears in a sample because of well construction or design,
and the situation cannot be corrected, then it may be worthwhile to perform both the dissolved
(filtered) and total metals (unfiltered) analyses. If filtration occurs (i.e., a dissolved metals test
is to be performed), the metals samples are to be filtered immediately on-site by the field
sampler before adding preservative (EPA 1987a).
Sampling protocols in general practice often recommend that samples from ground water
monitoring wells to be used for metals analyses be field-filtered under pressure before
preservation and analysis. The filtered samples collected for metals are usually acidified.
Acidification of unfiltered samples can lead to dissolution of minerals from suspended clays.
The sample should be filtered as soon as possible after it is collected, preferably in the field.
Where field filtration is not practical, the sample should be filtered as soon as it is received in
the laboratory (American Public Health Association, et al. 1989, EPA 1976).
POTENTIAL SOURCES OF PROBLEMS
This section contains a brief discussion of the predominant mechanisms wherein undesirable
suspended solids, in the form of fine particles, become entrained in well water. The
predominant mechanisms are through inadequate well construction, development, and
maintenance and well purging and sampling.
Well Construction. Development, and Maintenance
The proper construction and development of monitoring wells is essential to the collection of
representative water samples. Improperly developed monitoring wells will produce samples
containing suspended sediments that may both bias chemical analyses of collected samples and
cause clogging of field filtering mechanisms (EPA 1987b).
When constructing monitoring wells, the drilling process may cross contaminate aquifers with
loosened fine particles of topsoil, possibly laden with agricultural or industrial chemicals (Keith
1990). Installation of a screen with oversized slots, a poorly designed filter pack, improper
screen placement, and removal of cement holding the sand grains together around the well
screen also contribute to the movement of fine-grained materials into a well.
Monitoring wells must be developed to provide water free of suspended solids. There are many
ways to develop wells. The first step in a common method of well development involves the
movement of water at alternatively high and low velocity into and out of the well screen and
11-399
-------
gravel pack to break down the mud cake on the well bore and loosen fine particles in the
borehole. This step is followed by pumping to remove these materials from the well and the
immediate area outside of the well screen. If the flushing procedure is too harsh, the gravel
pack may be dislodged leading to possible screen damage or creation of a conduit for small
fines to enter the screen. Inadequate pumping will leave sediments in the well. These
sediments can become entrained as suspended solids in samples.
Improper maintenance can lead to the incrustation of carbonates, metal hydroxides, and biofilms
of iron bacteria which can slough off as suspended solids (Driscoll 1989). These incrustations
can markedly affect the chemistry of the well water.
Purging and Sampling
Because biochemical and geochemical reactions and other factors alter the quality of water
stored in a well casing, the stored water must be removed before obtaining a sample
representative of the quality of water in the aquifer. The amount of water to be purged from
a well prior to sample collection varies from well to well. If a sample is collected too early
before complete purging, it may not reflect the quality of water in the aquifer. If collected too
late, water or contaminants from areas removed from the well can be drawn into the sample,
possibly resulting in a sample which is not representative of aquifer quality at the well location.
Often, samples are collected after a standard number of well volumes are purged (e.g., 2 to 10)
or when the purged water appears to become "stabilized", determined by the presence of water
that appears to be unclouded (Brown and Egan 1989, EPA 1983).
Well water that appears to be clear may contain paniculate matter in suspension, particularly
if the water is from new or little used wells, such as ground water monitoring wells or standby
municipal supply wells. The amount of sediment discharged from a well is affected by the type
of pump, well construction, size and type of screen, the purging rate, and other factors. Often,
fine grained materials near a well intake erode due to water pressure and well construction.
These pass through a well screen and accumulate as sediments in the bottom or on the sides
of the well casing. When a bailer or pump intake is activated for sampling, the sediments can
be disturbed and entrained as suspended sediments in the water sample (Brown and Egan 1989,
Bloese 1983).
Bailers are commonly used for both purging and sampling water from small diameter, shallow
wells because of their relatively low cost and portability. However, without very careful
control, the movement of a bailer often mixes well water, resulting in a potential for aeration
and degassing of the sample. The aeration is the result of repeated submergence and removal
of the bailer during sampling, which may result in turbulent flow of water in the wellbore.
Further aeration can occur as a result of pouring the collected sample from the top of the bailer
into the sample bottle (Keith 1988). Such aeration and degassing causes physical and chemical
changes in water quality, creating suspended solids in the form of hydroxides and other
precipitates.
Bottom-draw bailers, and suction-lift, gas-displacement, and other types of pumps have been
used to minimize the problems of aeration and turbidity. All of these devices have drawbacks
(e.g., slow withdrawal rate, degassing), compared to the simplicity of the bailer (Keith 1988,
EPA 1983).
Water samples containing suspended sediments derived from well disturbance do not represent
true ground water quality. The results of analyses of metals from such samples (if unfiltered)
would be biased high relative to true levels in ground water, due to metal release from the
disturbed sediments.
11-400
-------
Laboratory Storage and Pre-treatment
The accepted time limit for the storage of metals samples is 180 days (EPA 1987a, EPA 1976).
During this period of time, the acidic environment of the sample may cause decomposition of
the suspended solids, thereby releasing metals into solution. Also, heat and acid of the
laboratory pre-treatment procedure may release metals through decomposition of the suspended
solids. These processes do not affect determination of a "total metals" concentration in a
sample. However, the effects (i.e., release of metals into solution) of these two processes
negate the ability to obtain representative differentiation of the colloidal and dissolved fractions.
Variable Practices
The difficulty of obtaining a representative ground water sample in light of the suspended solids
problem is complicated by the lack of consistency in sample filtration and sample acidification.
Delays in filtration and preservation and the sequencing of each process result in additional
complications. Currently, there is no commonly followed practice for the filtering of ground
water samples (Puls and Barcelona 1989b).
Quality control is not implemented uniformly with respect to the preservation of a ground water
sample with acid. Often, it is standard practice to preserve a sample by adding a standard
amount of acid (e.g., 5 drops), with the intent of creating a pH<2 in the sample. However, due
to variation in the buffering capacity of ground waters in different parts of the country, the
pH of the a sample may vary from <2 to >5 following addition of the acid. The pH is seldom
verified with a pH meter and corrected to <2.
In an examination of field quality control methods in general practice, Keith (1988) found a
number of procedures and areas of disparity at the time of sampling and sample preservation
that contribute to variances in the quality of the collected water. These practices include:
aeration and degassing of sample during field filtration
delaying acidification
delaying filtration or filtering after acidification
lack of necessary temperature reduction for successful
stabilization of certain samples (e.g., mercury, chromium,
cyanide) during transport.
Delay in the preservation of metals samples can lead to substantial variation in the reported
concentration. For example, an experiment has shown that the concentration of iron in a sample
acidified immediately after collection was 11.6 mg/1; whereas, the concentration of a duplicate
sample acidified seven hours after collection was 0.33 mg/1. Replicate samples from another
site, acidified in the same manner showed similar results (5.74 to less than 0.08 mg/1).
Significant changes were also observed for other metals (Keith 1988).
A major concern related to the timing of acidification of a metals sample relates to aeration of
the sample. When ground water is in a reduced state, the addition of oxygen can cause metal
precipitation. Aeration of the sample can occur during transfer from the sampling device to a
sampling bottle, transfer to a holding container prior to filtration, or during filtration. If fixation
of the metals in the sample by addition of acid occurs after filtration, metal precipitates (e.g.,
metal hydroxides and metals adsorbed to the hydroxides) could be removed by the initial field
filtering and not be available for laboratory analysis. The turbulence and associated aeration
11-401
-------
of a metals sample during filtering affects sample quality much more than simply holding a
sample which has not been acidified, hi fact, studies indicate that aeration of the sample during
nitration can have as much affect on sample quality as the sampling activity itself.
Acidification immediately after sample collection and prior to filtering minimizes precipitate
formation after sampling (Brown and Egan 1989, Keith 1988).
POSSIBLE SOLUTIONS TO THE PROBLEM
Attainment of Turbidity Standard Prior to Sampling
It may be possible to restrict the entrainment of suspended solids into a sample. A monitoring
well can be developed in such a way that it is basically free of sediments from construction
activities. Although not always possible and only if it has been properly designed and
developed, a monitoring well can be maintained in such a way that the screen does not become
clogged and the incrustation of carbonates, metal hydroxides, and biofilms of iron bacteria are
controlled. However, such chemical and physical maintenance techniques are difficult to
perform without destroying the representativeness of samples. Excepting severe damage during
well development, a monitoring well can be purged and sampled in such a way that its clarity
is equal to that of drinking water (e.g., maximum contaminant level turbidity standard of 5
nephelometric turbidity units).
There are precedents relating to the establishment of clarity hi well water before sampling. The
goal of several Federal ground water sampling programs (e.g., EPA monitoring program
objectives under the Resource Conservation and Recovery Act, the Air Force Installation
Restoration Program, and the Superfund remedial program) is to develop, purge, and sample
monitoring wells in such a way as to assure clarity in the collected samples of water (Puls and
Barcelona 1989b, EPA 1989).
Due to time and resource constraints, there are several problems inherent in the attainment of
a turbidity standard of clarity before sampling during a site inspection. These problems include:
Duration of sampling. Some monitoring wells are so laden with fine sediments
that purging rates need to be as low as two liters per hour. Some monitoring
wells may require up to seven hours (Keith 1988) to complete an adequate
purging and sampling effort, a time and resource requirement which may not be
achievable under the conditions of an initial ground water screening.
Verification. A frequent nephelometric measurement would be required to
confirm attainment of a turbidity standard. Although relatively easy to perform,
this would be a burdensome task for site inspection personnel, given limited time
and resources. The additional sample handling could increase the probability of
sample contamination and alteration of the chemical characteristics of the sample.
Well Development The development of a well for the purposes of producing
water of potable quality is very time consuming and relatively costly compared
to the time and resource constraints associated with the installation of a
monitoring well for screening purposes.
Well Maintenance. Monitoring wells installed for site inspections may be
sampled once after installation and never again. They may remain unattended
for many months or even years between sampling events. Without periodic
screen and gravel pack cleaning, treatment for incrustation and biofouUng, and
other maintenance activities, clarity of samples cannot be assured.
11-402
-------
These problems can be overcome in a long-term monitoring program where wells are sampled
periodically (e.g., every quarter). Drawdown rates for such wells are known and the proper
well-specific optimal purge volumes determined through records of periodic sampling. The
routine, repetitious sampling regimen of such long-term monitoring programs allows for the
provision of adequate time for sampling and well maintenance activities. The time and resource
constraints of a screening process does not allow such extensive quality control procedures.
Field Filtration
Field filtration of ground water samples to separate colloidal and dissolved solids from
suspended sediments is desirable, preferably with vacuum filtration to expedite the filtering
process. The disadvantage of field filtration relates to quality control. Under field conditions
it would be difficult to avoid sample contamination while coping with several procedures
inherent in the filtering process. The filter disks need to be washed with successive volumes
of distilled water (American Public Health Association, et al. 1989) and then prewashed with
sample water to equilibrate the filter disks with sample water (disk will initially sorb certain
metals). However, this problem could be overcome through use of pre-washed disposable
filtration devices. Problems also arise with control over the build up of a "filter cake" and
resultant clogging of filters associated with high concentrations of suspended solids. During
sampling, handling, and filtration, aeration could result in unintentional metal precipitation.
Field filtration has become a routine practice in some monitoring programs, but an exacting
expectation for sample representativeness and quantitation may preclude field filtering due to
the above mentioned quality control problems. The additional resource burden associated with
the filtration of ground water samples in the field may be excessive, given the limited resources
available for site inspections.
Laboratory Filtration
Filtration in a fixed laboratory, such as a laboratory under the CLP, is an attractive alternative
compared with field filtration. Conditions are conducive for controlled analytical measurement
and sample handling.
The disadvantage to filtration in the laboratory relates to the time lag from sample collection
to analysis, the greater this time lag, the more the entrained sediments become dissolved by the
acidic preservative. Filtration in the laboratory would involve immediate analysis vs. the current
practice of metals sample storage for a prolonged period of time, bringing about a new concept
in metals analysis. However, the analysis of metals samples upon receipt by the laboratory
is logistically feasible, because the CLP requires a rapid turn around for the analysis of other
types of hazardous substances.
Preservation and Storage
The filtration of colloids and dissolved solids from large suspended solids requires special care
in sample preservation and storage in order to minimize degradation of the large suspended
solids by acid. If filtration is conducted in the field, dissolution of the suspended solids
fraction is minimal. However, if laboratory filtration is performed, special care must be taken
to minimize chemical reactions after acidification (which "fixes" dissolved solids already in
solution).
One apparently ideal method of minimizing the chemical reactions which can breakdown the
suspended solids in acidified metals samples is to lower the temperature of the samples. This
11-403
-------
is a standard technique for the preservation and storage of many types of samples. Water
reaches its maximum density at 4ฐC. For many decades, it is at this temperature that samples
of waterbome coliform bacteria, pathogens, and organisms were stored for immediate transfer
to a laboratory for culturing, plating, and analysis. Field samples easily are maintained at a
temperature of 4ฐC by means of an ice slurry (or wet ice) in an ice chest. Maintaining
chemical samples in an ice slurry is common practice for certain metals affected by biotic
activity (e.g., mercury, chromium, colorimetric analysis of copper, cyanides) and for volatile
substances. When metals samples at ambient temperature are placed in an ice slurry (or wet
ice), the samples attain a temperature of 4ฐC within three hours (Keith 1988). This cooling
process could aid in stabilizing the various metal fractions until receipt at the laboratory for
analysis.
There are a few disadvantages associated with the cooling of metals samples. One disadvantage
is that a decrease in temperature of a sample will increase its oxygen saturation level,
contributing to aeration of the sample and possible hydroxide formatioa However, acidification
of the sample should mitigate problems associated with such aeration. The cooling of a ground
water sample from a 20ฐC temperature of a warm, shallow aquifer to 4ฐC can raise its pH by
as much as one-half of a pH unit (Diehl 1970). However, if the sample is acidified properly
to a pH<2, any change in the pH due to cooling should have insignificant effects on precipitate
formation.
Freezing metal samples is another alternative. Freezing samples will minimize chemical
reactions and inhibit breakdown of suspended solids, but presents several problems. The
freezing action (unless flash frozen; e.g., with liquid nitrogen) can create a phase separation
wherein water free of acid becomes frozen first leaving the remaining liquid more acidified,
possibly creating problems in metals recovery in the laboratory (e.g., during the CLP
pie-treatment analysis. The field logistical requirements for freezing involve special transport
and handling of the freezing agent (e.g., dry ice), extra cost of materials and equipment, and
special training of field personnel. The receipt, storage, handling, and thawing of frozen
samples in the laboratory may present added logistical and analytical problems.
SUGGESTED FILTRATION PRACTICES
It is recommended that ground water metals samples be acidified immediately upon collection
in the field and cooled to a temperature of 4ฐC for transport to a fixed laboratory for analysis.
The acidification to pH<2 should be verified in the field prior to cooling the samples.
The metals samples should be filtered for the separation and analysis of colloidal and dissolved
solids immediately upon receipt at the fixed laboratory. After filtration, the filtrate should be
acid and heat pre-treated using the current CLP procedure for the pre-treatment of metals
samples.
The filter pore size used for filtration should be large enough to allow the bulk of the colloidal
particles to be recovered, but small enough as to exclude larger suspended sediments. A
commercially available, acid resistent, 5 micron pore size filter is available in standard sizes
(e.g., 2.2 cm to 4.7 cm) and is recommended. A larger pore size (10 micron) filter is
available, but is not recommended for the size range of colloids associated with ground water.
It may be possible that a more preferable 2 micron pore size, acid resistent, filter is
commercially available, but its availability needs to be confirmed.
Field filtration is not recommended for ground water metals samples. Should a decision be
made to filter ground water metals samples in the field, the following procedure is
recommended. Immediately upon collection, the samples should be subjected to mild acid
11-404
-------
treatment (e.g., nitric acid pH 3 to 4) for 10 minutes to free sorted dissolved and colloidal
contaminants from large suspended solids. Then, filtration should be preformed using a
5 micron (or 2 micron if available) pore size filter. The filtrate should be acidified immediately
upon collection in the field and cooled to a temperature of 4ฐC for transport to a fixed
laboratory for analysis. The acidification of the filtrate to pH<2 should be verified in the field
prior to cooling the samples.
Coupled with the above, relevant site inspection guidance should be developed, focusing on the
use of various techniques to minimize the entrainment of suspended sediments in ground water
metals samples.
The problem of sediments entrained in ground water samples is associated with an overestimate
of the concentration of metals in ground water. Samples collected for organic compounds
analyses should not be filtered. This is consistent with common practice to not filter samples
collected for the analysis of organic compounds (Keith 1991).
Ground water samples in Karst areas should not be filtered. The presence of suspended solids
larger than colloids is an intrinsic feature of these systems and is indicative of natural
background levels.
Both filtered and unfiltered surface water samples (split samples) should be used for metals
analyses. Data from unfiltered samples should be used for comparison with benchmarks such
as Ambient Water Quality Criteria (AWQC) which represent unfiltered concentrations. Data
from filtered samples should be used for comparison with benchmarks such as Maximum
Contaminant Levels (MCLs) which represent water delivered to a user of a public water supply.
Large suspended solids have been removed from such delivered water by various means
including sand filters, flocculation, and gravity settling in storage facilities. Even in private,
rural water supplies, paniculate matter is removed by settling in household compression tanks,
gravity and pressure filters, zeolite softeners, and other ion-exchange units for the removal of
unwanted hardness.
SUGGESTED CONFIRMING STUDIES
A number of studies need to be conducted to confirm that the recommended sample
preservation and filtration procedures are appropriate. The studies should be conducted by a
laboratory familiar with the filtration of colloids and dissolved solids from ground water
samples containing high concentrations of suspended solids. The following are some of a
number of questions which should be addressed by such studies.
What is the most appropriate type of filter and filter pore size in terms of
availability and applicability, given acidified conditions and the need to extricate
colloidal and dissolved metal particles from ground water samples?
What portion of the total metals concentration of a sample is associated with
suspended solids greater than 5 microns and greater than 2 microns in diameter?
Do colloids represent a significant amount of the metals concentration of a
sample (excluding dissolved solids)?
Does the cooling of acidified metals samples significantly reduce the breakdown
of suspended solids?
11-405
-------
Arc the quality control problems associated with field filtration too great to
warrant field filtration in lieu of laboratory filtration?
Do the metal concentrations degrade significantly during transport between the
field and the laboratory and during temporary storage prior to filtration?
If samples are not analyzed immediately in a fixed laboratory, will the suspended
solids be significantly degraded, given acidification and cold (4ฐC) storage?
RAMIFICATIONS OF SUGGESTIONS
The use of filtration to separate the colloidal and dissolved metal fractions for analysis, thereby
removing large suspended solids from the sample, represents a "forced" control over a sampling
problem which may not be controllable in the field. The resultant data, derived from analysis
of the filtrate, would be more representative of conditions representing background and site
contamination.
Requiring filtration would eliminate the occurrence (though infrequent) when samples were
collected to represent background and site contamination, where one of the samples was filtered
and the other sample was not filtered.
Data from filtered ground water samples are more appropriate for comparison against drinking
water benchmarks. This comparability of sample comparisons applies to both surface water
as well as ground water.
The use of filtration recognizes the fact that under the screening conditions of a site inspection,
the problem of the entrainment of suspended sediments in ground water metals samples is not
easily solved by quality control procedures. The time and resource constraints of a site
inspection also may preclude field filtration in lieu of filtration in a fixed laboratory.
Filtration will increase the number of and types of metals samples to be collected and tracked.
For example, hi surface water, filtered and unfiltered metals samples will need to be collected
at each sampling point through the use of split samples. In Karst aquifers, metals samples
would not be filtered. For surface water, the data user must be assured that filtered samples
are compared with filtered samples and vice versa. All reported water data will need to be
flagged with respect to whether the samples were filtered or unfiltered.
The requirements for site inspection personnel would be increased through the implementation
of a filtration policy. Field acidification would require verification. Samples will need to be
cooled and maintained at a temperature of 4ฐC and rapid transport to the laboratory assured.
Improved quality assurance and quality control requirements relating to purging and sampling
may be required.
Contracts with fixed laboratories may need to specify a new pre-treatment protocol in the
statement of work for inorganic analyses. The new laboratory procedure would shorten the
holding time for metals samples from 180 days to less than 48 hours, resulting in a marked
change in routine laboratory procedures.
The removal of colloidal particles represents a new filtration practice involving more extensive
quality control procedures. Although the precedent of filtration is firmly established, the
separation of colloidal particles would represent a new way of thinking in contrast with the
traditional viewpoint of "dissolved" versus "suspended" solids.
11-406
-------
SUMMARY
A new approach to ground water sampling and analysis of metals is proposed. The approach
involves the separation of colloids and dissolved solids from ground water samples by means
of filtration within a fixed laboratory. This approach will not eliminate all uncertainties, but
represents a "forced" control over a sampling problem which may not be controllable in the
field during initial screening investigations of an uncontrolled hazardous waste site.
REFERENCES
American Public Health Association, American Water Works Association, and the Water
Pollution Control Federation. 1989. Standard Methods For the Examination of Water and
Wastewater, 17th Edition. Washington, DC
Blatt, Harvey, Gerard Middleton, and Raymond Murray. 1972. Origin of Sedimentary Rocks.
Prentice-Hall, Englewood Cliffs, New Jersey
Bloese, Rod. 1983. Field Filtering Ground Water Samples. Internal EPA memorandum to Joe
Petrilli, dated Marck 22, 1983
Brown, Richard D. and David E. Egan. 1989. Site Assessment Media-Specific Considerations:
Lessons Learned from a Data User (Seminar presentation). Detailed outline published in
Proceedings of The 6th National Conference On Hazardous Wastes and Hazardous Materials,
The Hazardous Materials Control Research Institute, Greenbelt, Maryland
Champ, D.R., W.F., Merritt, and J.L. Young. 1982. Potential for Rapid Transport of Pu in
Ground Water as Demonstrated by Core Column Studies. In, Scientific Basis for Radioactive
Waste Management, Vol.5. Elsevier Science Publishers, New York
Diehl, Harvey. 1970. Quantitative Analysis: Elementary Principles and Practice. Oakland
Street Science Press, Ames, Iowa
Driscoll, Fletcher G. 1989. Groundwater and Wells (2nd. Ed.). Johnson Filtration Systems
Inc., St. Paul, Minnesota
Enfield, C.G. and G. Bengtsson. 1988. Macromolecular Transport of Hydrophobic
Contaminants in Aqueous Environments. Ground Water 26(1): 64-70.
Gschwend, P.M. and M.D. Reynolds. 1987. Mono-disperse Ferrous Phosphate Colloids in An
Anoxic Ground Water Plume. J. of Contaminant Hydrol. 1: 309-327.
Keith, Lawrence H. 1991. Environmental Sampling and Analysis: A Practical Guide. Lewis
Publishers, Chelsea, Michigan
Keith, Lawrence H. 1990. Environmental Sampling: A Summary. Environmental Science and
Technology 25(5):610-617
Keith, Lawrence H. 1988. Principles of Environmental Sampling. American Chemical Society,
Washington, DC
Nielsen, David. 1991. Practical Handbook of Ground-Water Monitoring. Lewis Publishers,
Chelsea, Michigan
11-407
-------
Puls, R.W. 1990. Colloidal Considerations in Groundwater Sampling and Contaminant Transport
Predictions. Nuclear Safety 31(1): 58-65
Puls, Robert W. and Michael J. Barcelona. 1989a. Filtration of Ground Water Samples for
Metals Analysis. Hazardous Waste & Hazardous Materials 6(4): 385-393
Puls, Robert W. and Michael J. Barcelona. 1989b. Superfund Ground Water Issue: Ground
Water Sampling for Metals Analyses. EPA/540/4-89/001. Center for Environmental Research
Information, U.S. Environmental Protection Agency, Cincinnati, Ohio
Puls, Robert W. and Hinrich L. Bonn. 1988. Sorption of Cadmium, Nickel, and Zinc by
Kaolinite and Montmorillonite Suspensions. Soil Sci. Amer. J. 52(5): 1289-1292
Stumm, W. and J. Morgan. Aquatic Chemistry: An Introduction Emphasizing Chemical
Equilibria in Natural Waters (2nd ed.). John Wiley and Sons, Inc., New York
Tillekeratne, S., T. Miwa, and A. Mizuike. 1986. Determination of Traces of Heavy Metals
in Positively Charged Inorganic Colloids in Freshwater. Mikrochimica Acta B: 289-296.
U.S. Environmental Protection Agency. 1989. Risk Assessment Guidance For Superfund:
Human Health Evaluation Manual, Part A. EPA/540/1-89/002. Office of Solid Waste and
Emergency Response, Washington, DC
U.S. Environmental Protection Agency. 1987a. Statement of Work: Inorganic Analyses.
Contract Laboratory Program, Washington, DC
U.S. Environmental Protection Agency. 1987b. Handbook: Ground Water. EPA/625/6-87/016.
Robert S. Kerr Environmental Research Laboratory, Ada, Oklahoma
U.S. Environmental Protection Agency. 1983. Characterization of Hazardous Waste Sites - A
Methods Manual: Volume II. Available Sampling Methods. EPA-600/4-83-040. Environmental
Monitoring Systems Laboratory, Las Vegas, Nevada.
U.S. Environmental Protection Agency. 1976. Manual of Methods for Chemical Analysis of
Water and Wastes. EPA-625-/6-74-003a. Environmental Monitoring and Support Laboratory,
Cincinnati, Ohio
1-408
-------
QO DETERMINATION OF TARGET ORGANICS
90 IN AIR USING LONG PATH SPECTROSCOPY
Richard D. Spear Ph.D. 0)
Pamela D. Greenlaw (2), Raymond J. Bath Ph.D. (2)
The EPA Region II, Environmental Services Division (ESD),
Surveillance and Monitoring Branch (SMB) has recently acquired a
transportable system to perform long path remote sensing of air
contaminants. This remote sensing system consists of spectrometers
which identify and guantify target organic chemicals in ambient air
Pathlengths, up to 500 meters, are defined by use of a
retroreflector, a specially constructed mirror assembly which
reflects and collimates the signals generated by the spectrometers.
The spectrometers used are: a Fourier transform infrared (FTIR);
with a resolution of 0.5 cm-(1) and a liquid nitrogen cooled mercury-
cadmium-telluride (MCT) photodetector and a long path ultraviolet
(LPUV) with a prism monochromator and a photo diode array detector.
With meteorological monitoring, this system can be used to monitor
the air for many environmental applications: site investigations
for Hazardous Ranking System (HRS); fenceline monitoring of
industrial sites; off-site health and safety monitoring during
remediation or removal projects; monitoring of lagoons for
potential air release; and in emergency response to community
complaints on air quality. This paper will present the design,
application and interpretation of data for the EPA Region II,
ESD/SMB, LPUV/FTIR.
(1) U.S. Environmental Protection Agency, Region II, Edison, NJ 08837
C2) NUS Corporation, 1090 King Georges Post Road, Edison, NJ 08837
11-409
-------
99 MEASUREMENT OF TOXIC ORGANIC COMPOUNDS IN LANDFILL GAS
SAMPLES USING CRYOGENIC TRAPPING AND FULL SCAN GC/MS
Steven D. Hoyt
Environmental Analytical Service
170-C Granada Drive
San Luis Obispo, CA 93401
(805)541-3666
A Nutech automated cryogenic concentrator with adjustable sample volume
loops is used for analyzing landfill gas samples using full scan GC/MS and selected
ion monitoring (SIM). This method is able to quantitate VOC compounds over
concentration ranges of 0.5 ppbv to 1000 ppbv. Landfill samples can be effectively
collected in evacuated SUMMA passivated canisters and most VOC compounds
have a holding time of 14 days. A 0.5 to 500 ml landfill gas sample is loaded into the
Nutech Automatic Concentrator and then analyzed with an HP 5890 GC using a 30
meter DB-5 fused silica capillary column connected directly to the source of an HP
5790 MSD. The capillary column is temperature programmed from -40 to 150 C to
analyze compounds from F-12 to trichlorobenzene. The relative standard deviation
for the method is less than 10% for most compounds and the MDL is about 0.5 ppbv
depending on sample size and the carbon dioxide content of the sample. The
sampling methods, instrument modifications for analyzing landfill gases will be
discussed along with the examples of data, and the limitations of the method.
11-410
-------
100 THE DETERMINATION OF THE HEAT OF COMBUSTION AND
WATER CONTENT OF INCINERATOR FEEDS USING
NEAR INFRARED SPECTROSCOPY
Dr. Nilesh K. Shah. Senior Analytical Chemist, Methods Development, Dr. Peter A.
Pospisil, Manager, Methods Development, Rita A. Atwood, Analytical Chemist,
Chemical Waste Management Inc., Technical Center, 150 West 137th Street,
Riverdale, Illinois 60627;
Dr. David L. Wetzel, Research Analytical Chemist, Arnold J. Eilert, Associate
Analytical Chemist, Kansas State University, Schellenberger Hall, Manhattan,
Kansas 66502
ABSTRACT
Near Infrared Spectroscopy (NIR) allows the simultaneous determination of the
heat of combustion and moisture content of a broad range of heterogeneous
incinerator feeds, with no sample preparation.
RCRA regulations require the determination of the heat of combustion on all
incinerator feeds to determine if they are above the 5000 BTU/lb level. Water
content is necessary for proper operating conditions of the incinerator. To satisfy
these requirements, a large number of samples are currently analyzed using both
bomb calorimetry and Karl Fischer titration, which are labor intensive and time
consuming methods. The NIR procedure utilizing selected absorption bands
eliminates all sample preparation, while simultaneously determining both
parameters.
NIR technology was used to generate heat of combustion and moisture data on 73%
of 564 incinerator feeds at a 90% success level, subsequent to software screening to
classify the incinerator feeds into physico-chemically unique types. The 73% can be
increased to 95% and the success level increased, by consolidating feed type
calibration curves and by improving the prescreening software. Additional
parameters may be added as the database is expanded. The runtime of two minutes
per sample entails an 80% analytical cost savings.
INTRODUCTION
RCRA regulations require the determination of the heat of combustion on all
incinerator feeds to determine if they are above the 5000 BTU/lb level. Chemical
Waste Management Inc. incineration facilities receive a broad range of liquid
hydrocarbon-based wastes requiring incineration. Incinerator feed type
compositions cover very wide ranges of constituents with heats of combustion
ranging from 1,000 to 20,000 BTU/lb, water contents from 0.1% to 100% and
halogen contents from 0.1% to 70%. Heat of combustion and water content are
critical sample composition parameters that affect incinerator performance and
blend feeds before incineration. Because these analytical parameters critically
affect incinerator performance and efficiency, each feed requires chemical analysis
using conventional bomb calorimetry and Karl Fischer titration methods, which are
both labor intensive and time consuming.
1-411
-------
Page 2
Near infrared reflectance (NIR) spectroscopy is a rapid and sensitive measurement
technique that has found many applications in analyzing agricultural and
pharmaceutical materials [1,2]. The near infrared spectral region, which spans from
1100 to 2500 nm, has high information content in the form of many overlapped
bands arising from overtones and combinations of X-H stretching modes of
vibration. The widespread use of NIR spectroscopy can be greatly attributed to the
introduction of powerful computerized data processing techniques for interpretation
of complicated NIR spectra. The development of this technique for quantitation is
primarily due to the availability and use of multilinear regression analysis; however,
quantitation is limited to samples of controlled composition. The qualitative
information available in the NIR spectral region is used by pattern recognition
techniques for the identification and classification of samples of unknown origin.
This paper reports a classic example of using near infrared spectroscopy and
chemometrics methods for analyzing hazardous wastes. Because of the nature and
spectroscopic complexity of hazardous wastes, a two-step chemometrics approach
must be used to successfully extract useful information from the near infrared
spectra. The first step is to extract qualitative spectroscopic features from the near
infrared spectra for pattern recognition analysis. The second step, then, is
quantitation of heat or combustion and water content for multivariate calibrations.
Mahalanqbis distance pattern recognition analysis is used to develop the
classification models from near infrared spectra. Multivariate calibration models
are developed by multilinear regression analysis for each of the defined classes. A
reasonable degree of accuracy is obtained in predicting the heat of combustion and
water content of liquid incinerator feeds provided appropriate calibration is used.
THEORY
Symbols and Notations
The following discussion explains the symbols and notations used in this paper to
describe the theory of Mahalanobis distance pattern recognition analysis and
Multilinear regression analysis. Bold letters are used to denote matrices and lower-
case letters to denote scalars (italic) and vectors (bold). A vector is always a column
vector if no transpose is attached. Transposed vectors are denoted by single quote
('). The symbol x is for a NIR spectrum and c is the concentration of the chemical
constituent of interest. In addition, i is the number of training set samples
(observations) and k is the number of spectral values (wavelengths). With this
notation, the model consists of i observations of k dimensions and the two sets of
data are denoted by c and X.
The training set is defined as the samples that are used to develop the classification
and calibration models. The test set is defined as the samples that are used to
evaluate the classification and calibration models and are samples that are not used
in the training set.
11-412
-------
Page3
M ahalanobis distance pattern recognition analysis
The Mahalanobis distance technique assumes a multivariate normal distribution
N(p, E) for the class population. The class model consists of a single point in
multidimensional space, the class centroid /*. The distance between an zth sample,
x,-, to the centroid is given by the generalized squared distance:
MDJ = (x../0'E-1 (x.-,,) ................................................. 1
where E is the training set's variance-covariance matrix which explains the
dispersion of data around the centroid. In practice, the true centroid /* and the
variance-covariance matrix E of the class population are unknown and, therefore,
must be estimated by the mean vector ~x and the variance-covariance matrix S from
a sample training set of n. The sample Mahalanobis distance can then be calculated
from equation 2:
MD? = (x; -It)' S"1 (x. -Ic) ................................................ 2
and
s =
Geometrically, the Mahalanobis distance class model is an ellipsoid-shaped cluster
with the population mean at its centroid. A spectrum is classified as a member of a
group if the Mahalanobis distance is less than 6 as compared to the Mahalanobis
distance for that sample with other groups. An excellent review of the theory of
Mahalanobis distances is given by Mark and Tunnell [3].
Multilinear Regression Analysis
Regression analysis is used for predicting BTU values from a collection of
independent variables such as wavelengths. The procedure consists of two phases:
calibration and prediction [4]. A data matrix is constructed from the NIR
instrument response X (absorbance) at selected wavelengths for a given set of
calibration samples. A vector of heat of combustion values c is then formed using
an independent method such as bomb calorimetry method.
One of the objectives of the calibration phase is to develop a model that relates the
NIR spectra to the heat of combustion values obtained by the bomb calorimetry
method. In regression analysis, a linear combination of the variables in X is
calculated such that the model's estimates of the heat of combustion values of the c
in the calibration set are as close to the known values of c as possible (minimizes the
errors in reproducing c). Mathematically, the linear regression model with a single
response (BTU value) can be explained by equation 4:
-------
Page 4
where ft is a vector of regression coefficients and c is a vector of errors or residuals
associated with the regression model. To "fit" the model given in equation 4 to the
known values of c, we must determine the values for the regression coefficients /9
and the residual e consistent with the available data.
The method of least squares selects regression coefficients estimates, p, using
equation 5:
p = (X'X)'1 X'c 5
and the estimated response c using equation 6:
a =
The regression estimates p are consistent with data whose sum of squared
differences (ซ ) from the observed c is as small as possible.
n
where deviations are the residuals estimate, c/ is the multilihear regression
estimate of the BTU value and cg is the actual BTU value for the Ith sample.
EXPERIMENTAL
564 samples received from various Chemical Waste Management sites were
analyzed by a Bran+Luebbe (Technicon) NIR 400 filter instrument in the
dispersive reflectance mode. About half of the samples were also analyzed by a
scanning Bomem MB155 FTIR/NIR instrument in transmission mode. Most of the
samples contain some amount of solid material, which settles after a period of
standing. A few also had two liquid phases. When necessary the samples were
shaken or sonicated to minimize j>hase separation. A QA/QC program has also
been developed for the Bran+Luebbe NIR spectrometer during the method
development process.
A Bomem MB155 FTIR/NIR connected to a Compaq 386 20 Mhz personal
computer was used to acquire NIR spectra. The absorbance data were collected in
the NIR spectral range from 10,000 to 4,000 cms'1 (or 1100 to 2500 nm). Sixteen
scans at 8 cms'l resolution were averaged for Fourier data processing. Using the
complete NIR spectrum range provided visual information for identifying spectral
patterns responsible for C-H and O-H overtone bands.
The data for pattern recognition and the calibration models were acquired using a
Bran+Luebbe NIR 400 filter instrument and a Compaq 386 20 Mhz computer. The
11-414
-------
PageS
Bran+Luebbe 400 instrument consists of 19 filters mounted on a filter wheel which
were configured with different wavelengths. The NIR spectral region selected for
the present work ranged from 1600 to 2450 nm since most of the useful spectral
information was present in this region. The samples were analyzed using a stainless
steel cell covered with a quartz plate to obtain a thin layer of film which was
measured in reflectance mode by the filter instrument.
Customized software written in Microsoft Quick Basic program was used for the
Technicon 400 instrument data acquisition. The data were then imported to
Bran+Luebbe IDAS software for developing Mahalanobis distance pattern
recognition and multilinear regression calibration models. Two separate equation
files, written in ASCII format, were read by a custom software to predict samples of
unknown origin. The custom software first classified the "spectroscopic type" of the
sample based on the Mahalanobis distance pattern recognition analysis and then
used the appropriate calibration to obtain a quantitative results for heat of
combustion and water content.
RESULTS AND DISCUSSION
Conventional Methods Overview
The standard technique for determining the heat of combustion of liquid incinerator
feeds is the bomb calorimetric method. The heat of combustion, measured in
British Thermal Units per pound (BTU/lb), is determined by burning a previously
weighed sample in an oxygen calorimeter under controlled conditions. The energy
required to raise the temperature of a given volume of water is measured by
observing the temperature before firing the bomb and after a stable temperature is
reached. These observations are made and recorded by the calorimetry apparatus
which also reports the heat of combustion (BTU/lb).
The standard technique for determining the water content of liquid incinerator
feeds is the Karl Fischer titration method. The percent water content is determined
by titrating a known amount of sample with standardized Karl Fischer Reagent
(KFR) to its endpoint. When there is an excess of KFR, the solution color changes
to a dark brown due to presence of free iodine. The Karl Fischer reagent is
standardized by titrating KFR with a known amount of water. Using an automatic
titrator, the endpoint of the reaction can also be electrometrically determined.
NIR Spectroscopy
Near infrared spectroscopy is based upon molecular heteroatom vibrations
producing a charge distribution, which interacts with electromagnetic radiation.
The interaction intensity is directly proportional to the dipole moment of the
molecular bond, and produces the characteristic absorption patterns representative
of the chemical composition of the sample. The mid-infrared region (25.0 to 2.5 pm
or 400 to 4000 cms"1) is the most well known range of analysis of organic materials.
The sharp spectral bands produced by the fundamental vibrational frequency of the
heteroatom bonds are directly related to skeletal and functional structures of
organic compounds. Near infrared absorption bands are produced by vibrational
overtones, and for each mid-infrared band there are four to seven near infrared
11-415
-------
Page 6
overtones. This crowding produces broad plateaus arising from superimposed
harmonics. The loss of structural information is offset by a gain in signal to noise
ratio. This makes NIR spectroscopy especially suited for the analysis of hazardous
waste liquid incinerator feeds.
The near infrared region (1100 to 2500 nm) is attractive for heat of combustion and
water analysis because most of the absorption bands observed in this region arise
from overtones and combinations of C-H and O-H stretching vibrations. Near
infrared spectra of three types of liquid incinerator feeds are shown in Figure 1.
The spectra show prominent bands for each type of incinerator feeds. For example,
type 6 feeds have a broad band at 1940 nm that is characteristics of O-H stretching
and second overtone vibration. Absorption bands are particularly strong above 2300
nm due to presence of two or more types of hydrogen bonded molecular complexes.
Type 1 feeds are primarily fuel oil (hydrocarbon) types of hazardous waste and,
therefore, the NIR spectra of such type materials contain a broad C-H overtone
band around 1720 nm. PCB type of materials are responsible for peaks at 1650 and
2175 nm in type 2 feeds. The absorptivity of these bands is largely independent of
the remainder of the molecule, but does depend on the concentration of the
absorbing functional group and, therefore, can be used for predicting the heat of
combustion and water content of the liquid incinerator feeds.
1.5 -
1 -
Near Infrared Spectrum
of Incinerator Feed Types
16OO
iaoo
8OOO
Wavelength (nm)
24 OO
Figure 1: Near Infrared Spectra of Incinerator Feeds
Incinerator feeds have been identified into seven types based upon their NIR
spectral patterns. Table I is a summary of matrix types responsible for the seven
groups of incinerator feed types. The distribution of the seven groups of incinerator
feeds analyzed by the NIR spectroscopy is shown in Figure 2.
11-416
-------
Page?
Table I: Summary of matrix types for pattern recognition analysis
Feed Type
1
2
3
4
5
6
7
Matrix Type
Fuel Oils
Haloeenated solvents (e.\ PCB)
Unknown
Aromatic solvents
Unknown
Polar solvents (e.g , methanoH ??
Aqueous solution (e.g.. water)
50.89%
14.54%
7.98%
2.13%
14.18%
Di
D2
4
5
17
Figure 2: Distribution of seven feed types analyzed by NIR spectroscopy
Near infrared analysis depends on the development of an empirical linear equation,
in which the constituent concentration is related to some combination of optical
measurements, usually expressed in absorbance or reflectance. To use this
empirical approach, the analyst must have a set of samples having known values
generated by another method (training set samples). From this set of knowns, the
system is trained through an iterative process. Using regressive and correlative data
processing, the analyst generates a multiterm linear expression making suitable use
of the analytical data. With sufficient experimentation and statistical treatment of
the data, this produces a final working calibration curve.
Mahalanobis Distance Pattern Recognition Analysis
In the Mahalanobis distance classification technique, two or more wavelengths are
used for classification of samples. The classification of spectra was based on the
generalized square distance of an observation from the centroid of a cluster. In
addition, only one mathematical model was constructed for all incinerator feed
types. In our present work, four wavelengths gave adequate discrimination to
identify seven groups of incinerator feeds based on their NIR spectral patterns and
Mahalanobis distance pattern recognition analysis. In Figure 3, a three dimensional
11-417
-------
PageS
plot of wavelength 2100, wavelength 2139 and wavelength 2348 shows good
discrimination between various classes. Adding a fourth wavelength 2310, the
pattern recognition model was able to classify various feed types more accurately.
The legends 1... 7 used in Figure 3 are explained in Table I.
3D Scatter Plot
of Incinerator Feeds
WL 21001.2
OS
/'T~M^
4 *y
ซ 4 ?ป*
"^ * 'Xli/Jj
'^S4i
4 1
4ฅ*-
ki ซ....
1.2
WL2139
0.4
Figure 3: 3-Dimensional scatter plot of wavelength 2100 vs wavelength 2348 and
wavelength 2139. See Table I for explanation of legends
The mathematical model for the Mahalanobis distance pattern recognition consists
of two matrices: the group-mean matrix and the mversed pooled variance-
covariance matrix. Using the model, the Mahalanobis distances between groups
were calculated for the training set data from which the model was developed. In
addition, greater the Mahalanobis distance between groups, the greater the
difference in their patterns. The results for Mahalanobis distances between groups
are summarized in Table n. According to Table II, only group 1 and group 4 are
close to each other, suggesting a similarity in spectral patterns between them.
Table II: Mahalanobis distances between groups used in the training set
to
from
Group 1
Group 2
Grouo 3
Grouo 4
fJrQUD 5
Group 6
Group 2
5.2952
Group 3
7.1974
5.8535
Group 4
32439
4.9845
9.0612
Group 5
62445
5.4749
7.5224
5.5473
Group 6
61238
7.7708
11.6443
5.0761
10 2863
Group 7
7.3523
8.4692
97214
7.6695
10.2894
6.1324
11-418
-------
Page 9
Multilinear Regression Analysis
Regressions were performed on the absorbance data (log 1/R) without any prior
data pretreatment. Individual models, and the corresponding regression
coefficients, were developed and evaluated to achieve maximum inference from the
regression analysis. Three types of calibration curves for three groups of liquid
incinerator feeds; type 1, type 2 and type 4, have been developed for heat of
combustion and water content determination. These three groups comprise about
73% of liquid incinerator feeds analyzed by this technique. Table III summarizes
the regression statistics on each constituent for the three incinerator feed types.
Table III: Regression statistics for heat of combustion and water content equations
Tvpel BTII
H2O
Tvpe2 BTU
H20
Tvpe4 BTU
H20
F ratio
295.479
40.517
162.045
25.431
23.735
31.462
Corr Coef
0.968
0.821
0.950
0.768
0.861
0.890
SEE (W\
0.645
2.229
0.575
0.345
1.421
0.916
SEP (%\
0.730
2.301
0.589
0.398
1.436
0.932
Range*
Ilr000-20r000
0.1-10.0
5000-11 000
0.1-1.0
9.000-18rOOO
0.1-10
Range for BTU is BTU/lb and % moisture for water content
In general, the regression statistics given in Table III are used to evaluate the
validity of the regression model. The F-ratio for regression is a quality measure for
the regression that puts an overall goodness of the regression into one number. A
high value of "F1 is indicative of a good fit obtained from many samples with a small
number of wavelengths. These kinds of calibrations will be more robust against
small variations and time. The multiple correlation coefficient is a measure of error
versus total variation and should tend to unity. For a given range, the standard of
error of estimate (SEE) and standard error of prediction (SEP) evaluates the
calibration and the prediction model and should be as small as possible.
Besides evaluating the regression statistics given in Table III, the residuals must also
be examined to evaluate the adequacy of the regression model. Figure 4 is a plot of
NIR predicted heat of combustion (BTU/lb) values against the actual heat of
combustion values obtained using bomb calorimetry procedure for type 1
incinerator feeds. All sample informations on lack of fit is contained in the
residuals.
11-419
-------
Page 10
10 12 14 16
Actual BTU
18
20
Figure 4: Plot of NIR predicted BTU versus actual BTU from Bomb Calorimetry
If the regression model is valid, the residuals are the estimates of the model error,
which are assumed to have a normal distribution around the mean (/z = 0) and
constant variance. Figure 5 is a plot of residuals against the predicted BTU values
for type 1 feeds. According to Figure 5, the residual plot for type 1 feeds has a mean
equal to zero and a constant variance, suggesting the robust nature of the calibration
curve. Residual plots were also evaluated for type 2 and type 4 feeds before using
the calibration models to predict the heat of combustion and the water contents of
incinerator feeds.
Predicted BTU
Figure 5: Plot of residuals versus predicted BTU for type 1 incinerator feed
QA/QC Procedure
Instrument performance parameters must be evaluated on a daily basis to
demonstrate that the instrument is performing properly. Two instrument diagnostic
checks and one instrument performance standard were developed for QA/QC
procedure. The two instrument diagnostic checks are checking for the front end
board of the instrument and the amount of light energy passing through the filters.
11-420
-------
Page 11
Kerosene was chosen as the instrument performance standard because of its
consistency and wide use in bomb calorimetry procedure.
In addition to the instrument reliability, the method must also be proved reliable.
Method performance is monitored throughout the day by using a quality control
(QC) check sample. A QC check sample is a material which represents the sample
matrix being analyzed. Sample duplicate and fortified samples are used to measure
the precision and accuracy or the method.
CONCLUSIONS
NIR technology can be used for analyzing incinerator feeds for heat of combustion
and water content. About 73% of the incinerator feeds have been successfully
analyzed by the NIR technology. Additional calibration curves will increase the
percent of samples analyzed as the database is expanded. The elimination of
conventional bomb calorimeter and Karl Fischer titration for sample preparation
drastically reduces the analytical costs by streamlining sample analyses. The
runtime of two minutes per analysis entails an 80% cost savings.
REFERENCES
1. Wetzel, D.L.;AnaL Chem., 1983,55,1165A-1176A.
2. Shah, N.K.; Gemperline, PJ.;Anal Chem., 1990,62, 465-470.
3. Mark, H.L.; Tunnell, D.;AnaL Chem., 1985, 57,1449-1456.
4. Johnson, R.A; Wichern, D.W.; Applied Multivariate Statistical Analysis, Prentice
Hall: NJ, 1988, pp 300-314.
1-421
-------
SOURCE SAMPLING AND ANALYSIS GUIDANCE
A METHODS DIRECTORY1
Merrill D. Jackson and Larry D. Johnson, Quality Assurance
Division, Atmospheric Research and Exposure Assessment
Laboratory, U. S. Environmental Protection Agency, Research
Triangle Park, North Carolina 27711; Kim W. Baughman, Ruby H.
James and Ralph B. Spafford, 2000 Ninth Avenue South, Southern
Research Institute, Birmingham, Alabama 35255
ABSTRACT
Sampling and analytical methodologies are needed by EPA and
industry for testing stationary sources for specific organic
compounds such as those listed under the Resource Conservation
and Recovery Act (RCRA) Appendix VIII and Appendix IX and the
Clean Air Act of 1990.
A computerized directory, Problem POHC Reference Directory,
has been developed that supplies information on available
field sampling and analytical methodology for each compound in
those lists. Existing EPA methods are referenced if
applicable, along with their validation status. At the
present, the data base is strongly oriented toward combustion
sources. The base may be searched on the basis of several
parameters including name, Chemical Abstracts Service (CAS)
number, physical properties, thermal stability, combustion
rank, or general problem areas in sampling or analysis. The
methods directory is menu driven and requires no programming
ability; however, some familiarity wit dBASE III+ would be
helpful.
INTRODUCTION
There are a large number of chemical compounds listed under
Appendix VIII1 and Appendix IX2 of RCRA and the Clean Air Act
of 19903, that are regulated by the U.S. Environmental
Protection Agency (EPA). EPA has several sampling and
analytical methods which are validated for many of these
compounds. Other of the listed compounds may be analyzed by
these methods but they have not been validated. EPA or State
permit writers and industry personnel may not be familiar with
each compound and its status; therefore, a data base of each
compound listed along with its methodology, has been prepared.
If the methodology has been validated for a compound, a
reference is given; however, if no method has been validated,
1. This paper has been reviewed in accordance with the U.S.
Environmental Protection Agency's peer and administrative
review policies and approved for presentation and publication.
The mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
11-422
-------
the best method to try is indicated. Since the data base was
originally developed for use with incinerators, it has an
orientation towards combustion methodology.
COMPUTER AND SOFTWARE REQUIREMENTS
An IBM PC or compatible system with a hard disk using DOS 2.0
or higher is required. Version 1.0 of our program requires
have dBASE III+ in order to run and is available from the
National Technical Information Service (NTIS) under the name
"Problem POHC Reference Directory"6. This version contains
only the compounds listed under RCRA in Appendix VIII.
Version 2.0 will also include the compounds listed under
Appendix IX and the Clean Air Act of 1990, and it is scheduled
to be released shortly. It will be titled "Source Sampling and
Analysis Guidance, Version 2.0" and will be available from
NTIS. We plan to have Version 2.0 in the compiled format;
therefore, this version of our program will not need a data
base program such as dBASE I11+ or IV in order to run.
DATA BASE CONTENTS
The following information for each compound is given if
available: (1) name of compound (The Appendix VIII name is
given first with either the Appendix IX or the Clean Air Act
name given next. If a common name that had not been used is
known, then it is given also.),(2) the CAS registry number,
(3) chemical formula, (4) molecular weight, (5) compound
class, (6) University of Dayton Research Institute (UDRI)
thermal stability class and ranking4, (7) heat of combustion,
(8) combustion ranking5, (9) boiling, melting and flash points
and water solubility, (10) information on toxicity, (11)
sampling and analysis methods, (12) validation status of the
compound in the methods, (13) general and specific problems,
(14) a description of the problems, and (15) solutions (if
known) . The data in the base is not complete by any means and
is constantly being revised. Yearly updates are planned.
RUNNING THE PROGRAM
The first screen seen after opening the program is the main
menu shown in Figure 1.
Selection of an option will start a new sequence. Selection
of option 1 will print the entire data base (warning: This
will take about 1.5-2 hours.). This option will probably be
used only once to provide a complete hardcopy of everything in
the data base; additional copies can be photocopied. Selection
of option 4 will print a list of all the compounds with their
CAS numbers and data base record number. This is a very useful
tool to have available since the data base record number is
11-423
-------
needed when using option number 2.
Figure 1. Opening Screen
MAIN MENU
1. PRINT ALL RECORDS IN DATABASE
2. PRINT A SPECIFIC DATABASE RECORD
3. LIST COMPOUNDS BY PHYSICAL PROPERTY,
THERMAL STABILITY, OR COMBUSTION RANK
4. LIST COMPOUNDS BY NAME AND/OR CAS REGISTRY NUMBER
5. LIST COMPOUNDS BY PROBLEM AREAS
6. EXIT
ENTER YOUR CHOICE (1-6) FOR THE ABOVE:
Using selection number 2 will bring up the Records Menu
(Figure 2).
Figure 2. Records Menu
PRINT A SPECIFIED DATABASE RECORD.
SPECIFY THE RECORD TO BE PRINTED BY:
1. RECORD NUMBER
2. COMPOUND NAME
3. CAS REGISTRY NUMBER
OR
4. EXIT TO MAIN MENU
ENTER YOUR CHOICE (1-4) FOR THE ABOVE:
Upon the entry of choice 1, 2, or 3, the question "DO YOU WANT
A HARD COPY OF THE DATA? (Y/N)" will appear. Selection "yes"
will create a printed copy, whereas a "no" answer will only
bring the data on screen. The search routine is such that the
record number is the fastest way to locate an entry; however,
if you do not know the data base record number, you may search
by either the name of the compound or its CAS Registry Number.
The Records Menu will probably be the most used menu since it
provides the complete information on a given compound.
Selection of the third option on the Main Menu brings up the
Specific Compounds Menu (Figure 3).
11-424
-------
Figure 3. Specific Compounds Menu
LIST COMPOUNDS ON THE BASIS OF:
1. UDRI THERMAL STABILITY CLASS
2. UDRI THERMAL STABILITY RANKING
3. MOLECULAR WEIGHT
4. BOILING POINT
5. MELTING POINT
6. COMBUSTION RANK
7. COMBINATION OF ANY TWO PROPERTIES,
8. RETURN TO MAIN MENU
After selecting any of options 1-6, the user will be prompted
to input a range for that option before again asking if he
wants a hard copy. Selection of number 7 will result in a
request for the two properties and the range for each
property. This search and listing option can be particularly
helpful in Principal Organic Hazardous Constituent (POHC)
selection for trial burns, since compounds can be listed by
incinerability category and by physical properties.
The fourth selection on the Main Menu (Figure 1) will provide
an alphabetical list of the compounds with the data base
number. This provides you with the easiest method of searching
with option number 1 of the Records Menu (Figure 2).
The Problem Menu (Figure 4) is selected from option 5 of the
Main Menu.
Figure 4. Problem Menu
1. LIST ALL PROBLEM COMPOUNDS
2. LIST COMPOUNDS BY GENERAL PROBLEM
3. LIST COMPOUNDS BY SPECIFIC PROBLEM
4. RETURN TO MAIN MENU
ENTER YOUR CHOICE (1-4) FOR THE ABOVE:
The first option will list every compound that is recorded to
1-425
-------
have any problem. The second choice brings up the screen shown
in Figure 5.
Figure 5. General Problem Types
1. ANALYSIS
2. HAZARDOUS
3. SAMPLING
SPECIFY GENERAL PROBLEM TYPE (1, 2, OR 3):
A selection here will list all problem compounds in the area
selected. The third choice on the Problem Menu probably is
the most useful one since it allows a more limited selection.
The menu which goes with the third choice is shown in Figure
6.
Figure 6. Specific Problem Types
GENERAL PROBLEM
SPECIFIC PROBLEMS
1. ANALYSIS
2. HAZARDOUS
3. SAMPLING
E. SENSITIVITY
F. RECOVERY
G. DECOMPOSITION
A. CHROMATOGRAPHY
B. INTERFERENCE
C. WATER SOLUBLE
D. BLANK
A. CORROSIVE
B. EXPLOSIVE
C. INCOMPATIBILITY
D. TOXIC
A. BLANK
B. BREAKTHROUGH
C. DECOMPOSITION
D. REACTIVE
SPECIFIED GENERAL PROBLEM TYPE (1,2, OR 3)
After the user selects the general type from the Specific
Problem Types menu, then the program prompts the user to
select a specific problem type from the selections on the
right.
11-426
-------
Samples of printouts of individual compounds showing the
actual information available are presented in Figures 7 and 8.
On Figure 7, points of interest are that hexachlorobenzene is
listed on both Appendix VIII and the Clean Air act of 1990 but
not on Appendix IX. It has a UDRI class and ranking. Only
compounds listed on Appendix VIII have UDRI ratings at the
present time. The record also indicates that we have several
areas not filled in yet. The data base is not complete, and
data will be added as we become aware of it. The sampling and
analytical methods for this compound are listed as suggestions
since they have not been validated. The heat of combustion is
listed for help in determining which compounds in a waste
mixture should be selected as POHCs. Figure 8 shows a fully
documented compound, benzene. The sampling and analytical
methods have been validated, and the references are given. The
specific problem type is a blank problem, and suggestions are
given on how to overcome this problem.
SUMMARY
A data base program listing sampling and analysis methods
along with several characteristics of each compound listed
under RCRA Appendix VIII, is available for use with dBASE
III+. The data base permits those personnel who need field
sampling and analytical procedures for regulation purposes to
have a single reference for this information. A second
version covering RCRA Appendix VIII, Appendix IX, and Clean
Air Act 1990 compounds will be available in late 1991. The
second version will be a compiled program, which will not
reguire any additional software (ie dBASE III+ or IV) to
operate.
REFERENCES
1. Code of Federal Regulations, 40, Part 261, Appendix VIII,
p 90-98, July 1, 1990.
2. Code of Federal Regulations, 40, Part 261, Appendix IX,
p 98-117, July 1, 1990.
3. Clean Air Act, Title III, Public Law 101-549, 1990.
4. Guidance on Setting Permit Conditions and Reporting Trial
Burn Results. Volume II of the Hazardous Waste
Incineration Guidance Series p 105-123, EPA-625/6-
89/019, January 1989.
5. Guidance Manual for Hazardous Waste Incinerator Permits,
Mitre Corp., NTIS PB84-100577, July 1983.
11-427
-------
6. Baughman, K.W., R.H. James, R.B. Spafford and C.H.
Duffey, Problem POHC Reference Directory, EPA-600/3-
90/094, January 1991.
Figure 7. Data Output for Hexachlorobenzene
RECORD NUMBER: 361 DATE OF LATEST ENTRY: 04/10/91
COMPOUND: Hexachlorobenzene
CAS REGISTRY NO: 118-74-1
FORMULA: C6-(C1)6
MOLECULAR WEIGHT: 284.80
COMPOUND CLASS: Chlorinated aromatic
APPENDIX 8? Y APPENDIX 9? N CLEAN AIR ACT OF 1990? Y
UDRI THERMAL STABILITY CLASS: 1
UDRI THERMAL STABILITY RANKING: 31
BOILING POINT, CELSIUS: 323
MELTING POINT, CELSIUS: 231
FLASH POINT, CELSIUS:
SOLUBILITY, IN HATER: Insol 0.035 ppm
HEAT OF COMBUSTION, KCAL/MOLE: 567.70
COMBUSTION RANKING: 65
TOXICITY DATA:
SAMPLING METHOD: SW-846 No. 0010 (MM5)
ANALYSIS METHOD:
SW-846 No. 8270 (Extraction, GC/MS)
VALIDATION STATUS:
GENERAL PROBLEM TYPE(S):
SPECIFIC PROBLEM TYPE(S):
DESCRIPTION OF PROBLEMS:
SOLUTIONS:
11-428
-------
Figure 8. Data Output for Benzene
RECORD NUMBER: 77 DATE OF LATEST ENTRY: 12/13/90
COMPOUND: Benzene
CAS REGISTRY NO: 71-43-2
FORMULA: C6-H6
MOLECULAR WEIGHT: 78.11
COMPOUND CLASS: Aromatic hydrocarbon
APPENDIX 8? Y APPENDIX 9? Y CLEAN AIR ACT OF 1990? Y
UDRI THERMAL STABILITY CLASS: 1
UDRI THERMAL STABILITY RANKING: 3
BOILING POINT, CELSIUS: 80.1
MELTING POINT, CELSIUS: 5.5
FLASH POINT, CELSIUS: -11.00
SOLUBILITY, IN WATER: Sol
HEAT OF COMBUSTION, KCAL/MOLE: 780.96
COMBUSTION RANKING: 47
TOXICITY DATA: Cancer suspect agent; flammable liquid
SAMPLING METHOD: SW-846 No. 0030 (VOST)
ANALYSIS METHOD:
SW-846 No. 5040 or Draft No. 5041{Therm. Desorb./P and Trap-GC\MS)
VALIDATION STATUS:
The VOST method has been validated for this compound (See "Validation
Studies of the Protocol for the VOST" JAPCA Vol. 37 No. 4 388-394, 1987).
(Also see "Recovery of POHCs and PICs from a VOST" EPA-600/7-86-025.)
GENERAL PROBLEM TYPE(S): Sampling
SPECIFIC PROBLEM TYPE(S): Blank
DESCRIPTION OF PROBLEMS:
Cancer suspect.
Blank problem with Tenax
Benzene is a common PIC. This may complicate interpretation of results,
and make it difficult to achieve acceptable ORE with low waste feed
concentrations.
SOLUTIONS:
Level of lab blank should be determined in advance. Calculations should
be based on waste feed concentration to determine if blank level will be
a significant problem. Benzene should not be chosen as a POHC at very low
waste feed levels because it is likely to make blank or PIC problem
significant.
1-429
-------
A FIELD INVESTIGATION OF GROUNDWATER MONITORING WELL
PURGING TECHNIQUES
Van Maltbv. Research Scientist, Jay P. Unwin, Regional
Manager, National Council of the Paper Industry for Air and
Stream Improvement, Inc., Central-Lake States Regional Center,
(NCASI), Western Michigan University, Kalamazoo, MI 49008.
ABSTRACT
A field investigation of commonly used monitoring well purging
techniques was conducted under different conditions including
type of pump, pump inlet location, and the use of packers.
Tracers including deionized water, fluorescent dye, and
lithium chloride were used to define the amount of stagnant
water at any given time in the pump discharge. Tests were
conducted in shallow 5 cm (2 in) diameter wells. The effects
of drawdown were examined.
All runs conducted in the absence of drawdown with the pump
inlet in a fixed position at or above the screen showed a
highly variable and unpredictable inclusion of stagnant water.
The use of packers did not completely prevent the inclusion of
stagnant water into the pump inlet. The inclusion of stagnant
water into a sample was minimized by purging from some dis-
tance above the screen followed by relocation of the pump
inlet into the screen for sample collection. In wells where
drawdown occurred during purging, stagnant water inclusion was
minimized by reduced pumping rates to allow for sample
collection during periods of well recharge. Real time
monitoring of indicator parameters such as pH, temperature and
specific conductance was not generally successful in indicat-
ing when purging was complete.
MONITORING WELL PURGING
It is generally recognized that the composition of the
stagnant water within a monitoring well above the screened
section is probably not representative of the overall ground-
water quality at the sampling site. The water standing in the
well casing is commonly referred to as being stagnant, that
is, the water has been isolated from the aquifer at least
since the last time the well was sampled. During that time,
the chemical quality of the stagnant water may have changed by
(a) direct introduction of foreign material into the well, (b)
interactions with the well casing or at the interface with the
atmosphere, or (c) biological activity. Even without such
alterations, the stagnant water would not reflect any changes
in the groundwater quality that may have occurred since the
last time the well was sampled. Because the investigator
cannot be certain which, if any, of these influences has
occurred or whether inclusion of some of the stagnant water in
a sample from the well would significantly change the conclu-
11-430
-------
sions that might be drawn from the data, the safe thing to do
is to prevent or minimize such inclusion.
One presently used purging technique presented as a coopera-
tive agreement by the Illinois States Water Survey (ISWS) and
Illinois State Geological Survey (ISGS) (1) is to pump the
well and calculate the percentage of water at any given time
in a pump discharge that can be attributed to drawdown. This
approach is based on the knowledge of time-drawdown charac-
teristics of a well and does not account for contributions of
stagnant water from any source other than drawdown. Another
commonly used purging technique presented by the U.S. Geologi-
cal Survey (USGS) (2) is to pump a well until indicator
parameters such as pH, temperature, and conductivity stabi-
lize. This approach ignores the possibility that a near
constant contribution of stagnant water into the sample may
result in stabilized readings for the observed parameters. It
also fails to account for contributions of stagnant water that
are too small to notably affect the measured parameters, but
which may significantly alter the outcome of an analysis.
Probably, the most commonly used purging practice is to purge
an arbitrary number of bore volumes (well casing) with little
or no regard to drawdown or indicator parameters.
This research reflects the need for documentary evidence
regarding the hydraulic behavior of a monitoring well during
purging. By examining truly trace concentrations, the
fraction of stagnant water entering a pump inlet can be better
defined as a function of bore volumes pumped (or time) and
inlet position.
CRITICAL REVIEW OF EXISTING INFORMATION ON MONITORING WELL
PURGING
Illinois State Water Survey and Illinois State Geological
Survey ISWS and ISGS have published "Procedures for the
Collection of Representative Water Quality Data From Monitor-
ing Wells" (1) which describes in detail guidelines for
monitoring well purging. The basic assumption made in this
research was that during the initial pumping of a small
diameter monitoring well, a significant fraction of the pump
discharge comes from stagnant water within the well casing.
This effect is due to drawdown. The procedure uses an
equation which develops time-drawdown data based on individual
monitoring well hydrologic data obtained during pump tests.
The resulting theoretical drawdown curve is used to predict
the time at which the effects of casing storage due to
drawdown become negligible. This curve is intended to be used
as a guideline, in conjunction with the observation of indica-
tor parameters for the selection of an appropriate pumping
rate and number of bore volumes to be pumped prior to sample
11-431
-------
collection.
Six monitoring wells at different locations within the State
of Illinois were used for the development of this purging
protocol. NCASI encountered several difficulties while
reviewing the results of the study on these six wells. These
difficulties included (a) reproduction of the theoretical
drawdown curves presented, and (b) interpretation and conclu-
sions drawn from the monitored indicator parameters. NCASI
and others have been unable to reproduce any of the published
theoretical drawdown curves for six wells examined by ISWS and
ISGS. For each well, ISWS and ISGS have provided theoretical
drawdown curves derived from the Papadopulos and Cooper
equations which generally show good agreement with the actual
drawdown curves presented. NCASI has used these equations in
the manner described by ISWS and ISGS to produce theoretical
drawdown curves which bear little resemblance to those
published. A careful examination by NCASI has not revealed
the reason for these discrepancies.
The cooperative agreement examined the effects of well purging
on the chemical composition for six monitoring wells. Five of
the six wells were described as having site specific limita-
tions which hindered interpretation of the results. The
single well (Site 5), in which a clearly indicated effect of
purging on indicator parameters was noted, directly contra-
dicted information from the pump test portion of the study.
In spite of limitations described for each of the six sites
examined, ISWS and ISGS concluded that "the chemical data from
this portion of the study have verified the theoretical ratios
of aquifer to stored water predicted during the pump tests".
A subsequent publication by the ISWS "Practical Guide For
Ground-Water Sampling (3) has endorsed the above mentioned
purging protocol.
United States Geological Survey The United States Geological
Survey (USGS) (2) states that in order to obtain a representa-
tive sample from an aquifer at a given location, a well must
be pumped until indicator parameters such as pH, temperature,
and conductivity are constant. Measurement of drawdown during
the purging period is recommended because changes in the indi-
cator parameters may reflect water from different zones of the
aquifer being drawn into the well. This procedure is stated
as the minimum required precaution for insuring that a sample
adequately represents the water quality in the aquifer.
Guidelines for indicator parameter stability have been
presented by Gibs and Imbrigiotta (4) . Research by Slawson
et al. (5) examined the variability of indicator parameters
and other constituents in well discharges during continuous
pumping. Appreciable changes were observed in several of the
11-432
-------
parameters, most notably conductivity. These changes were
attributed to naturally occurring vertical and horizontal
variability within the aquifer from natural influences. Such
a situation would not allow for the universal use of indicator
parameters to determine when a groundwater sample should be
collected.
Consideration should also be given to the general nature of
the specified indicator parameters. Conductivity and pH are
may be affected by changes in temperature and pressure during
sampling. Pressure changes can cause rapid degassing of
carbon dioxide and other gases that could affect sample pH and
specific conductance.
NCASI Laboratory Purging Investigations NCASI conducted a
laboratory investigation to examine factors other than
drawdown that could cause stagnant water to enter a pump inlet
(6) . All tests were conducted at constant head, thereby
disregarding the effects of drawdown. In the research,
stagnant water in a well column was spiked with a fluorescent
dye. Care was taken to minimize density differences induced
by either temperature gradients between stagnant and aquifer
water or concentration induced density gradients. The well
was then sampled with the pump inlet in various positions
while the tracer concentration in the pump effluent was
continuously monitored. Results demonstrated that an average
of about 2 to 4 percent of the water pumped from locations
above the screen and an average of about 1 percent of the
water pumped from within the screen of the monitoring well
came from the stagnant water located above the pump inlet.
The mechanism that caused the overlying stagnant water to
reach the pump inlet was not investigated, though mixing
caused by turbulence around the pump inlet was hypothesized.
University of Waterloo Robin and Gillham (7) conducted a
study using non-reactive tracers to judge the effectiveness of
various purging procedures. The results suggest a sharp
interface and little mixing between fresh water in the screen
or below a pump inlet and the stagnant water in the casing.
For wells not completely evacuated, pumping from immediately
beneath the air/water interface for 2 or 3 bore volumes was
deemed sufficient to collect a representative sample. Three
tracers were used for the study: deionized water, NaCl
(conductivity), and bromide. Of these, NaCl was demonstrated
to be an inappropriate tracer due to mixing caused by density
differences between the tracer and the fresh water. While
deionized water was determined to be an appropriate tracer
(verified with bromide) an even greater density difference
existed between deionized water and the natural groundwater at
the site. Although the deionized water was less dense than
the groundwater, the effect of this density difference may
1-433
-------
have acted to discourage mixing around the pump inlet.
EXPERIMENTAL PROCEDURE
Wells
Two shallow monitoring wells were selected for this research.
Both wells were constructed of 5-cm ID PVC (2-in) with commer-
cially available PVC screened sections (Timco 0.010 slot).
Both wells are situated in a shallow glacial unconfined
aquifer composed of sand and clay. Each well is approximately
9.2 m (30 ft) deep and has a standing water level within 1.5
m (5 ft) of the ground surface. Although these wells were
only approximately 62 m (200 ft) apart, local variability
within the aquifer accounted for marked difference in the
hydraulic performance of each well. At the purging rates used
in this research, one well experienced very minimal drawdown
(less than 0.6 cm, 0.25 in), and the other well experienced
extreme drawdown and could easily be pumped dry.
Equipment
Two pumps were used for purging in this research. The
majority of the purging runs were conducted using an above
ground peristaltic pump (Masterflex, #70-15 head) with a
maximum flow rate of approximately 800 ml/min. Several runs
were conducted with a submersible pump (Keck #84) with a flow
rate of approximately 4.5 L/min.
The fluorescent tracer concentration in the purging pump
discharge was detected with a fluorometer (Turner #111) with
a flow-through cell for continuous measurement. Conductivity,
drawdown, temperature, and pH were monitored continuously
using methods described elsewhere (7). Drawdown was moni-
tored with a submersible pressure transducer. All data from
the instruments with the exception of pH were recorded on a
portable computer. Values for pH were recorded manually due
to pH signal recording difficulties. A portable electric
generator provided electrical power where needed.
The amount of stagnant water in the pump discharge at any
given time was measured directly by the use of one or more of
the following tracers: Rhodamine WT, lithium (as lithium
chloride), or deionized water. Rhodamine WT was used as the
tracer of choice during this research. Rhodamine WT is a non-
toxic fluorescent xanthene dye commonly used in percolation
studies, potable water systems, and surface water systems.
Additionally, Rhodamine WT exhibits both low reactivity and
sorption tendencies (8) making it well suited for this res-
earch. The initial concentration after the dye had been added
11-434
-------
to the stagnant water was approximately 200 ppb. At this
concentration in water, the dye imparts no color detectable
with the naked eye. In several of the purging tests in the
well in which drawdown occurred, the presence of turbidity
interfered with the detection of the fluorescent dye. In this
situation, lithium was used as the tracer. For several of the
tests, deionized water was used as a tracer in conjunction
with Rhodamine WT.
General Procedures
At the start of each purging test, the static water level was
measured. Tracer was added to the stagnant water in the well
in a manner that resulted in a homogeneous concentration
within the stagnant water column, without migration into the
screen area. To accomplish this, an inflatable packer was
used to hydraulically isolate the screened portion of the well
from the cased portion above. The packer was designed as a
flow-through device so that water could be collected from the
screen area during the time the packer was inflated and in
position.
The stagnant water above the packer was pumped out and
collected in a container at the surface. Aquifer water from
the screen area was pumped to the surface at the planned
purging rate in order to zero the fluorometer and obtain
background readings for pH, conductivity and temperature.
With the fluorometer zeroed for the aquifer water, the
previously collected storage water was pumped through the
fluorometer in a closed loop system. Tracer was added to the
casing water until the fluorometer readout was 100 percent.
This casing water containing the tracer was poured back into
the well to a water level slightly below the static water
level so that the volume displaced by the packer would be re-
placed by aquifer water moving into the screened section
rather than by spiked water moving down into the screened
section when the packer was removed.
The stagnant water containing the tracer was kept isolated
from the screened section with the inflated packer for a
minimum of 12 hours to ensure that the undisturbed tempera-
ture-depth profile of the stagnant water would be re-estab-
lished. A preliminary investigation of the wells used for
this research, revealed that the temperature-depth profile
determined after a month of non-pumping, would be re-estab-
lished approximately 8 hours after having been disturbed.
A run was initiated by slowly deflating the packer and
carefully removing it from the well. The pump inlet was
placed at a predetermined depth for the run. Pumping and
automated data collection were started. Fluorometry readings
11-435
-------
were recorded every two seconds, other readings except pH were
recorded every five seconds. Readings for pH were recorded
every 105 seconds. The computer and software allowed for
real-time observation of data throughout the purging run.
At the end of a test, the pump inlet was raised to the
air/water interface while pumping continued until the fluorom-
eter readings returned to background. At this point new
tracer could be added for the next run.
For the runs where it was applicable, the number of bore
volumes pumped was based on the volume between the pump inlet
(bottom of sample line) and the top of the well screen. For
the runs where the pump inlet was placed immediately above the
screen top, the duration of a run was measured in the number
of liters pumped, rather than bore volumes. Because the
measured tracer concentration in the aquifer water was zero
for all runs, the fraction of the pumped volume that came from
the stagnant zone above the pump inlet is simply the ratio of
the measured concentration of tracer in the pumped water to
the initial concentration in the stagnant zone.
Two packers were investigated to determine their effect on
stagnant water concentration during purging and sampling.
Such packers form a seal against the inside of the well casing
to isolate the stagnant water above from the pump inlet. In
effect, the volume of standing water above the pump inlet is
reduced. One packer was a commercially available unit
attached to the top of the Keck pump. The other packer was
laboratory made and designed as a flow through device.
The effect of purging and sampling in a well experiencing a
significant degree of drawdown was investigated. During
drawdown, the balance of the water in the pump discharge not
accounted by flow into the well from the screen comes from the
stagnant casing water above the pump inlet. A sample was
collected from such a well by purging the well at a rate great
enough to produce drawdown and thus reducing the level of
stagnant water. Sample collection occurred at a reduced
flowrate during water level recovery. Purging and sampling in
this manner allows the stagnant water/aquifer water interface
to move upward and away from the pump inlet, reducing the
chance for stagnant water to become captured by the pump
inlet. As noted previously, due to excessive turbidity in
this well, lithium chloride at an initial concentration of 62
mg/L was used as the tracer.
The following purging configurations were investigated during
this research: (a) fixed pump inlet positions at approximately
5 cm (2 in) below the air/water interface, mid-casing, and at
the top of the well screen, (b) a comparison of Rhodamine WT
11-436
-------
and deionized water as tracers with the pump inlet reposi-
tioned between the screen and the air/water interface several
times during the run, (c) purging from above the screen,
followed by sampling within the screen, (d) packers, (e)
drawdown and recover. Details specific to each test are given
in the section below.
RESULTS AND DISCUSSION
Drawdown did not become significant during the following tests
(maximum 0.6 cm, 0.25 in) at the specified purging rates.
Peristaltic Pump. Inlet Near Static Water Level
Figure 1 graphically presents the results of a purging test
with the pump inlet located approximately 5 cm (2 in) below
the static water level in the monitoring well. The purging
rate during this test was 1.01 L/min.
During most of the time required to remove the initial bore
volume of water from the well (14.3 minutes), the concentra-
tion of stagnant water in the pump inlet was 100 percent. As
the column of fresh water moving up the well casing approached
the pump inlet, there was a corresponding rapid decrease in
the stagnant water concentration detected in the pump dis-
charge. During the removal of subsequent bore volumes the
concentration of stagnant water continued to decrease but was
still detectable in the pump discharge for a relatively long
time (9.0 bore volumes, 128 minutes). At this point the
majority of the stagnant water had been removed from below the
pump inlet, however, it may not have been entirely removed
from above the pump inlet. Earlier research (9) using
visible dye concentrations in transparent well casings
provided evidence that the intermittent inclusion of stagnant
water detected in the pump discharge may have been related to
turbulence and subsequent mixing around the pump inlet. The
reason for the relatively small concentrations of stagnant
water detected at the end of the run is probably related to
the fairly small volume of stagnant water above the inlet,
(0.1L). This volume would most likely have been diluted by
mixing during the pumping of 9 bore volumes (130L) . At 9 bore
volumes, the pump inlet was lowered into the screen area.
With the pump inlet located in this position, one bore volume
of water (14.4 liters) was collected without the detection of
stagnant water in the pump discharge. Figure 2 presents a
five minute expanded segment of Figure 1 between 4.0 to 4.3
bore volumes. It shows the intermittent nature of the
intrusion of stagnant water into the pump inlet. The bore
volume axis has been converted to time. While the contri-
bution of stagnant water is relatively small at this point, it
is nonetheless measurable, and for intervals up to 30 seconds.
1-437
-------
456
BORE VOLUMES
10
Fig. 1- Inlet near static water level
BORE VOLUME 4.0 TO 4.3
57.5 58 58.5 59 59.5 60
TIME, (min)
-A-
60.5 61 61.5 62
Fig. 2- Expanded view, bore volume 4.0 to 4.3
11-438
-------
I
a
M
^
c
o
1
>-
>
JH
o
a
-7
o
1 U
9 -
J
8-
7 -
6-
4-
3-
2-
1 -
n -
DEPTH
pH
\^/^
\ CONDUCTIVITY
TEMPERATURE
. L. -A
r \ . . ^ ~ ^ ~
'
-/4
-23
-22
-21
-20
-19
-18
-17
-16
-15
-1 4
1 *T
-13
-1 O
j:
x-
uj
Q
^^s
O,
a."
u
i
0
456
BORE VOLUMES
10
Pig. 3- Indicator parameters
One implication of this is that approximately 500ml of water
could be collected as a groundwater sample over a 30 second
time interval. Depending on the purpose for which the sample
was being collected, inclusion of almost 1 percent stagnant
water could be problematic.
As shown in Figure 3, temperature, pH, and conductivity during
this run displayed an initial response during the removal of
the first bore volume. However, they exhibited little or no
change during the pumping of subsequent bore volumes. A
visual inspection of these parameters between bore volumes 1
and 2 indicate little if any change, whereas the tracer
concentration shown in Figure 1 indicates a stagnant water
concentration decreasing between 11 and
2 percent. These parameters were too insensitive for the
detection of trace concentrations of stagnant water in this
run. An examination of these parameters for all of the other
purging runs revealed similar results.
Peristaltic Pump. Inlet at Mid-casing
11-439
-------
Figure 4 displays the results of a purging test with the pump
inlet located in the mid-casing position with equal amounts of
PUMPING TIME 90 MINUTES
INLET LOWERED TO SCREEN
AT 10.6 B.V.
567
BORE: VOLUMES
10 11 12
Fig. 4- Inlet at mid-casing
water above and below the inlet. Examination of Figure 4
reveals results similar to the purging run displayed in Figure
1 in that the majority of the stagnant water below the pump
inlet was removed with the initial bore volume of water.
While the stagnant water concentration tapered off more slowly
between bore volumes 2 and 4 with the inlet position located
in the mid-bore position, the subsequent bore volumes pumped
after the fourth bore volume appear to have elevated stagnant
water concentrations when compared to the run displayed in
Figure 1.
Peristaltic Pump, Inlet at Screen Top
Figure 5 displays the results of a purging run with the pump
inlet situated immediately above the well screen.
Because there was no volume of storage water below the pump
inlet, the lower x-axis in Figure 5 refers to the number of
liters pumped. The absence of storage water below the pump
inlet is evidenced by the lack of an initial interval during
1-440
-------
20
18-
16 -
z 12-
<
i 10-
LJ
<
O
ex.
LJ
D-
6 -
2-
0
0
PUMPING TIME 80 MINUTES
INLET RAISED 10 FEET
IMLET LOWERED TO SCREEN
i
20
30 40
LITERS PUMPED
50
60
70
Fig. 5- Inlet at top of screen
which the pump discharge was 100 percent stagnant water. This
indicates that the pump inlet was positioned at the stagnant
water/fresh water interface. The results suggest that the
stagnant water detected in the pump discharge during the 80
minute pumping period was caused by the stagnant water
overlying the inlet mixing with the fresh water from the
screen. At 60 minutes, the pump inlet was lowered 5 cm (2 in)
to the top of the screen with little effect on the stagnant
water concentration. At 69 minutes, the inlet was raised 3 m
(10 ft) which increased the stagnant water concentration to
100 percent.
Source of Stagnant Water
The results of the preceding three purging trials suggest that
the stagnant water detected in the pump discharge was from the
dyed water overlying the pump inlet. The reasons are twofold.
First, the concentration of stagnant water detected in the
pump discharge was apparently affected by the volume of dye
overlying the inlet. A mathematical average of the stagnant
water concentration detected between the fifth and the sixth
11-441
-------
bore volumes was determined to be 0.026 percent, 0.199
percent, and 0.103 percent for Figures 1, 4, and 5 respective-
ly (Fig.5 concentration determined by examining the data be-
tween 38 and 45 minutes, an interval equal to that selected
for the concentration determined in Figure 4) . One expla-
nation for the average stagnant water concentration in Figure
1 being approximately one order of magnitude smaller than that
of Figures 4 and 5 is the smaller volume of stagnant water
(0.1L -vs- 7.2L and 14.4L) available for mixing above the pump
inlet. The much smaller volume would have been diluted to a
greater extent over a given time period, accounting for the
smaller percentage of stagnant water in the pump discharge.
Second, the results of the run displayed in Figure 5 verify
that the pump inlet was positioned at the stagnant water/fresh
water interface. The detection of stagnant water for the
duration of the 80 minute run indicates that the tracer above
the pump inlet was being captured.
Multiple Inlet Positions. Dual Tracers
In an attempt to repeat a purging test conducted by Robin and
Gillham (6), deionized water was used as a tracer in addition
to Rhodamine WT. During this purging test, varying inlet
positions were used starting from the screen top. The purging
rate was 1.1 L/min. The water above the screen was replaced
with deionized water containing Rhodamine WT using the
inflatable packer in the manner described previously. The
well was pumped with the inlet immediately above the screen
while the fluorescence and conductivity were being monitored.
The well was pumped in this manner until the fluorescence
approached zero. The inlet was then raised 1.2 m (4 ft) and
pumping was continued at the new level until the fluorescence
approached zero. This process was repeated several times
until the air/water interface was reached.
The results of this test are presented in Figure 6. In
general, the shapes of the curves obtained with the fluo-
rescent dye are very similar to the initial portions of curves
shown in Figures 1, 4, and 5 in that removal of the majority
of'the stagnant water below the inlet occurs fairly quickly.
The curves from the deionized water conductivity measurements
however, contradict those for the fluorescent dye. The
deionized water tracer suggests background water quality was
reached within a few minutes of each relocation of the pump
inlet, whereas the curves from the fluorescent dye tracer
still show stagnant water detectable in the percent range
several minutes later. The pattern of response for each of
the tracers was repeated each time the pump inlet was raised.
Deionized water may be too insensitive a tracer when used to
differentiate water sources in a monitoring well. Conclusions
11-442
-------
CONDUCTIVITY
(DEIONIZED WATER)
10
20
30 40
TIME, (min)
50
60
1.2
-1
STAGNANT WATER
(FLUORESCENT DYE)
70
TJ
O
i_
D>
.*
O
o
.a
_
W
o
0)
E
O
O
Fig. 6- Comparison of Rhodamine WT and deionized water as
tracers
as to required purging volumes based on deionized water tracer
tests may, therefore, be inappropriately low for many purpos-
es.
Peristaltic Pump. Pump and Lower
Figure 7 displays the results of a purging test during which
the pump inlet was initially positioned five feet above the
screen for approximately five bore volumes (18.4L) and then
repositioned into the screen for sample collection. The
advantage of purging and sampling in this fashion is that the
entire column of stagnant water need not be purged in order to
collect a representative groundwater sample. This test was
conducted in triplicate with similar curves observed for all
three runs. During the removal of the initial five bore vol-
umes of water a clean zone generally free of stagnant water
was developed below the pump inlet. After apparently lowering
the pump inlet, a minimum 40 liter pumping period followed,
during which stagnant water was not detected in the discharge.
II-443
-------
PUMPING TIME 105 MINUTES
INLET LOWERED TO SCREEN
(18,4 LITERS)
INLET RAISED
10 20 30 40 50 60 70 80 90 100 110
LITERS PUMPED
Fig. 7- Pump and lower
Stagnant water was eventually detected in the pump discharge.
This occurred while the pump inlet was still in the screen
area meaning that the stagnant water had migrated over the
five foot separation between the initial inlet position and
the screen. The mechanism for this migration has not been
investigated, but is likely due to a combination of diffusion
and advection.
PACKERS
Figures 8 (Keck pump) and 9 (laboratory manufactured packer)
present the results of purging runs using packers. Tracer was
added to the well for each trial in the manner described
earlier. At the start of each test the packer used for tracer
addition was removed. The packer used for sampling was slowly
lowered so that the pump inlet for each trial was positioned
immediately above the screen. The packer was inflated and
pumping was initiated.
Neither packer functioned as anticipated. This is evidenced by
the unexpected contribution of stagnant water to the pump dis-
II-444
-------
Of.
Ld
^_
1
z
z
o
1
z
Ld
o
cr
1 1
LlJ
Q_
ZU
18-
16-
14-
12-
10-
8-
6-
4-
2-
o-
1
PUMPING IIME /4 WINUILS
,[
^iJlliilwiiilLi,, i, i, Li J. L i
i i
0 50 100 150 200 250 300 350 4C
LITERS PUMPED
Fig. 8- Keck Pump with packer
charge. While stagnant water was detected in both runs, there
are marked differences between Figures 8 and 9. Both curves
exhibit the same general shape. However, the curve for the
Keck pump (Figure 8) is dramatically shifted to the right. A
portion of the curve from Figure 8 is superimposed onto Figure
9 for comparison. There are three possible causes for the
higher than expected contribution of stagnant water in these
tests. First, leakage past the packer induced by a decrease
in head below the packer during pumping may have caused stag-
nant casing water to migrate past the packer into the pump
inlet. Such a leak could have been caused by an irregularity
or crack in the casing wall. Second, in the case of the Keck
pump and packer, a significant volume of water is displaced
(2.66 L) when the pump is lowered into position. This
volume is equivalent to a water level rise of 1.31 m (4.3 ft)
in a 5 cm (2 in) well casing. It is assumed therefore, that
stagnant water was displaced downward through the screen into
the aquifer. This may have caused the dramatic shift to the
right of the curve for the Keck pump. Finally, stagnant water
held between the packer and the pump inlet may have been drawn
in by the same phenomenon observed in other tests were
1-445
-------
20
18-
16
<
<
8-
LJ c J
O 6 I
o:
2-
0
0
PERCENT STAGNANT WATER
FROM KECK PUMP/PACKER
INLET RAISED TO
NEAR SURFACE
AT 28 MM.
_ ....1 ....ป- A ซ
j
10 15
LITERS PUMPED
20
25
Fig. 9- Peristaltic pump with packer
stagnant water was above the pump inlet. The inlet on the
laboratory manufactured flow-through packer was immediately
below the packer. However the Keck pump inlet was situated
approximately 30 cm (12 in) below the bottom of the packer.
This may explain why the test with the Keck pump showed much
more frequent and higher spikes of stagnant water than did the
flow-through packer.
Purging/Sampling With Drawdown
Figure 10 displays stagnant water concentrations for a well
sampled by the drawdown and recover method described
earlier. The pump inlet was positioned at the screen top for
the duration of the test. The initial pumping rate for the
first 25 minutes was approximately 750 ml/min. During the
initial portion of this time period when the rate of drawdown
was the greatest (as indicated by the slope of the curve
marked "depth") a relatively larger portion of the pump
discharge would have been from stagnant water. This is
confirmed by the larger lithium concentrations. As the rate
of drawdown decreases (indicated by decreasing slope of the
11-446
-------
10
20 30 40
TIME, (min)
50
60
Pig. 10- Purging/sampling with drawdown
"depth" curve) the lithium concentration also decreases,
indicating a smaller fraction of stagnant water in the pump
discharge. At 25 minutes the pumping rate was decreased to
300 ml/min to allow the well to recharge. Evidence of the
stagnant water/fresh water interface, moving upward away from
the pump inlet is shown by the rapid decrease in lithium con-
centration. Sample collection during the period of recovery
would minimize the inclusion of stagnant water.
The principle illustrated here is related to that used in the
ISWS procedure (1). However, since the mathematical methods
used there were inadequate to predict the drawdown for this
well, it was necessary to measure drawdown directly in order
to determine when a sample could be obtained. Direct measure-
ment is not difficult in such a low yielding well, and it
could be part of the sampling protocol for such a well.
SUMMARY AND CONCLUSIONS
The results of the research presented here show that the
procedures used to sample a well can have an effect on the
1-447
-------
amount of stagnant water in a groundwater sample. All runs
conducted in the absence of drawdown with the pump inlet in a
fixed position at or above the screen showed a highly variable
and unpredictable inclusion of stagnant water. This inclusion
of stagnant water may have been caused by turbulence around
the pump inlet. Packers were not generally effective in
preventing the inclusion of stagnant water into the pump
inlet. Deionized water as a tracer in this study was general-
ly ineffective and possibly misleading when compared to
Rhodamine WT. The results of tests conducted with deionized
water as a tracer which suggest pumping 2 to 3 bore volumes
from near the air/water interface may be inappropriately low.
Real time monitoring of indicator parameters such as pH,
temperature and conductivity was not successful in indicating
when purging was complete. Research by Gibs and Imbrigiotta
(4) resulted in a similar conclusion.
The inclusion of stagnant water into a sample was minimized by
purging from some distance above the screen followed by
relocation of the pump inlet into the screen for sample
collection. In wells where drawdown occurred during purging,
stagnant water inclusion was minimized by reduced pumping
rates to allow for sample collection during well recharge.
The Illinois State Water Survey procedure for calculating the
effect of drawdown was found not to be usable for the wells in
this research.
REFERENCES
(1) Illinois State Water Survey and Illinois State Geologi-
cal Survey, Cooperative Groundwater Report 7, "Proce-
dures for the Collection of Representative Water Qual-
ity Data from Monitoring Wells", Gibb, J.P., Schuller,
R.M., and Griffin, R.A., State of Illinois Department
of Energy and Natural Resources, 1981.
(2) Wood, W.W., "Guidelines for Collection and Field Analy-
sis of Groundwater Samples for Selected Unstable Con-
stituents", U.S. Geological Survey Techniques of Water
Resources Investigations, Book 1, p24, 1976.
(3) Barcelona, M.J., Gibb, J.P., Helfrich, J.A., and Gar-
ske, E.E., Practical Guide for Ground-Water Sampling.
Illinois State Water Survey, November 1985.
(4) Gibs, J., and Imbrigiotta, T.E, "Well-Purging Criteria
for Sampling Purgeable Organic Compounds", Ground
Water, Volume 28, Number 1, January-February 1990.
11-448
-------
(5) Slawson, G.C., Jr, Kelly, K.E., and Everett, L.G.,
"Evaluation of Ground-Water Pumping and Bailing Meth-
ods-Application in the Oil Shale Industry", Ground
Water Monitoring Review. Summer 1982.
(6) Unwin, J.P., and Maltby, C.V., "Investigations and
Techniques for Purging Ground-Water Monitoring Wells
and Sampling Ground Water for Volatile Organic Com-
pounds", Ground-Water Contamination: Field Methodsf
Collins, A.G., and Johnson, A.I., Eds., ASTM STP 963,
1988.
(7) Robin, M.J.L., and Gillham, R.W., "Field Evaluation of
Well Purging Procedures", Ground Water Monitoring
Review. Fall 1987.
(7) "Guide to Groundwater Sampling", NCASI Technical Bul-
letin No. 362, January 1982.
(8) Replogle, J.A., Myers, L.E., and Brust, K.J., "Flow
Measurements with Fluorescent Tracers", Journal of
Hydraulics Division. September 1966.
(9) Unwin, J.P., and Huis, D., "A Laboratory Investigation
of the Purging Behavior of Small-Diameter Monitoring
Wells", Proceedings of the Third National Symposium on
Aquifer Restoration and Ground-Water Monitoring. Niel-
sen, D.M., Eds., National Water Well Association, May,
1983.
1-449
-------
103 ANALYSIS OF POLYCHLORINATED BIPHENYLS IN WATER AND STACK
EMISSIONS BY HIGH RESOLUTION GAS CHROMATOGRAPHY/ HIGH
RESOLUTION MASS SPECTROMETRY.
Edwin A Marti, Hani S Karam, Jakal Amin, Triangle Laborato-
ries, Inc., 801-10 Capitola Dr., Research Triangle Park,
North Carolina 27713; Timothy J Yagley, Alan F Weston,
Occidental Chemical Corporation, Niagara Falls, New York.
ABSTRACT
Polychlorinated biphenyls (PCBs) in environmental samples
are generally analyzed by high resolution gas chromatogra-
phy/low resolution mass spectrometry (HRGC/LRMS) or high
resolution gas chromatography with electron capture detec-
tion (HRGC/ECD). Detection limits reported using these
techniques for water samples are on the order of 50-500 ppt
for the mono-deca PCBs (HRGC/LRMS) or 50-100 ppt for Aro-
clor characterization by GC/ECD. HRGC/LRMS analysis of air
samples (collected on XAD-2) typically show detection
levels of 50 to 500 ng for the mono-deca PCBs.
High resolution GC/high resolution MS (HRGC/HRMS) is used
routinely for the analysis of polychlorinated dioxins and
furans (PCDDs/PCDFs) in water and air samples, with detec-
tion limits as low as 10 parts per quadrillion (ppq) for
water and 50 picograms (pg) for air.
This HRGC/HRMS technique has recently been utilized for the
analysis of PCBs in water and air samples and the sample
results indicate that the detection limits of these species
are at least two orders of magnitude lower than achieved
using the low resolution mass spectrometric technique.
Using this technique, PCBs are reported as totals for each
congener group (mono-deca) as well as congener specific
analysis for 11 congeners, seven of which are quantified by
isotope dilution mass spectrometry.
INTRODUCTION
The past few years have witnessed an increasing need for
new methodologies that are capable of measuring very small
quantities of toxic substances in various matrices, i.e.,
low parts per trillion (ppt) for soils, parts per quadril-
lion (ppq) for water and picograms (pg) for air samples
collected on solid absorbents.
The proposed method determines polychlorinated biphenyls in
II -450
-------
sample extracts representing one liter of water or stack
emission (air) samples collected on XAD-2. The method of
analysis utilizes high resolution gas chromatography/high
resolution mass spectrometry (HRGC/HRMS) operated at a re
solving power of 8,000 to 10,000 in the selected ion moni-
toring (SIM) mode. This method was based on isotope dilu-
tion mass spectrometry during which nine 13Ci2 - labeled
internal standards were used to characterize and quantify
all 209 PCB congeners. By using published retention times
of the 209 congeners 1f five retention windows bracketing
the ten congener groups could be monitored to determine
total PCBs by isomer groups (mono through deca) as well as
specifically quantify eleven PCB congeners.
EXPERIMENTAL METHOD
For stack emission sampling of stationary sources, the
XAD-2 resin was spiked with 10 ng of surrogate standards
prior to sampling (Table 1). Following the sampling ses-
sion, the samples (XAD-2, glass-fiber filter, front half
and back half solvent rinses, impinger water and impinger
rinses) were returned to the laboratory. The front half
and back half rinses were concentrated, then placed inside
a Soxhlet extractor along with the rest of the solid frac-
tions of the sampling train. The sample was spiked with 10
ng of PCB internal standards (Table 1), then Soxhlet ex-
tracted with 750 mL of methylene chloride. The impinger
water was spiked with 10 ng of alternate surrogate stand-
ards (Table 1). The water was then liquid-liquid extracted
in a separatory funnel using 3 X 60 mL methylene chloride.
Both the impinger water extract and the Soxhlet extract
were concentrated then combined. ' The extract was split
50:50, with one-half being archived and the other half
subjected to an acid/base wash cleanup. The sample extract
was then concentrated to a final volume of 50 microliters.
For water samples, one liter of sample was spiked with the
nine internal standards (in acetone) at 10 ng. The sample
was allowed to equilibrate for one hour. The sample was
then extracted with 3 X 60 mL portions of methylene chlo-
ride in a separatory funnel. The extract was concentrated
using a K-D apparatus and put through an acid/base wash
cleanup. The extract was then concentrated to a final
volume of 100 microliters. Before analysis of the PCB
extracts, 5 ng of recovery standards (Table 1) are added to
the extracts.
CALIBRATION
The mass spectrometer response was calibrated by using the
11-451
-------
set of five initial calibration solutions shown in Table
1. Each solution was analyzed once and the analyte rela-
tive response factors (RRF) were calculated.
An acceptable calibration must meet the following criteria:
1) The percent relative standard deviations (RSD) for the
mean response factors from each of the unlabeled ana-
lytes must be less than 25 or 30 percent depending on
the analyte (Table 2).
2) The signal-to-noise ratio (S/N) for the GC signals
present in every selected ion current profile must be >
10:1.
3) The ion abundance ratios must be within the specified
control limits (Table 3).
A continuing calibration was demonstrated every 12 hours by
injecting one uL of solution number 2 from Table 1. The
RRFs are calculated and compared to the mean RRFs obtained
during the initial calibration procedure. An acceptable
continuing calibration run must meet the following crite-
ria:
1) The measured RRFs (for the unlabeled PCBs) obtained
during the continuing calibration run must be within 25
or 30 percent depending on the analyte (Table 2) of the
mean values established during the initial calibration.
2) The ion-abundance ratios must be within the allowed
control limits listed in Table 3.
3) The signal-to-noise ratio (S/N) for the GC signal
present in every selected ion current profile must be >^
10:1.
At the beginning of every 12-hour shift during which sam-
ples are analyzed the fused-silica capillary GC column
performance was verified by injecting a 1-uL aliquot of the
PCB window defining mixture (Table 4). This was necessary
to identify the various retention time windows for each
group of analytes, which are grouped in five mass descrip-
tors. Figure 1 shows the tetra-PCB first and last eluter
chromatogram with the corresponding tetra-PCB internal
standard.
RESULTS AND DISCUSSION
In order to evaluate the analytical method's ability to
11-452
-------
detect and quantify small quantities of analyte present in
water, three sets of five samples at 0.5, 5.0 and 25 ppt
were analyzed. The results for the 0.5 ppt matrix spike
are given in Table 5 and on Figures 2 and 3. The MDL and
LOD values were calculated for the lowest point (0.5 ng/L)
samples. The mean %Accuracy for the most congeners were
approximately equal to 100%. The mean %Accuracy ranged from
75.6% for 2255-T-PCB to 100% for 2234455-Hp-PCB. The
%Recovery of the internal standards range from 44.62% for
22455-Pe-PCB to 280.0% for 334455-Hx-PCB. The high recov-
ery of the carbon labeled 334455-Hx-PCB was due to an
interference problem. The %RPD range from 0.0% for 2234455-
Hp-PCB to -13.6% for 223344556-Nona-PCB. The results of
5.0 ng/L and 25.0 ng/L were also similar to 0.5 ng/L sam-
ples. For both 5.0 ng/L and 25.0 ng/L the internal stand-
ards recoveries of 22455-Pe-PCB were the lowest and highest
for 334455-Hx-PCB. For 5.0 ng/L mean %Accuracy range from
66.60% for 2255-T-PCB to 119.20% for 244-Tr-PCB. For 25.0
ng/L the mean %Accuracy range from 2255-T-PCB for 111.12%
for 3344-T-PCB. The %RPD ranged from -0.80% for 44-Di-PCB
to -33.40% for 2255-T-PCB for 5.0 ng/L samples. The %RPD
range from 2.56% for 244-Tr-PCB to -28.20% for 2255-T-PCB
of 25.0 ng/L samples.
In all three points, the ฑ3C12-22455-Pe-PCB gave the lowest
recovery. One possible explanation for the low recovery
might be the compound was not in the same mass descriptor
as the recovery standard. Another explanation is that the
concentration of carbon-labeled standards are not measured
using isotope dilution method. When the corresponding
analyte, 22455-Pe-PCB was measured using the isotope dilu-
tion method, the mean %Accuracies were 103.6%,110.20%, and
93.84% for 0.5 ng/L, 5.0 ng/L, and 25.0 ng/L samples,
respectively. The concentrations of 224455-Hx-PCB were
computed using 13d2-(245)3-Hp-PCB with mean %Accuracies of
86.80%, 87.02%, and 78.79% for 0.5 ng/L, 5.0 ng/L, and 25.0
ng/L, respectively. This was done because when
13Ci2-334455-Hx-PCB was used to compute the analyte concen-
tration, the results were erratic due to the high percent
recovery of the internal standard caused by an interfer-
ence. The 13Ci2-224455-Hx-PCB was not used to compute the
corresponding analyte concentration because carbon-labeled
standard was used as the recovery standard. In the future,
the analyte, 224455-Hx-PCB, will be computed using the
corresponding internal standard (13C12-224455-Hx-PCB) and
the T3Ci2-334455-Hx-PCB will be used as the recovery stand-
ard.
The MDL values for 0.5 ng/L samples were calculated using
11-453
-------
the formula:
MDL = S * ttm_a., i_ซ _ 0.99)
where S = Standard Deviation
t = Student t.
For the present study, the MDL value for the 0.5 ppt spike
was 0.072 ppt. A more rigorous determination of the MDL
can be determined spiking seven replicate samples at 0.05
ppt. In lieu of this, the Limit of Detection (LOD) can
still be calculated with this data set. Using So (the
value of the standard deviation as concentration approaches
zero), the LOD (Limit of Detection) was computed using the
formula:
LOD = 3 * S0.
The LOD for the mono-PCB isomer was 0.036 ng/L.
The matrix spike evaluation for the stack emission samples
is currently in progress. The preliminary results show
recoveries between 80 and 140% for the eleven PCB target
analytes with %RPDs ranging from 5 to 40% for three matrix
spikes.
CONCLUSIONS
The extraction, cleanup and analysis procedures described
in this method for the trace analysis of mono through deca
polychlorinated biphenyls in water are adequate for the
isolation and measurement of individual PCBs to detection
limits in the low ppq range.
The Limit of Detection (LOD) was calculated to be 36 ppq
(parts per quadrillion) for the mono-PCB isomer.
The overall accuracy of the method, as determined by a
series of five matrix spikes at three different concentra-
tion levels (33 total measurements), was 94.5% [ranging
from 66.6% for the 2255-tetrachlorobiphenyl to 120% for the
3344-tetrachlorobiphenyl both at the 5 ppt spike level].
The precision of the method, as calculated from the mean of
the 33 analyses, was 6.4% relative standard deviation
[ranging from 2.49% for the 3344-tetrachlorobiphenyl to
19.7% for 2255-tetrachlorobiphenyl both at the 5 ppt spike
level].
11-454
-------
Table 1. Composition of the Initial Calibration Solutions
^ ^" ^"^ ป ป ป ป ป ป ป ป ป ^ Bป BB BB ป * BB BB BB BB BB BB BBBB BM BB BB BB BB BB ป BB Bป BB BB ^ ^ ^ ^ ^B BB BB BB BB BB BB BB BB BB BB
Compound Concentrations (pg/uL)
Solution Number 1 234
Un labeled Analytes
2 -Chlor obipheny 1
44 ' -Dichlorobiphenyl
244 ' -Trichlorobiphenyl
22 '55' -Tetrachlorobiphenyl
33 ' 44 ' -Tetrachlorobiphenyl
22 ' 455 ' -Pentachlorobiphenyl
22 ' 44 ' 55 ' -Hexachlorobiphenyl
22 ' 344 ' 55 ' -Heptachlorobiphenyl
22 ' 33 ' 44 ' 55 ' -Octachlorobiphenyl
22 ' 33 ' 44 ' 55 ' 6-Nonachlorobiphenyl
Decachlorobiphenyl
0.5
0.5
0.5
1.0
1.0
1.0
1.0
1.5
1.5
2.5
2.5
5
5
5
10
10
10
10
15
15
25
25
10
10
10
20
20
20
20
30
30
50
50
50
50
50
100
100
100
100
150
150
250
250
100
100
100
200
200
200
200
300
300
500
500
Internal Standards (13Ci2)
4-Chlorobiphenyl(3y
44'-Dichlorobiphenyl
244'-Trichlorobiphenyl
33'4,4 '-Tetrachlorobiphenyl
22'455'-Pentachlorobiphenyl
22'44'55'-Hexachlorobiphenyl<4>
22'344'55'-Heptachlorobiphenyl
22'33'44'55'-Octachlorobiphenyl(2 >
Decachlorobiphenyl
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
Surrogate Standard (13Ci2)
33'55'-Tetrachlorobiphenyl
33'44'5-Pentachlorobiphenyl
22'344'5'-Hexachlorobiphenyl
22'33'55'66'-Octachlorobiphenyl
Alternate Standard (13C12)
2 2'3 3'4 4'-Hexachlorobiphenyl
100 100 100
100 100 100
100 100 100
100 100 100
100 100
100 100
100 100
100 100
100 100 100 100 100
Recovery Standards (13Ci2)
22'55'-Tetrachlorobiphenyl
33'44'55'-Hexachlorobiphenyl{4
200 200 200 200 200
200 200 200 200 200
11-455
-------
The method evaluation of stack emission samples is current-
ly in progress with results expected at the time of publi-
cation of this paper.
REFERENCES
(1) Mullin, M. D.f Pochini, C. M., McCrindle, S., Romkes,
M., Safe, S. H. and L. M. Safe (1984). High Resolution PCB
Analysis: Synthesis and Chromatographic Properties of All
209 PCB Congeners. Environ. Sci. Technol., Vol. 18, No. 6,
p. 468-476.
11-456
-------
Table 1 (continued):
Notes:
1) Based on 100 uL final extract volume,this corresponds to a
calibration Range from 50 pg to 10 ng for mono-PCB).
2) The labeled octa-PCB (Internal standard) is used to compute
response factors of unlabeled nona-PCBs.
3) The Mono-chloro-Biphenyl internal standard is a 13C6 and not
al3p
*-12
4) The 22'44'55'-hexa-PCB (internal standard) and 33'44'55'-hexa-PCB
(recovery standard) reflect the current method. The original
method validation had them switched.
1-457
-------
Table 2. Initial and Continuing Calibrations Response Factors
Minimum Requirements
Compound
Relative Response Factors
I-Cal Con-Cal
%RSD %Delta
2-chlorobiphenyl
44 ' -dichlorobiphenyl
244 ' -trichlorobiphenyl
22 ' 55 ' -tetrachlorobiphenyl
33 ' 44 ' -tetrachlorobiphenyl
22 '455 ' -pentachlorobiphenyl
22 ' 44 ' 55 ' -hexachlorobiphenyl
22 ' 344 ' 55 ' -heptachlorobiphenyl
22 ' 33 ' 44 ' 55 ' -octachlorobiphenyl
22'33'44'55' 6-nonachlorobiphenyl
decachlorobiphenyl
30
25
25
30
25
30
30
25
25
30
25
30
25
25
30
25
30
30
25
25
30
25
13C6-4-Chlorobiphenyl
13Ci2-44'-Dichlorobiphenyl
13Ci2-244'-Trichlorobiphenyl
i3Cj.2-33'44'-Tetrachlorobiphenyl
X3Ci2-22'455'-Pentachlorobiphenyl
13C12-22'44'55'-Hexachlorobiphenyl
13CX2-22'344'55'-Heptachlorobiphenyl
13Cj.2-22 ' 33' 44' 55' -Octachlorobiphenyl
13Ci2-Decachlorobiphenyl
30
30
30
25
30
30
30
25
30
30
30
30
25
30
30
30
25
30
13CX2-33'55'-Tetrachlorobiphenyl
X3Ci2-33'44'5-Pentachlorobiphenyl
13C12-22'344'5'-Hexachlorobiphenyl
i3Ci2-22'33'55'66'-Octachlorobiphenyl
13C12-22'33'44'-Hexachlorobiphenyl
25
25
25
25
25
25
25
25
25
25
Notes:
1) Isomers that have 25% criteria are
X3Ci2-labeled standards.
those with corresponding
2) The labeled tetra-PCB (IS) will be in a different mass
descriptor than the unlabeled tetra-PCB analyte.
11-458
-------
Table 3. Ion-Abundance Ratio Acceptable Ranges
Number of
Halogen
Atoms
Ion Type
Theoretical
Ratio
Control Limits
Lower
Upper
1 Cl
2 Cl
3 Cl
4 Cl
5 Cl
6 Cl
7 Cl
8 Cl
9 Cl
10 Cl
M/M+2
M/M+2
M/M+2
M/M+2
M/M+2
M+2/M+4
M+2/M+4
M+2/M+4
M+2/M+4
M+4/M+6
3,
1,
1,
0,
0,
1,
1
08
54
03
77
61
24
04
0.89
0.78
1.18
2.62
1.31
0.87
0.65
0,
1.
52
,05
0.88
0.76
0.66
1.00
3.54
1.77
1.18
0.89
0.70
1.43
1.20
1.02
0.90
1.36
11-459
-------
Table 4. PCB Window Defining Mix
PCS isomer Group Number
2-chloro (F) 1
4-Chloro (L) 1
2,6-dichloro (F) 2
4,4'-dichloro (L) 2
2,4,6-trichloro (F) 2
2,3,5-trichloro 2
2,2',6,6'-tetrachloro (F) 2
3,4,4'-trichloro (L) 3
2,3,3',4-tetrachloro 3
3,3'4,4'-tetrachloro (L) 3
2,2',4,6,6'-pentachloro (F) 3
2,2',4,4',6/6'-hexachloro (F) 3
3,3',4,4',5-pentachloro (L) 4
2,2',3,4,4',6-hexachloro 4
2,2',3,4,5,6'-hexachloro 4
3,3',4,4',5,5'-hexachloro (L) 4
2,2',3,4',5,6,6'-heptachloro (F) 4
2,2',3,3',4,4',5-heptachloro 4
2,2',3,3',5,5',6,6'-octachloro (F) 4
2,3,3',4,4',5/5'-heptachloro (L) 5
2,2',3,3',4,4',5,5'-octachloro JL) 5
2,2' ,3,3',4,5,5',6,6'-nonachloro (F) 5
2,2',3,3',4/4',5,5',6-nonachloro (L) 5
decachloro 5
Note: The table contains the order of elution for specific
isomers.
11-460
-------
Table 5 Matrix Spike Replicate Analytical Results
(0.50 ppt Spike)
Analytes
2 -Mo
44-Di
244-Tr
2255-T
3344-T
22455-Pe
224455-Hx*
2234455-Hp
22334455-Oc
223344556-No
Dec a
Cj.2
Internal
Standards
4 -Mo
44-Di
244-Tr
3344-T
22455-Pe
224455-Hx
2234455-Hp
22334455-Oc
Dec a
MSI
0.46
0.54
0.55
0.72
1.20
1.10
0.86
1.50
1.70
2.10
2.60
MS2
<
0.44
0.53
0.53
0.77
1.00
1.00
0.88
1.40
1.50
2.20
2.50
MS 3
[ Concentration
0.45
0.57
0.53
0.73
1.00
0.98
0.88
1.50
1.60
2.10
2.60
MS4
in ppt)
0.48
0.59
0.53
0.77
1.20
1.00
0.89
1.60
1.70
2.20
2.80
MS 5
0.49
0.57
0.59
0.79
1.20
1.10
0.83
1.50
1.70
2.20
2.70
Mean
0.46
0.56
0.55
0.76
1.12
1.04
0.87
1.50
1.64
2.16
2.64
%Recoveries
68.5
140.0
90.1
116.0
56.6
301.0
98.4
122.0
90.3
78.0
149.0
105.0
125.0
52.8
211.0
103.0
114.0
92.3
78.1
174.0
112.0
123.0
42.8
228.0
100.0
112.0
88.0
73.2
161.0
105.0
115.0
34.3
246.0
101.0
116.0
90.0
85.3
157.0
107.0
124.0
36.6
414.0
107.0
131.0
89.1
76.6
156.2
103.8
120.6
44.6
280.0
101.9
119.0
89.9
*13Ci2-(245)3-Hp was used as the internal standard.
1-461
-------
riiป
> ilf.BZ*j r:j.2ff.fS23 T-.f
L*ปOK*TOXJfS. INC. : :ซ HXS rCB:
100
P5j
to.
sj
0:
75:
70.
<5.
CO.
55.
50.
45.
40.
35.
30.
25:
20.
ซ
10
5
0
22:00 23:00 24:00 25:00 26:00 27:00 21:00 29:00 30:00 31:
rilป:*H2125 Aey:2f-tOO(-fl 07:12:33 tUft 301.9626 F:3
Sปmplป Tuct:SANPXC riiป TซJrt:ntZANCLE IABORATOXZES, INC.: :KI RES PCS; :
24 10
r-r-fcf
<
23:03
23:41
L 3.3'.4,4'-T-PCS " 0.01 ppt
t
P5j
*0j
ป5J
OJ
75J
70J
C5.
ซ0
55.
50j
45.
40:
3S
30
25
20.
153
10.
Cซrbon-I*b.iaปd-2, 2' 5, S' -T-PCB
30:10
Cซrlx>ij-I*b.Jl.<(-3, 3', 4, 4'-T-PCB
22:00
23:00
24:00
25:00
26:00
27:00
28:00
29:00
30:00
00 XTNC
S.9E7
.S.SE7
.I.OC7
.7.6E7
.7.1E7
.6.7E7
.6.3C7
L5.8E7
5.4E7
4.9E7
4.SE7
.4.0E7
.3. 6E7
.3.1C7
.2.7E7
.2.2E7
.1.BE7
.1.3C7
.t.PEC
.4.see
O.OZO
31: 00 TIH.
Figure 1
11-462
-------
PCB HIGH RESOLUTION GC/MS ANALYSIS
WATER MATRIX SPIKE REPLICATE RESULTS
Q.
CL
o
LLJ
O
z
o
o
ACTUAL SPIKE
MEAN
%RPD
: %RSD (N = 5)
0
Figure 2
-------
PCB HIGH RESOLUTION GC/MS ANALYSIS I
WATER MATRIX SPIKE REPLICATE RESULTS
HI ACTUAL SPIKE
f///A
MEAN
'/'#
%RPD
III!
%RSD
CL
Q.
o
\-
2
LLJ
O
Z
O
o
10 b
LLJ
O
DC
LU
DL
Figure 3
-------
CONTINUOUS ANALYSIS OF VOCs IN AIR USING A NEW,
PHENYL-METHYL SILICONE STATIONARY PHASE FOR
HIGH-RESOLUTION CAPILLARY GC
Rene P. M. Dooper and Nico Vonk. Chrompack Inc., 1130 Route
202 South, Raritan, New Jersey 08869. Henk J. Th. Bloemen,
RIVM, P. O. Box 1, Bilthoven, The Netherlands
INTRODUCTION
Monitoring (very) volatile organic compounds in outdoor air
becomes more and more important, as these compounds are
involved in smog formation and are known ozone precursors.
Also the US 1990 Clean air Act clearly indicates the need
for accurate data. In air, volatile organic components are
numerous and often very similar to each other. In most
cases they occur in the gaseous phase at ambient
temperature at the (sub) part-per-billion level. For these
reasons, chromatography is the most suitable method. These
considerations and the specifications for a monitoring
system have formed the basis for the design of a monitoring
system for volatile organic compounds in air, described in
this paperthe VOC Air Analyzer (Chrompack International,
Middelburg, The Netherlands) .
INSTRUMENTAL
The VOC Air Analyzer is a system for unattended continuous
automatic analysis of air containing (very) low levels of
organic components. The VOC Air Analyzer takes samples at
regular intervals over a selected time. The volatile
organic components are concentrated on an adsorbent tube.
When sampling has been completed the adsorbent tube is
heated to release the components and transport these to a
liquid nitrogen-cooled fused silica trap. Here the sample
components are refocussed in a narrow band. The trap is
then flash heated by which the sample is introduced into
the capillary column for analysis.
A schematical presentation of the VOC Air Analyzer is given
in figure 1. It has been designed to simultaneously sample
and analyze air using (cryo) adsorption and thermal
desorption techniques. Control of the heating and cooling
devices, valves, and sampling pump, as well as the
synchronization with the gas chromatograph and the
integrator-data processor, is performed by the VOCAA-
controller.
In the sample collection mode, air is drawn through the
valve and the adsorption trap by means of a sample pump
with a flow ranging from 10 to 70 mL/min. The adsorption
trap filled with appropriate trapping materials such as
Tenax, Carbosieve, or Carbotrap, or a composition of these
materials, is cooled using liquid nitrogen to a temperature
ranging from -20ฐC to ambient.
11-465
-------
Just before the end of a sampling period, the capillary
trap is cooled to a temperature with the range of -180ฐC to
subambient (cooling rate 100ฐC/min).
In the sample transfer mode, the trapped compounds are
transferred from the adsorption trap to the capillary trap
using a reversed carrier gas flow. The capillary trap
consists of a wide bore fused silica capillary coated with
liquid phase and/or filled with adsorbent. The transfer is
induced by heating the adsorption trap (heating rate
190ฐC/min) and switching of the valve. Only in this mode
are the sampler and the injector connected. While in this
mode, the temperature of the capillary trap is maintained
at low temperature set during the precool. To minimize
discrimination of the higher boiling compounds and sample
transfer time, the transfer flow is higher than the flow
defined in the restriction of the capillary column. This
is achieved by opening the desorption vent. After the
completion of the transfer, the valve is switched again.
To remove any remaining compounds the adsorption trap is
heated for a short period to a temperature higher than the
one in the sample transfer mode and again using a reversed
carrier gas flow. Before sampling is restarted, the
adsorption trap is cooled to the desired temperature.
Analysis time is optimized to allow separation of the
components of interest, cool down to, and equilibration at
the initial temperature setting before the sampling period
is over. In this way a continuous, unattended operation is
possible, based on a one hour cycle time.
It is necessary for most air samplers to remove the
moisture in the sample stream, as it might block the
adsorption tube or the cold trap with ice. Moisture can be
selectively removed on-line by passing the sample stream
through a two-stage dryer [Nafion tubes, DuPont Corp.
(Wilmington, Delaware), see the right part of Figure 1]
reducing the dew point to -55ฐC. This two-stage dryer is
self-regenerating. The drying force is the moisture
gradient generated by the underpressure (0.1 atm absolute)
in the first stage. The dry air stream is used in the
second stage to dry the sample stream. If dry air is
available, then a single Perma-Pure (Nafion tube) dryer
system can be used, which also removes the water from the
sample stream through the semi-permeable wall of the dryer
tube. The dry air stream is typically 4-5 times higher
than the air sample stream to get the desired drying
effect. The complete set of dryers, pump, valves, and
Nafion tubes is built in a new dryer/pump unit, which is
controlled by the control unit of the VOCAA. A
disadvantage of the Nafion dryer is that it partially
eliminate polar components which might be present in the
air samples. If these types of compounds have to be
analyzed, an option is built in to collect the samples
without the dryer in line in the sample stream, after which
the adsorption tube is heated from -20ฐC up to +10ฐC and
11-466
-------
pre-flushed with carrier gas (helium) to remove the water.
In this way, however, a substantial part of the C2-C4
hydrocarbons in also flushed of the adsorption tube and
lost for analysis. For the separation both thick film,
apolar phases capillary columns (such as CP-Sil 5 CB, 50 m
x 0.32 mm i.d. df = 5.0 /urn) and A1203/KC1 PLOT columns can
be used. Dual detection (FID and ECD) should be used by
splitting the effluent from the analytical column to the
two detectors. In order to cover the whole range from C2-
C12 hydrocarbons, the apolar column must start at -20ฐC and
programmed up to 210ฐC. Monitoring compounds in relation
to the biospheric ozone formation, such as aliphatic and
olefinic hydrocarbons as well as the alkyl aromatics,
required a chromatographic column with a high resolution of
the very volatile organic compounds, such a ethene and
ethane. For this purpose, the A1203/KC1 PLOT column is
selected. Temperature program then can start at 40ฐC,
which eliminates the use of a cryogenic unit in the gas
chromatograph. On this column, however, it is difficult to
analyze some of the halogenated hydrocarbons, such as
unsaturated freons.
The above-described analyzer is used by the Dutch National
Air Quality Monitoring Network and by the EPA during the
Summer 1990 ozone precursor study in Atlanta, Georgia. The
VOC Air Analyzer can be used where (very) low
concentrations of volatile organic components in air have
to be monitored continuously without operators present.
Th range of components that can be analyzed is:
1. Hydrocarbons C2-C^g
2. Halogenated hydrocarbons up to trichlorobenzene
3. Aromatic hydrocarbons up to trimethyl benzene
The application field is air pollution control and in some
cases industrial hygiene (especially where levels to
control are in the ppb range).
Table 1. shows the composition of an EPA calibration
mixture, which can be separated on a 5 /im CP-Sil 5 CB
column under the above described conditions. For reasons
of sensitivity and identity-conformation dual detection
should be used.
RESULTS AND DISCUSSION
Figures 2A and 2B show the results of a 200 mL outdoor air
sample at Raritan, New Jersey, being the FID and the ECD
signals. Here the apolar column program was started at
40ฐC and programmed up to 200ฐC. In Figure 3 the increase
in retention and separation is shown, using an A12O3/KC1
PLOT column under the same conditions. Typical
concentrations of the components are in the 0.1-10 ppb
range. Under the conditions described in Figure 3, a
continuous monitoring of air was realized during several
months. Part of the quantitative results are plotted in
11-467
-------
Figure 4. Similar plots are made for C4, C5, C6, C7-8 and
aromatic components.
A new method for the analysis of C2-C6 hydrocarbons and
halogenated compounds, including C5-C12 hydrocarbons, is
to inject and separate the collected air samples on two
capillary columns simultaneously. Such a combination could
be an A12O3/KC1 PLOT column and an apolar liquid phase,
such as CP-Sil 5 CB (thick film). In this way the analysis
can start at 40ฐC-45ฐC, so cryogenic cooling of the GC oven
is avoided, which gives a substantial reduction in the
liquid nitrogen consumption. In order to optimize the
separation of the chlorinated compounds a new, slightly
more polar liquid stationary phase was developed, CP-Sil 13
CB. This is also a polysiloxane phase, containing an
average of 14% phenyl/86% methyl groups in the polymer.
CP-Sil 13 CB has an excellent selectivity for the
halogenated hydrocarbons, as mentioned in EPA 624 and 502-
2. By combining FID and ECD detection, some co-eluting
peaks can be quantified independently. The stationary
phase does not contain electro-negative groups. Combined
with the low bleeding of the column this results in a very
stable baseline on the ECD trade.
CONCLUSION
The most significant characteristic of the VOC Air Analyzer
is the simultaneous sampling and analysis of (very)
volatile organic compounds at a frequency of 1 hour or
less. Using thick film WCOT or porous PLOT columns, the
compounds that can be monitored range from the unsaturated
and saturated alkanes, benzene, and the substituted
aromatics and various halogenated compounds. The high
resolution power of capillary columns allows high quality
identification and quantitation and produces information
concerning individual compounds relevant in atmospheric
processes.
11-468
-------
FIGURE CAPTURES.
Figure 1.
Figure 2A.
Figure 2B.
Schematic diagram of the Chrompack VOC Air
Analyzer, including the new pump-dryer unit.
FID signal of a 200 mL outdoor air sample,
collected at Raritan, New Jersey. Column:
CP-Sil 5 CB. Temperature: 40ฐC (4 min) to
200ฐC. Sample collection: -20ฐC.
Desorption: 220ฐC. Peak identification:
1 = benzene; 2 = toluene; 3 = ethylbenzene;
4 = p.m-xylene; 5 = 0-xylene.
200 mL outdoor air
in Figure 2A.
sample.
Peak
Figure 3.
Figure 4.
BCD signal of a
Conditions as
identification:
1 = trichlorofluoromethane; 2 = methylbromide;
3 = trichloroethane; 4 = carbontetrachloride;
5 = trichloroethene; 6 = tetrachloroethene;
7 = tetrachloroethane; 8 = hexachlorobutadiene
(see original)
Plot of the C2-C3 concentration fluctuation
during a continuous monitoring of outdoor air
at Bilthoven, The Netherlands. Sample cycle
time was one hour, sample volume: 333 mL.
11-469
-------
SAMPLE PUMP
GDป
iAHPLEl OUT
SAMPLE COLLLCTtON,*
4 DRYER AIR IN
DRYER AIR OUT
VENT
PUMP, CAL.SAMPLE,
BRYER OPTION
CALIBRATION
SAMPLE
CYLINDER
COLUMN*
-------
60 COMPONENT CALIBRATION MIXTURE
(1) Acetylene
2) Ethylene
3) Ethane
4) Propylene
(5) Propane
(6) Isobutane
(7) 1-Butene
(8) n-Butane
(9) trans-2-Butene
(10 cis-2-Butene
(11 3-Methyl-1-Butene
(12 Isopentane
(13) 1-Pentene
(14) n-Pentane
(15) Isoprene
(16) trans-2-Pentene
(17) cis-2-Pentene
(18) 2-Methyl-2-butene
19) 2,2-Dimethylbutane
20) Cyclopentene
21) 4-Methyl-l-Pentene
(22) Cyclopentane
(23) 2,3-Dimethylbutane
2-Methylpentane
3-Methylpentane
2-Methyl-1-pentene
n-Hexane
Chloroform
trans-2-Hexene
cis-2-Hexene
(31) Methylcyclopentane
32) 2,4-Dimethylpentane
33) 1,1,1-Trichloroethane
34) Benzene
35 Carbon tetrachloride
36 Cyclohexane
37 2-Methylhexane
38) 2,3-Dimethylpentane
(39) 3-Methylhexane
(40) Trichloroethylene
2,2,4-Trimethy1pentane
n-Heptane
Methylcyclohexane
2,3,4-Trimethy1pentane
Toluene
(46) 2-Methylheptane
(47) 3-Methylheptane
(48) n-Octane
(49) Perchloroethylene
(50) Ethyl benzene
(51) p-Xylene
'52) Styrene
53) o-Xylene
54) n-Nonane
(55) Isopropylbenzene
(56) n-Propylbenzene
(57) a-Pinene
(58 1,3,5-Trimethylbenzene
(59 1,2,4-Trimethy1 benzene
(60 /J-Pinene
1-471
-------
ro
E
\
z
QL
h-
Z
u
o
z
O
O
VERY VOLATILE ORGANIC COMPOUNDS
C2-C3
ethane
ethene
HOURS PAST 00:00 MAY 29 1990
3
O propane
propene
propadiene
-------
*
*
* 2A
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* * ^
* * <*
* *
* * 'T1
* * o
* *
* * in
* * T-ซ X
* * ..
* * o
* * M ^ ^
* * .. ,Tl O
V 5*C O 'Tป jT|
* * -r-l _| .
* * ! ฃti
* * O M
* * "> ro U. U.
ID
mm
P
O
4^
f^
3
C
*J
-------
*
*
*
*
*2B
*
*
*
*
*
*
*
*
*
*
*
*
*
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
jjc sje
1-1
Tf
iT)
6
(\i
C"D v.
ro
c\i _t ,-1
<7* O
O* 'T1 ฃCf
1-1 i-t >
1 Lฃ
O ^
^ i a
^ o
i-i ni
'T>
1 ซ ซ C
O O
CO 01 0) "i
i 4* e 4*
^"* fl3 fl3 ^
O TJ C 3
C
.. Q| QJ (U
i I i 1 4*
ai -r-i ---I .p
-P U. U.
-------
13
10
9
LjJl
11 12
uu
14
Figure 3 Chromatogram of an air sample. Sample: Outdoor air at Bilthoven
Station, June 22,1990, at 15.00 hr. Column: fused silica 25m x 0.53mm
i.d. AI2O3/KCI, df = 10 \JJTI. Temperature: 40 ฐC, isothermal 1 min, pro-
grammed to200 "Cat 10ฐC/min, 200 ฐC, isothermal30min. Carrier: He
0.4 bar. Detector: FID 1, 10E-12 A, 2 mV full scale, 275 ฐC. Adsorption
trap: Carbosieve SHI, Carbotrap, Carbotrap C (7.5 x 0.29 cm). Capillary
trap: Poraplot U fused silica 0.53 mm i.d., 20 \im. Sample coll.: -20 ฐC,
35 min. Sample flow: 9.5 mL/min. Sample vol: 333 mL Sample desor.:
250 ฐC, 5 min, desorption vent flow 2.5 mL/min. Cryofocusing: -150 ฐC,
5 min. Injection: 125 ฐC, 10 min. Back flushing: 270 ฐC, 10 min, flow: 20
mUmin. Valve: 200 ฐC. Injection block: 200 ฐC. Peak identification: 1 =
ethane; 2 = ethene; 3 = propane; 4 = propene; 5 = 1-butane; 6 =
prodadiene; 7 = n-butane; 8 = trans-2-butene; 9 = 1-butene; 10 =
1-butene; 11 = cis-2-butene; 12 = cyclopentane; 13= 1-pentane; 14 =
n-pentane; ISA = 3-methyl-1-butene; 15B = trans-2-pentene; 15C =
2-methyl-2-butene; 15D = 1-pentene; 15E= 2-methyl-1 -butene; 15F =
cis-2-pentene; 21A = methylcyclopentane; 21B = cyclohexane; 21C =
2-methylpentane; 21D = 3-methylpentane; 25 = n-hexane; 26 = n-hep-
tane; 27 = benzene; 28 = n-octane; 29 = toluene.
29
27
27min.
-------
GENERAL
-------
J05 DEVELOPING A UNIFORM APPROACH FOR COMPLYING WITH EPA METHODS
Peggy Sleevi. Corporate Director of Quality Assurance, Enseco, Incorporated,
2612 Olde Stone Road, Midlothian, Virginia 23113; Deborah Loring, Director of
Quality Assurance, Enseco-East, 2200 Cottontail Lane, Somerset, New Jersey
08873; Jerry Parr, Chief Organic Scientist, Enseco Incorporated, 4955 Yarrow
Street, Arvada, Colorado 80002; Nancy Rothman, Chief Scientist, Enseco
Incorporated, 205 Alewife Brook Parkway, Cambridge, Massachusetts 012138
ABSTRACT
Since the late 1970's, EPA has developed methods using GC and GC/MS technology
to support regulatory initiatives. These methods have been promulgated,
distributed, and used as contract mechanisms. Commercial laboratories have
faced a bewildering array of "approved" methods, generally utilizing identical
technology but varying in detail.
As stated in a recent EPA report to the U.S. Congress "Improved coordination
is needed in the Agency's methods development program", Enseco supports the
activities of the Environmental Methods Management Committee, created to
respond to EPA's recommendation and has drafted an approach which results in
regulatory control combined with analytical flexibility. The approach
controls critical method elements such as the procedural details, calibration,
and quality control requirements but eliminates superficial differences that
currently exist in EPA methods.
Using as a model the methods available to analyze volatile organics by purge
and trap GC/MS (624, 524.2, 8240, 8260, etc.), information is presented
comparing and contrasting the differences between EPA methods from various
sources. Data will be presented discussing the impact of varying the method
details. Finally, an approach will be presented which discusses how
laboratories can balance productivity, technical enhancement and method
compliance issues.
INTRODUCTION
Since the late 1970's, EPA has developed methods using GC and GC/MS technology
to support regulatory initiatives. These methods have been promulgated,
distributed, and used as contract mechanisms. Commercial laboratories have
faced a bewildering array of "approved" methods, generally utilizing identical
technology but varying in detail.
In response to a recent EPA report to the U.S. Congress (Adequacy,
Availability and Comparability of Testing Procedures for the Analysis of
Pollutants Established under section 304(h) of the Federal Water Pollution
Control Act, also referred to as the 518 Report), EPA has formulated the
Environmental Methods Management Committee (EMMC) to address the issue of
methods consolidation. The Committee's efforts have been described to reduce
the number of method variations labs must integrate and to allow more methods
to be developed. The methods integration group has stipulated that quality
control is an intrinsic part of the methods (1). The approach to
consolidation must eliminate superficial differences which result in
laboratories needing to run duplicative methods with differing requirements
11-479
-------
which do not impact the quality of the data. Since these methods are
performed in response to regulatory requirements, it is incumbent upon
laboratories to meet the specific requirements of each method, irregardless of
the technical merit. This results in redundancy of effort and increased
analytical costs to ensure that specific method details are met.
As a solution to this dilemma, EPA has indicated that the various methods
utilizing the same technological approach will be combined into one master
method by using the "best practices" from each method. We are concerned that
this approach will result in methods which have such stringent criteria as to
be virtually unusable. Furthermore, the methods will not have the flexibility
to meet the Agency's various regulatory needs.
This "best practices" approach is also contrary to an earlier statement by EPA
in Environmental Lab. As stated by David Friedman:
"The approach we have been taking when promulgating
analytical methods often has been counter productive. It
has stifled creativity; it has led to poor analytical
results; and it has, in some cases raised the cost of
testing ... We have to move toward performance
standards, not design standards. We must specify what
needs to be done, including data quality objectives, not
how to do it." (2)
We are presenting here an alternative approach to EPAs recommendation to
use the "best practice" from each method to get one method. We propose a
minimum acceptable practice (MAP) to be used in concert with Data Quality
Objectives (DQOs) to define the analytical requirements for each project.
The DQOs and analytical requirements must be documented in a project
specific QAPjP which is agreed to as part of the project planning process.
In addressing this problem we have evaluated what aspects of a method are
critical to the execution of the method. Minimum QC criteria should be
specified outside of the method as proposed for SW-846 (55 Federal Register
4440). The ASTM document "Standard Practice for Generation of
Environmental Data Related to Waste Management Activities: QA/QC Planning
and Implementation" addresses the minimum acceptable practices to assure
the quality of field and analytical activities. Types of control samples
that are used to monitor method performance are described externally to the
methods and apply universally to all techniques amenable to such external
controls. Additional use of matrix-specific QC must be related to the
project needs based on the DQOs and not be mandated as a laboratory
exercise. The analyte list and QC sample elements are therefore not
mandatory elements of the method. Calibration criteria, sample size and
preparation procedures are, however, inherent method elements. The
specifics of these method elements may be variable and subject to
validation.
11-480
-------
Supporting information with respect to analytes which may be determined by
a method, in addition to any performance criteria obtained under specific
conditions should be included as appendices to the method.
Such an approach will improve the quality and usability of data while
providing a cost-effective means to generate environmental data. It will
also allow for technical enhancements to a method to allow for innovation
and improvement. Under the current system you can follow the method
exactly and obtain poor quality, unusable data. If you deviate from the
method to improve the quality and usability of the data you may be guilty
of non-compliance. We must ensure data integrity by requiring those
elements that are critical for integrity, while recognizing those elements
that are not critical.
Each laboratory must be required to demonstrate method proficiency based on
the specifics of the method as performed in that laboratory. This will
allow for differences in GC columns, temperature programs, and target
analytes to suit the project requirements, and assure the performance of
quality control as applied to that method. Each time a change is made to
how the method is executed or if additional analytes are to be included a
rigorous validation procedure must be performed. A proposal for initial
demonstration of proficiency and validation is described in this paper.
In this paper we address the applicability of this approach to the methods
for the analysis of volatiles by GC/MS, specifically Methods 624, 524.2,
8240 and 8260. The procedural differences in these methods have been
adequately described elsewhere and were not the focus of our efforts (3).
MINIMUM ACCEPTABLE PRACTICES
An analytical procedure should provide enough detail to allow an
experienced laboratory unfamiliar with the procedure to generate equivalent
data. Thus, extensive procedural details are required to be written into
the method. Examples of this level of detail include concentration of
calibration standards, mass range, extraction solvent, sample size,
internal standards, usable method performance data, etc. However, very
few, if any of these details should be mandatory. Rather, as discussed in
more detail later, this descriptive information provides the basis for the
performance data presented in the method. Other techniques can be used
provided they result in equivalent or better performance.
This section presents two examples of the problems with the current
approach and presents a proposed solution.
The first example relates to the retention time window used for identifying
compounds by GC/MS. The following "requirements" were found:
Method 524: "11.2.1 The GC retention time of the sample component
(1983) must be within t s of the time observed for that same
compound when a calibration solution was analyzed.
Calculate the value of t with the equation:
t = (RT)l/3
11-481
-------
where RT = observed retention time (in seconds) of the
compound when a calibration solution was analyzed." (4)
Method 524.2: "11.1.1 The GC retention time of the sample component
(Rev. 2.0, 1986)* must be within 10 s of the time observed for that for
that same compound when a calibration solution was
analyzed." (5)
* Note: As a comment on the current disarray of methods, we found four
versions of Method 524.2, two of which on the surface were stated to be the
same (Revision 2.0) and are required in the regulations, but are
substantially different. Currently both the 1986 and 1988 versions are
promulgated (40CFR141.24). However, the 1988 revision is required for
compounds 9 through 18 and the 1986 version required for the remaining
compounds (56 Federal Register 3526). Thus, the entire volatile analyte
list for drinking water must be performed using two distinct versions of
the method with different requirements. Furthermore, Revision 3.0,
available in the public domain, is not an approved method.
Method 524.2: "11.6 The GC retention time of the sample component
(Rev. 2.0, 1988) should be within three standard deviations of the mean
retention time of the compound in the calibration
mixture." (6)
Method 624: "12.1.2 The retention time must fall within + 30s of the
retention time of the authentic compound." (7)
Method 8240: "7.5.1.1.1 The sample component RRT must compare within
+ 0.06RRT units of the RRT of the standard component.
For reference, the standard must be run within the same
12 hours as the sample." (8)
As shown above, our review found five different "requirements" for
establishing this retention time window. Complete and full method
compliance would require that laboratories assure that the "correct"
approach was used based on the method purportedly used. Since the
retention time window is a minor component of identification in GC/MS, we
believe that these "requirements" are generally ignored. As an interesting
exercise, we evaluated the data from a volatile calibration standard
processed using current GC/MS target compound software. We believe that
the system would correctly identify the compounds using any of the
definitions.
A more appropriate wording of this section would be:
Identification of target compounds is based upon both
retention time and mass spectral agreement. The data
contained in this method were based on a + 0.06 RRT
window. Other approaches may be used if they provide
equivalent performance.
As another example, consider the language in Section 7.2.3 of Method 8240
which states "Prior to use, condition the trap daily for 10 minutes while
11-482
-------
backflushing at 180ฐC with the column at 220ฐC." In this one sentence,
there are six requirements, "Prior to use", "daily", "10 minutes",
"backflushing", "180ฐC", and "220ฐC". Accordingly, if someone was to
develop a different trap which could be conditioned at 170ฐC for 8 minutes,
it could not be used. Again, we would rewrite this sentence to read:
The trap must be conditioned to maintain performance.
It is recommended that the trap be backflushed at 180ฐC
for 10 minutes. Other approaches can be used provided
the trap maintains performance as measured by analyses
of QC samples.
In summary, there are many imperatives contained in methods. We should
always be questioning the intent of the imperative and verifying that it
is a requirement. Is there no alternative? Do we really mean "must"?
Or, would we allow a deviation? If the "requirement" is merely a
description of what was done during method development/implementation,
then it should not be a requirement. We do believe that the imperative
writing style adds clarity. For example, "Purge 5 mis" is much clearer
than "5 mis is purged". However, these imperative statements should be
followed by language that indicates other approaches could be performed.
A review of EPA methods for measuring volatile organics by purge and trap
GC/MS indicate that there are thousands of requirements. For example, we
found 28 requirements in Section 7.3 of Method 8240 relating to daily
calibration. This is in addition to the other 50 or so calibration
requirements in Method 8000 and the requirement for initial calibration in
section 7.2. We believe these requirements could be reduced to a few
critical elements, minimum acceptable practices, and all others reworded
to indicate the conditions used. For example, for calibration by GC/MS we
believe the key requirements are:
1. The mass spectrometer must generate reliable mass spectra, as
demonstrated by the measurement of a reference mass marker
compound such as bromofluorobenzene.
2. A predictable relationship between response and concentration must
be established. This calibration response must be used to define
the upper and lower limits of quantitation.
3. The calibration must be shown to be in control during instrument
operation.
How then do we assure data quality and comparability? Two ways. First,
data quality objectives, established for each and every analysis, are used
to specify the requirements expected of the method. For example, if my
objective is to measure vinyl chloride in groundwater at 10 ppb with a
precision of less than 25%, I would establish the procedural details around
this objective. I might for example analyze a larger sample, use vinyl
chloride as my matrix spike and specify a 15% RSD for initial calibration.
11-483
-------
If my objective was to survey groundwater for the Appendix IX list of
volatiles, I might use representative compounds as matrix spikes and have a
larger %RSD for initial calibration, i.e., use the default conditions in
the method.
If the method is written to allow for the most general objective, then
requirements can be superimposed for more specific objectives.
The second process to assure data quality and comparability is to establish
a rigorous method validation process. This process, described later, will
allow virtually any change to the method, as long as the change is
validated. Otherwise, the information contained in the method become the
requirement. The two processes, DQOs and method validation, must be used
in tandem.
This approach is not significantly different from the approach in EPA-
600/8-83-020, "Guidelines and Format for EMSL-Cincinnati Methods". (4) In
fact, the original Method 524, appended to EPA's report, was the closest
example to our approach of the methods surveyed. Unfortunately, many of
the niceties of this method were eliminated in subsequent revisions. For
example, the original method allowed for alternate traps if "it has been
evaluated and found to perform satisfactorily". This language was
eliminated from the revisions.
DATA QUALITY OBJECTIVES
In 1984, the Quality Assurance Management Staff (QAMS) at EPA proposed that
the design of environmental data collection programs be based on the
development of Data Quality Objectives (DQOs). DQOs are statements of the
level of uncertainty that a decision maker is willing to accept in results
derived from environmental data, when used in a regulatory or programmatic
decision (9). The DQO process is designed to ensure that the quality of
data is compatible with the requirements of the decision making process.
By utilizing this process, the participating parties can design an
environmental data collection program and its associated QA/QC program
which results in data which satisfies the needs of decision makers in a
cost effective manner.
In the DQO process, the decision maker must describe the decision, why data
are needed and background on the problem. The type of information needed
for the decision is described with respect to the scope and type of data
required. The use of the data must be defined, along with the importance
of the data for making a decision. The consequences of an incorrect
decision resulting from inadequate environmental data are also described.
The extent to which false positives and false negatives can be tolerated
must be defined. In addition, a description of the available resources to
fund the project must be stipulated.
11-484
-------
The second stage of the DQO process impacts the laboratory. It is in this
portion that the specific data required are defined with respect to
analytes, matrices, spatial and temporal requirements. The results to be
derived from the data are stipulated. That is, are the data to be assessed
for a particular statistic, for values relative to a regulatory action
level, or for determination of baseline or background levels. The desired
performance with respect to precision and accuracy, and acceptable levels
of false positives or false negatives are detailed. It is this portion of
the DQO process which should effect the specific method requirements and
the laboratory should have the flexibility to establish and validate method
criteria to meet the needs of the project.
Matrix-specific QC should be defined in conjunction with these elements
rather than being stipulated as an inherent part of the method. The same
method can be used to meet a regulatory limit or provide data to determine
baseline levels and therefore the frequency and makeup of matrix-specific
QC (spike levels and spike components) should be controlled by the type of
information being sought, not specified in the method itself.
Every project should go through this evaluation and specification process
in which the DQOs are clearly defined. The consensus of the lab, the data
user and the regulator must be forged before the work begins. The process
should be formalized in a project-specific Quality Assurance Project Plan
(QAPJP).
We therefore believe that the methods should have fewer elements relating
to the project objectives and that these elements should be addressed by
the DQO process. For example, we believe that analytes should not be
listed in the method. Rather, a list of compounds evaluated by the method
and their performance (precision, bias, detection limits) should be
contained as an Appendix to the method. (As a side benefit, this list
could be expanded to include additional analytes and/or additional
performance data without rewriting the method.) As another example, the
components used as matrix spikes, the spike levels and spike frequency
should not be specified in the method, but in the project objectives. For
example, we are continually amazed at the number of customers who are
interested only in PCBs but who require representative pesticide compounds
be spiked, because the compounds are listed in Method 8080.
By segregating project objectives from method requirements, data users will
be forced to make decisions based on their objectives, and not on some
default condition in the method. For example, the ASTM Standard Practice
addresses the issue of matrix spikes by first defining a matrix spike as
"an aliquot of sample spiked with a known concentration of target
analyte(s)" and then requiring matrix spikes to be analyzed "based on the
DQOs of the data collection activity". (10)
Thus, the data user must decide on the compounds and the frequency.
Obviously, the laboratory must be more involved in the overall process.
However, we believe that this approach will improve the quality and
11-485
-------
usability of the data. As presented by Jim Barren of OERR at the
Analytical Methods Caucus in San Diego in March 1991, EPA has focused
almost entirely on data authenticity. (11) This approach implies that
"quality" is achieved when the method is followed exactly and that "fraud"
is committed when the method details are not performed. As stated by Mr.
Barron, the method can be followed exactly and unusable data of poor
quality can be generated. Conversely, useable data of high quality
obtained by a non-approved modification of the method can be rejected. We
believe that our approach will not only improve the quality and usability
of the data but will also restrict charges of fraud to those instances
where actual fraud occurs.
This approach obviously increases the complexity of the laboratory work and
could result in bottlenecks which would prevent work from ever being
performed. We believe the solution to this dilemma is to establish default
conditions which are used in many situations, such as those determined from
a general survey. In these situations, the default conditions could be
written into the method. For example, the default condition for an
Appendix IX volatile analysis could be the analytical conditions in the
method, representative target compounds for matrix spikes, matrix spikes
every twenty samples per project, a five point calibration for all
compounds with a 30% RSD for all compounds with two allowed out. The
default conditions for analysis of TCLP leachate for toxicity
characteristic compounds might involve modifying the sample size relative
to the method, using all target compounds as matrix spikes, spiking every
sample and requiring a 25% RSD for every compound with a three point
calibration.
This approach requires that the laboratory and the data user agree on the
objective, prior to the initiation of the project. We have developed a
project initiation checklist which addresses these types of issues. In our
process, we seek to obtain consensus from the data users on the following
issues:
o Sample containers, preservatives, holding times
o Operational details - sample size, calibration, etc.
o Quality control samples - frequency, spike components, spike
levels, control limits
o Detection limit requirements
o Report format, content
In those situations where the objective are not clearly known, we have
established default conditions which relate to these issues. These default
conditions are documented in our laboratory QAPP, in method SOP's, and in
other reference documents. For example, we have a document which lists the
spike components and spike levels for routine methods and another document
which lists analytes and reporting limits for every test performed.
11-486
-------
METHOD VALIDATION
As discussed in the Section 518 report, a validated method is a method
based on sound technical principles that can be used routinely to achieve
some base level of performance. (12) We believe that method performance
data obtained by the author of the method should be included in an appendix
so that potential users can evaluate the usefulness of the method.
We also believe that laboratories who propose to use the method must
demonstrate a basic capability to generate comparable data using the
method. While we believe that the existing approach in Methods 624 and
8240 adequately address this issue, we would recommend a more rigorous
approach. We propose that the method contain performance data for a
limited number of analytes, such as those on the priority pollutant or
target compound lists. This performance data would be used by laboratories
to assess their performance. Laboratories wishing to use the method would
be required to analyze seven spiked samples (ideally standard reference
materials) at concentrations spanning the working range of the method.
Statistical tests (f-test and t-test) would then be performed to
demonstrate equivalency.
This process would demonstrate that a laboratory has the basic capability
to perform the method for a limited number of analytes. On going quality
control activities would then demonstrate the laboratory's performance on a
continual basis.
The more important issue relates to expanding the analyte list or to
modifying the method. We believe that the methods should be sufficiently
flexible to allow for extensive modification. However, to provide a
measure of control, we recommend that any change to the method and any
addition to the basic analyte list be permitted only if a rigorous
validation is performed. While on the surface this recommendation may seem
overly stringent, if the proposed change will substantially improve the
method (better quality, cheaper, faster, safer, etc.) then the effort will
be justified. Otherwise, there is no need to change.
Method validation must address method characteristics such as:
o detection limit,
o working range,
o precision,
o ruggedness,
o matrix,
o analytes,
o comparability, and
o bias
Adding a new analyte is distinctly different from modifying an existing
validated method. In the first case, very little if anything is known
about the performance and thus a more extensive validation must be
performed. We are developing an internal approach which involves four
activities. The initial activity is to establish the working instrumental
range. Once the working range is known, spiked samples, spanning the
working range are carried through the method. If this process is
11-487
-------
successful, ruggedness testing is performed. Finally, blind spiked samples
are analyzed. (13)
From these analyses, sufficient information will be provided to thoroughly
document all of the method characteristics for the new analytes.
For an existing analyte in an existing method, the method characteristics
should be known. The key when changing the method conditions is to
demonstrate that the change did not significantly affect the performance.
The process involved would require 14 analyses of a spiked sample, seven
analyses under the existing method conditions and seven using the proposed
modification. Statistical tests (f-test and t-test) would then be
performed to demonstrate equivalency.
The previous section discussed replacing method requirements with method
descriptions. These method descriptions would be the default requirements
unless an equivalency study was performed. Thus, for many of the method
details, a proposed change would not be justified relative to the efforts
involved. However, if the change was important, e.g. packed column versus
capillary column there would be a system to allow for the change.
No equivalency process will address every sample and every method condition
that may be experienced. The purpose of this validation process is to
demonstrate that the proposed change is fundamentally sound. Laboratory
controls and project specific quality control activities are used on an
ongoing basis to assess the quality of the laboratory work.
SUMMARY
As David Friedman discussed in his Environmental Lab article, this approach
will give analysts (and laboratories) more freedom. For this approach to
work, the laboratories and data users must therefore accept more
responsibility to ensure that the work is performed correctly. Freedom to
change methods will not result in constantly varying methods. This
approach will not meet the laboratory needs any better than the current
requirements. The driving force for this approach is the achievement of
technically sound, defensible data that does not burden the laboratory with
overly restrictive requirements.
We recognize that rewriting all of the existing methods to incorporate this
approach is a formidable challenge. In the interim, we would request the
EMMC to introduce language into each method which will allow deviations
based on DQOs and validation data.
Finally, the establishment of DQOs prior to initiation of the work, must be
mandated by EPA. It can no longer be acceptable for data users to request
an analysis from a laboratory without specifying the requirements.
11-488
-------
REFERENCES
(1) Fisk, Joan, "Methods Consolidation Efforts Between Superfund and
RCRA," presented at the 14th Annual EPA Conference on Analysis of
Pollutants in the Environment.
(2) Friedman, David, "New Directions in Method Development,"
Environmental Lab, August/September 1990.
(3) Mealy, R.6., "Towards a Unified Approach", Environmental Lab,
September 1989
(4) EPA-600/8-83-020, "Guidelines and Format for EMSL-Cincinnati
Methods", August 1983.
(5) "Methods for the Determination of Organic Compounds in Finished
Drinking Water and Raw Source Water," EMSL, Cincinnati, September
1986.
(6) EPA/600/4-88/039 "Methods for the Determination of Organic Compounds
in Drinking Water," EMSL, Cincinnati, December 1988.
(7) 40CFR Part 136
(8) SW-846, Third Edition, November 1986.
(9) Development of Data Quality Objectives, Description of Stages I and
II, QAMS, July 16, 1986.
(10) ASTM ES 16, "Standard Practice for Generation of Environmental Data
Related to Waste Management Activities: QA/QC Planning and
Implementation."
(11) Barron, Jim, "Quality Assurance Issues" USEPA Analytical Methods
Caucus, March 1991.
(12) EPA/600/9-87/030, "Availability, Adequacy, and Comparability of
Testing Procedures for the Analysis of Pollutants Established Under
Section 304(h) of the Federal Water Pollution Control Act," September
1988.
(13) Winkler, P., et al, "Assuring Reliable Laboratory Data Via Rigorous
Method Validation," in Preparation.
11-489
-------
HQg PERFORMING TCLP ANALYSES TO GET MEANINGFUL DATA
Kyle Dolbow, Ph.D. and Jody Price, IEA Inc. of New Jersey, 62L
Route 10, Whippany, NJ 07981
The Hazardous Waste Regulations which were promulgated in March of
1990 have had significant impact on waste generators and the
commercial laboratory. The major change in the regulation is the
substitution of the Toxicity Characteristic Leaching Procedure
(TCLP) for the Extraction Procedure Toxicity test. Upon closer
examination of the regulation, IEA Inc. of New Jersey discovered
several difficulties for waste generators and many new challenges
for the laboratory. Typical problems are:
1.0 The waste generator may not be able to tell how many
analyses will be reguired (and the total cost) until
after samples have been submitted to the laboratory and
initial testing has been performed.
2.0 With many types of samples, matrix interferences severely
affect the analysis of one or more fractions.
3.0 For multi-phase samples, data from up to six complete
TCLP analyses has to be combined through defined and
complicated formulas to a final set of numbers which is
then compared to regulatory levels to make a
hazardous/nonhazardous decision.
4.0 The regulatory levels for the individual analytes in a
particular fraction, such as semivolatiles, cover a range
greater than the linearity of the instrument used to do
11-490
-------
the analysis.
5.0 Matrix spike levels and other QC requirements are much
different from that of "normal" SW846 methods.
All of the above concerns caused IEA, Inc. of New Jersey to take
an integrated approach to TCLP analysis. This approach includes
all laboratory staff understanding TCLP data quality objectives,
client communication, optimized sample preparation methods, TCLP
specific analytical schemes, and an automated computer data
handling strategy. The end result is a high quality product which
specifically addresses the data quality objectives of TCLP and
presents the results in a format that is both easy to read and
understand. This paper describes this integrated approach in
detail.
11-491
-------
107 TOTAL CYANIDE BY PHOTOLYSIS
Jenner Gutierrez, SAIC, 8400 Wespark Drive, Mclean, VA 22102
ABSTRACT
The purpose of this study was to develop and examine a viable method which
could provide better quantification of the total cyanide (both soluble and
insoluble organometallic complexes) content in a waste. The methods which
have been employed comprise the derivatization and gas chromatographic
separation of benzylic nitrile derivatives. The results have shown promise
in that the reaction occurs spontaneously and at room temperature. Present
studies are currently aimed at reducing the overall detection limit and at
determining the degradation efficiency of the cyano-metallic complexes through
photolysis. At the conclusion of the study, in-house samples will be prepared
to compare the prospective with the current EPA methods SW-846 9010 and 9012.
11-492
-------
-I QO AMMONIA and TOTAL KJELDAHL NITROGEN DETERMINATIONS USING FLOW
INJECTION ANALYSIS WITH GAS DIFFUSION
J. Philip Calvi, Perstorp Analytical, Inc., 2875 C Towerview
Road, Herndon, VA 22071; Bernard Bubnis, Novatek, 10 West
Rose Avenue, Oxford, OH 45056; Jan-Ake Persson, Tecator AB,
Box 70, S-26321 Hoganas, Sweden.
Abstract
Free ammonia and total Kjeldahl nitrogen (TKN) in wastewater
have been determined by flow injection analysis (FIA) using a
gas permeable membrane. Final effluent from ten sample sites
representing the top five standard industrial code (SIC)
classifications for the nitrogen parameters were tested.
Results comparing this new method with the established EPA
methodologies are presented.
FIA methods introduce an aliquot of sample using an injection
valve. The valve generally is capable of delivering 40 to 200
uL of sample into a reaction stream which produces ammonia
gas. The gas permeates an in-line membrane to enter into an
indicating acceptor stream where a color change takes place.
Data indicate that the gas diffusion methods give results
equal to or better than currently approved EPA protocols.
Operating ranges were determined to be from 0.02 to 10 mg/L
with a method detection limit (MDL) of 0.006 mg/L for ammonia
and from 0.2 to 10 mg/L with a MDL of 0.02 mg/L for TKN.
The gas diffusion technique is simpler than other automated
nitrogen procedures. It does not require harmful chemicals;
is not sensitive to Kjeldahl digest pH; and is capable of
producing a result in 70 seconds.
Introduction
The growing concern about the environment and the quality of
drinking water have caused a substantial increase in the
number and frequency of analyses performed by laboratories.
The nitrogen parameters (ammonia and TKN) are a particularly
important indicator of the quality of water and soils and are
monitored routinely. Traditionally these parameters are
measured using distillation techniques, ion selective
electrodes and a variety of automated colorimetric procedures.
1-493
-------
A number of drawbacks exist using the current EPA approved
chemistries. The phenate method for ammonia and TKN (EPA
350.1/351.1) requires the use of a high concentration of
phenol. Reagent toxicity and disposal issues certainly need
attention when this method is used. Similarly, the salicylate
methodology for TKN (EPA 351.2) uses sodium nitroferricyanide
which is also classified as a hazardous chemical. In addition
to the health risks associated with the phenol reagents used
in the EPA methods 350 and 351, the smell of the reagents can
be a nuisance.
The automated EPA chemistries can be difficult to operate.
Both methods are temperature dependent requiring close
control. The salicylate method is prone to precipitation
problems which cause clogging of the reagent channels. The
most troublesome aspect to using these methods is the
influence of pH on method sensitivity. Strong buffers are
required to maintain pH control. The TKN analysis is
particularly a problem in terms of final sample pH. During
sample digestion, the organic matter in a sample will consume
acid. Therefore it is possible that the indivdual digestion
tubes can contain slight acid variations.
Block digestion as a sample preparation for FIA gas diffusion
methods proved to be efficient and in general troublefree
during the testing phase of this work.
The FIA gas diffusion technique addresses the automated
chemistry method drawbacks. The reagents used are sodium
hydroxide, water and a colorimetric indicator. An aliquot of
sample is injected into a carrier stream which is merged with
sodium hydroxide to raise the pH of the sample. Under
alkaline conditions ammonium ion becomes ammonia gas which
passes through an in-line gas permeable membrane into an
interference free colorimetric indicator. The color change of
the indicator is monitored at 590 nm and is proportional to
the amount of ammonia that passed through the membrane,
Variations in the acid content of samples is overcome by using
an excess amount of sodium hydroxide. The cycle time of the
method is 70 seconds.
In 1990, Tecator AB (Sweden) field tested and submitted the
gas diffusion method to the EPA for nationwide method approval
and inclusion in the Federal Register. This process was
performed as outlined in the EPA bulletin entitled
"Requirements for Alternate Test Procedures for Inorganic
Parameters in Non-Continuous National Pollutant Discharge
Elimination System Monitoring". The results of this study are
summarized in this paper.
11-494
-------
Sample Site Selection
In accordance with the June 20, 1990 protocol, ten (10)
industrial sites representing five (5) SIC listings for each
parameter supplied by EPA were sampled (2 per SIC site) and
are listed in Table 1.
Table 1 SIC Listings for Nitrogen Parameters
Ammonia TKN
SIC Industry
4952 sewerage
2621 paper mills
2869 industrial chemicals
4911 electrical services
2911 petroleum refining
SIC Industry
4952 sewerage
2621 paper mills
2869 indust. chemicals
1475 phosphate rock
2611 pulp mills
Reference Methods
The methods chosen to compare the automated FIA procedures
were the ion selective electrode procedure EPA methods 350.3
and 351.4. These methods were chosen since the EPA had a
large data base to compare results generated by the proposed
methods. In the case of TKN, the electrode method requires
block digestion. This feature was appealing since our
procedure of choice is block digestion and not the macro
digestion procedure commonly used with the distillation
methods. Further, EPA was adamant that no deviations from the
referenced methods take place. Since the only other block
digestion sample preparation procedure approved used the
salicylate chemistry, our choice was to use the electrode
procedures.
Experimental
Sample Collection and Preservation
Samples were collected in either glass or polyethylene
containers. The samples were acidified by the addition of a
sufficient amount of cone H2S04 to lower the pH to <2 followed
by refrigeration at 4ฐC. Using this procedure, the maximum
allowable holding time is 28 days.
11-495
-------
Analysis Requirements
ATP submission to the EPA can be undertaken for nationwide use
(NW) or limited use (LU) by EPA Regional, State or commerical
labs and individual dischargers. Table 2 describes the number
and type of required analyses and quality control checks that
need to be presented in the ATP submission.
In summary, for nationwide ATP approval a method submittal
must include 250 analysis results from the top five (5) SIC
classifications for a particular parameter (125 each for the
approved and proposed methodologies) and 38 quality control
checks for a total of 288 pieces of data.
************************************************************
Table 2 Effluent Sample and Subsample Analytical Requirements
Type Applicant Analyses Quality Control
unspiked spiked total known unknown total
NW Any 10 240 250 25 13 38
LU EPA 5 120 125 13 7 20
Regional
State or
commerical
LU Indiv. 5 60 65 7 4 11
discharger
Digestion Procedure
A block digester was pre-heated to 160 ฐC. 100 mL of sample or
an aliquot of sample diluted to 100 mL was placed in each
digestion tube. Sulfuric acid-mercuric sulfate-potassium
sulfate solution was added to each sample tube. Two boiling
rods were placed in each tube. A fume exhaust manifold was
placed over the digestion tubes which were then lowered into
the preheated block for one hour. The block temperature was
then raised to 380ฐC for one and one half hours. The tubes
were removed from the digester; the boiling rods rinsed and
the residue diluted to volume.
11-496
-------
Apparatus
FIA System
Tecator FIA System to include:
Injection valve capable of injecting 40 to 200 uL
samples
Tecator gas diffusion chemifold
Thermostat
590 nm detection system
Digestion Apparatus
Tecator aluminum block digester (6 or 20 place)
Fume removal manifold
Digestion tubes, 250 mL
Boiling rods
FIA System Operating Information
Injection time 20s
Cycle time 70s
Analysis rate 50/hr
Sample loop 40 - 200 uL
Temperature 3 0 ฐ C
Wavelength 590 nm
Pathlength 10 mm
Flow Diagram Information (See Figure 1)
Mixing coil: 100 cm x 0.7 mm id
(temperature - 30"C)
Flow rates: Sample (blk/blk) =1.4 mL/min
Reagent 1 (or/or) = 1.8 mL/min
Reagent 2 (or/wh) =0.9 mL/min
Indicator (blk/blk) = 1.4 mL/min
Reagents:
Ammonia TKN
Reagent 1 water 5 N NaOH
Reagent 2 0.5 N NaOH 5 N NaOH
11-497
-------
Revised 3/91
Figure 1
Ammonia/TKN Flow Injection Gas Diffusion Manifold
Reagent 1
Reagent 2
Indicator
Thermostat
/ Gas
/ Diffusion
Cell
HA SYSTEM OPERATING INFORMATION
Injection time:
Cycle time:
Analysis rate:
Sample loop:
Temperature:
Wavelength:
Pathlength:
Evaluation:
Gain:
20s
70s
50/hr
30ฐC
590 nm
10 mm
peak height
1
FLOW DIAGRAM INFORMATION
Mixing coil 1: 100 cm x 0.7 mm i.d. (temperature ซ 30ฐC)
Flow rates:
Sample (blk/blk) = 1.4 mL/min
Reagent 1, NaOH (or/or) - 1.8 mL/min *
Reagent 2, NaOH (or/wh) = 0.9 mL/min
Indicator (blk/blk) = 1.4 mL/min
* For Aanonia Reagent 1 = Water
11-498
-------
Quality Assurance
Each laboratory using these methods in regulated environmental
monitoring is required to operate a formal quality
assurance/control program. The minimum initial requirements
of this program consist of the demonstration of the
laboratory's capability with these methods. On a continuing
basis,the laboratory should check its performance (accuracy
and precision) by analyzing reagent blanks and check
standards, fortified blanks and/or fortified samples
preferably at a minimum frequency of 10% of the total samples
analyzed by the methods. The laboratory should maintain the
performance records that define the quality of the data
generated with the method.
Method Detection Limit (HDL)
The procedure for determining the MDL is outlined in 40 CRF,
Part 136, Appendix B, Rev. 1.11. A refined second document in
Environmental Science and Technology (1981, 12, 1427) by EPA
personnel further explains the MDL calculation. The MDL is
defined as the minimum concentration of a substance that can
be measured and reported with 99% confidence that the analyte
concentration is greater than zero and is determined from
analysis of a sample in a given matrix containing the analyte.
The MDL for ammonia and TKN was experimentally calcuated to be
0.0064 and 0.023 mg/L respectively (Table 3).
Results
Linear Range
Calibration curves for the ammonia and TKN gas diffusion
methods were linear over the concentration ranges tested.
Figure 2 is the calibration curve for ammonia in the 0.02 -
2.0 mg/L range (gain setting =5). It is described by y = -
0.0092 + 0.5578x R = 1.00. A second calibration for ammonia
in the 0.2 - 10 mg/L range (gain setting = 1) was run and is
described by y = -0.0046 + 0.0969x R = 1.00. The TKN method
was linear over the 0.2 - 10 mg/L range. The TKN curve is
described as y = -0.003 + 0.0713x R = 1.00.
Accuracy and Precision
Data indicate that the gas diffusion methodology is capable of
measuring ammonia and TKN levels over the indicated ranges
with high accuracy and good precision. Results are presented
in Tables 4 and 5. Analysis of EPA unknown samples were
carried out over the course of the work (3 months). Results
are presented in Tables 6 and 7. EPA has yet to declare the
value but the five (5) individual ampules of each unknown is
consistent over the three (3) month testing period.
11-499
-------
Figure 2
1.0-
0.8-
0.6-
i
8
0.4-
0.2-
0.0
NH3, T-W-0020-1, High Range (Revised 3/91)
y = - 0.0046 + 0.0969x R . 1.00
8 10
Concentration (mg/L)
II-500
-------
SIC Sample Analysis
A sample data set which compares the EPA accepted methodology
(ISE) and the gas diffusion method for ammonia for a SIC
sample is presented in Table 8. Table 9 shows similar data
for a TKN SIC sample.
Compiled data for the entire data set are shown in Tables 10
and 11.
Table 3 MDL Information
TKN Ammonia
Mean (n = 7) 0.329 0.0175
St Dev 0.0063 0.0023
S 2 0.00011 0.000006
Su2 0.000039 0.000005
b
F.95C6.6, 4'28 4'28
S 2/S * 2.78 1.18
a ' b
S 0.0087 0.0024
pooled
MDL (mg/L) 0.023 0.0064
************************************************************
************************************************************
Table 4 Ammonia Accuracy and Precision
Known Mean St Dev RSD (%) Bias % Recovery
0.04
0.25
0.51
1.00
4.00
9.05
0.04
0.24
0.51
0.99
4.08
8.90
0.01
0.02
0.01
0.02
0.04
0.05
25.00
8.33
1.96
2.02
0.98
0.56
0.0
-0.01
0.0
-0.01
+0.08
-0.15
100.0
96.0
100.0
99.0
101.7
98.3
************************************************************
1-501
-------
************************************************************
Table 5 TKN Accuracy and Precision
Known Mean St Dev RSD (%)
Bias % Recovery
0.20
0.30
0.90
2.50
4.00
8.00
0.20
0.30
0.91
2.53
3.95
7.99
0.02
0.03
0.01
0.01
0.03
0.25
10.00
10.00
1.10
0.40
0.76
3.13
0.0
0.0
+0.01
+0.03
-0.05
-0.01
100.0
100.0
101.1
101.2
98.8
99.9
************************************************************
Table 6 Analysis of EPA Ammonia Unknown Samples
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5
Mean
St Dev
Unknown fl
20.66
21.13
20.69
20.35
20.91
20.75
0.26
Unknown #2
3.55
3.49
3.46
3.42
3.53
3.49
0.05
Unknown f3
0.98
1.
1,
,00
,18
0.96
0.93
1.01
0.09
Table 7 Analysis of EPA TKN Unknown Samples
Unknown fl Unknown I2 unknown #3
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5
Mean
St Dev
0.73
0.72
0.75
0.76
0.71
0.73
0.02
11.37
10.30
10.60
11.00
11.00
10.85
0.37
12.59
12.91
13.34
13.05
13.00
12.98
0.24
************************************************************
11-502
-------
Tecator AB Alternate Test Procedure Data
Nitrogen, Ammonia, Method T-W-0020-1
Automated FIA Gas Diffusion Reference No. N90 0018
Table 8
Sample A Sample B
ISE FIA
388 379
388 388
387 381
388 383
< 1 3
Source: SIC 4952 Sewerage Plant #1
Note: Ammonia results in mg/L; ISE Method EPA 350.3
s
CO
Technique
Trial 1
Trial 2
Trial 3
Mean
St Dev
ISE
0.19
0.17
0.16
0.17
0.01
FIA
0.20
0.20
0.20
0.20
0.0
Sample c
ISE FIA
809 794
780 802
789 800
793 799
10 3
Sample D
ISE FIA
1208 1189
1194
1183
1195
9
1183
1187
1186
2
-------
Tecator AB Alternate Test Procedure Data
Nitrogen, TKN, Method T-W-0021-1
Block Digestion, Automated FIA Gas Diffusion EPA Reference No. N90 0015
Table 9
Sample A
Technique
Trial 1
Trial 2
Trial 3
Mean
St Dev
I8E
0.18
0.19
0.18
0.18
0.0
FIA
0.14
0.17
0.17
0.16
0.01
Sample B
ISE
8.2
7.9
8.2
8.1
0.1
FIA
8.0
8.0
8.1
8.0
0.01
Sample C
ISE
16.1
16.7
15.7
16.2
0.4
FIA
16.1
15.8
16.4
16.1
0.2
Sample D
ISE
24.2
24.8
24.6
24.5
0.2
FIA
24.2
24.4
23.4
24.0
0.4
Source: SIC 4952 Sewerage Plant #1
Note: TKN results in mg/L; ISE Method EPA 351.4
-------
Table 10
Combined Ammonia Data
SIC 4952 Sewerage
Technique Mean
FIA 0.16
ISE 0.16
FIA 393
ISE 399
FIA 800
ISE 782
FIA 1202
ISE 1182
SIC 2621 Paper Mills
FIA 0.16
ISE 0.14
FIA 387
ISE 398
FIA 771
ISE 764
FIA 1209
ISE 1196
SIC 2869 Industrial Chemicals
FIA
ISE
FIA
ISE
FIA
ISE
FIA
ISE
0.19
0.17
388
398
794
793
1197
1198
SIC 4911 Electrical Services
FIA
ISE
FIA
ISE
FIA
ISE
FIA
ISE
0.18
0.16
403
388
808
794
1202
1191
SIC 2911 Petroleum Refining
FIA
ISE
FIA
ISE
FIA
ISE
FIA
ISE
0.20
0.15
392
392
806
799
1206
1200
Confidence Intervals
-3s +3s
0.12
0.10
357
361
791
722
1150
1148
0.24
0.22
429
437
809
859
1254
1226
0.01
0.01
355
367
712
676
1160
1151
0.12
0.02
401
373
780
768
1196
1160
0.31
0.29
419
429
830
852
1258
1241
0.13
0.08
333
387
768
775
1177
1182
0.25
0.26
443
409
820
811
1217
1214
0.24
0.34
405
403
836
820
1208
1222
0.08
0.12
386
371
782
772
1146
1156
0.32
0.18
398
413
830
826
1264
1244
Note: Ammonia results in mg/L
************************************************************
11-505
-------
************************************************************
Table 11
SIC 4952 Sewerage
Combined TKN Data
Technique
FIA
ISE
FIA
ISE
FIA
ISE
FIA
ISE
Mean
0.14
0.16
8.0
8.1
16.2
16.2
24.1
24.5
SIC 2621 Paper Mills
FIA
ISE
FIA
ISE
FIA
ISE
FIA
ISE
0.14
0.13
8.0
8.0
16.2
16.1
24.0
23.8
SIC 2869 Industrial Chemicals
FIA
ISE
FIA
ISE
FIA
ISE
FIA
ISE
0.18
0.17
8.1
8.0
16.0
16.2
24.1
23.9
SIC 1475 Phosphate Rock
FIA
ISE
FIA
ISE
FIA
ISE
FIA
ISE
0.18
0.14
7.8
8.2
16.2
16.2
23.8
24.0
SIC 2611 Pulp Mills
FIA
ISE
FIA
ISE
FIA
ISE
FIA
ISE
0.13
0.18
8.1
8.2
16.4
16.3
24.2
24.0
Confidence Intervals
-3s +3s
0.05
0.08
7.1
7.1
15.4
12.2
22.3
22.3
0.11
0.01
8.0
7.3
15.6
14.3
23.3
21.5
0.02
0.12
7.6
7.0
15.3
14.0
23.7
22.4
0.23
0.24
9.1
9.1
17.0
20.2
25.9
26.7
0.17
0.25
8.0
8.7
16.8
17.9
24.7
26.1
0.15
0.05
7.9
7.6
15.5
14.9
23.6
22.8
0.21
0.29
8.3
8.4
16.5
17.5
24.6
25.0
0.15
0.01
6.8
6.9
15.6
15.7
21.9
22.8
0.21
0.28
.8
.5
16.8
16.7
25.7
25.2
0.24
0.24
8.6
9.4
17.5
18.6
24.7
25.6
Note: TKN results in mg/L
11-506
-------
Discussion
Precision and accuracy for the FIA gas diffusion methods are
shown in Tables 4 and 5. For ammonia samples the standard
deviation did not exceed 0.05 and recovery ranged from 96.0 to
101.7 percent. For TKN the standard deviation did not exceed
0.25 and recovery ranged form 98.8 to 101.2 percent. Tables
6 and 7 show analyses of EPA unknowns for ammonia and TKN
respectively. Again, very good precision for the FIA gas
diffusion technique is exhibited.
Tables 8 and 9 show complete data sets for both the ISE
reference methods and the FIA methods from one of the sewerage
sites (SIC 4952). Four samples were tested in triplicated by
the ISE and FIA procedures for both TKN and ammonia. The
equivalency of the FIA gas diffusion technique to the ISE
reference method is obvious.
Tables 10 and 11 show the combined data for each of the two
sites for each parameter (TKN and ammonia) and each SIC code.
In almost all cases the precision of the FIA gas diffusion
technique is equivalent or better than the ISE reference
technique.
Conclusion
The FIA gas diffusion technique for testing ammonia and TKN in
wastewater has been shown to provide equivalent results to the
EPA reference ISE methods (350.3 and 351.4). At a rate of 50
samples per hour it offers an automated approach to TKN and
ammonia analysis of wastewater additionally it does not have
the drawbacks of the phenate and salicylate methods (350.1,
351.1 and 351.2) which were stated in the introduction.
References
"Methods for Chemical Analysis of Waster and Wastes". USEPA
March 1983, EPA 600/4-79-020
Requirements for Approval of Alternate Test Procedures for
Inorganic Parameters in Non-Continuous National Pollutant
Discharge Elimination System Monitoring, June 20, 1990, Nancy
S. Ulmer and Larry B. Lobring, Inorganic Chemistry Branch,
Chemistry Research Division, USEPA Office of Research and
Development, Environmental Monitoring Systems Laboratory,
Cincinnati, Ohio 45268
40 CFR Part 136 Appendix B page 510 - 512
Environmental Science & Technology 1981,12,1427, John Glaser,
Denis Foerst, Gerald McKee, Stephen Quave, William Budde
11-507
-------
109
AN OBJECTIVE CRITERION FOR TERMINATING PERMEABILITY TESTS
Mark S. Meyers, P.E., Geotechnical Engineer, U.S. Army Corps of
Engineers, St. Paul District, 1421 USPO, St. Paul, MN 55101
ABSTRACT
In present day (1991) geotechnical engineering laboratories, when
permeability testing is performed, whether the testing is being
performed for research purposes or design purposes, the criteria used
to terminate the permeability tests vary from laboratory to
laboratory. While an experienced engineer may be able to determine
when equilibrium has been reached, his judgment is based solely on
his past experience with the type of soil/permeant/permeability test
combination he is using: thus it is a subjective judgment.
Subjective judgments are used in engineering every day, especially in
the field of geotechnical engineering, with its highly variable soil
types and soil conditions. Most laboratory tests used to determine
distinct properties of a given soil have an objective criterion
associated with the tests. For exanple, the Proctor test uses a
specific energy input; the plastic limit uses a 1/8-inch thread; the
liquid limit uses the number of blows to close a given width groove
along a 1/2-inch length of the groove; and shear strength tests use a
given percent strain depending on the use and type of soil.
In recognition of the uncertainties and inadequacy of a totally
subjective test termination criterion, this paper will investigate a
new and more generally applicable objective criterion for terminating
a permeability test. This approach is applicable to the soil types
commonly tested in a geotechnical or materials testing laboratory.
The termination criteria developed are not intended to replace the
judgment of the engineer or researcher. The termination criteria are
tools to be used as an objective confirmation of the judgment of the
engineer or researcher in determining that a test has reached
equilibrium. In no case should judgment be overridden if the
engineer or researcher feels the test has reached equilibrium.
Background
The use of relatively impervious compacted clays and grouted sands
for such purposes as liners in hazardous and toxic waste landfills,
as cores in earth and earth-rock dams, and as cutoffs for dams
requires a thorough understanding of the capability of the soil to
satisfactorily perform its intended function. The major indicator of
the ease with which water is able to travel through a soil is
referred to as the permeability of the soil.
Permeability and Hydraulic Conductivity
The terms permeability and hydraulic conductivity are often
interchanged in the field of geotechnical engineering. From the
11-508
-------
fluid flow aspect, this is only correct if the fluids are held at 20
C. The fundamental difference is that to calculate intrinsic
permeability, the temperature (and thus viscosity) of the permeant is
taken into account. The use of the term "permeability11 throughout
this paper is to be taken as hydraulic conductivity, as no effort is
usually taken in soil mechanics laboratories to measure fluid
temperature.
Definition of Permeability
Permeability can be defined as the discharge velocity through a unit
area of soil under a hydraulic gradient of unity (Cedergren, 1989).
It is more commonly known as the coefficient 'k1 in Darcy's law for
laminar flow in a soil media. Darcy's law can be stated by the
equation
Q = kiAt Eq. 1
where Q is the quantity of seepage flowing through a cross section of
soil having an area A normal to the direction of flow, under a
gradient i, during a period of time t. If the terms in Equation 1
are rearranged, Equation 2 is obtained, which is the basis for the
experimental determinations of permeability that measure the amount
of seepage over a period of time under a given gradient.
k = Q/iAt Eq. 2
The coefficient of permeability has units of velocity and is usually
expressed in centimeters per second, cm/s, for soils having a low
coefficient of permeability. Other commonly used units include
ft/day and ft/yr.
The coefficient of permeability is usually assumed to be constant for
a given soil type. However, it can vary widely for a given soil type
or other material depending on a number of factors, as discussed in
detail by Taylor (1948), Daniel (1985), Bodocsi (1988), Bowers
(1988), Carson (1988), Cedergren (1989), and Conrad (1991).
Laboratory Methods for Determining Permeability
The two most commonly used methods for determining the coefficient of
permeability of a soil sample in a laboratory are the constant head
permeability test and the falling-head permeability test, utilizing
either a rigid wall or a flexible wall permeamster. A discussion on
the constant head and falling-head tests can be found in Cedergren
(1989). Daniel, et al., (1985), Carpenter (1986), and Evans (1986)
discuss permeameters in detail.
Terminating A Permeability Test
During a permeability test the measured permeability of a soil sample
often undergoes a prolonged period of transitional behavior before an
equilibrium value of permeability is reached. The decision as to
when an equilibrium value has been reached is not straightforward.
In general, no concensus criteria exist for terminating the
11-509
-------
permeability test procedure.
Die existence of a standardized objective criteria would allow for
more reliable comparisons of permeability between independent
research laboratories and commercial soil testing laboratories, would
aid in the elimination of inconsistencies in permeability test
results, and would eliminate unnecessary testing time (Pierce and
Witter, 1986).
Review of the Literature
A review of the literature has revealed several termination criteria
in use at the present time, with subjective judgment being the most
common method used.
Subjective Judgment
The method of subjective judgment can be described as a method in
which a subjective decision is made regarding whether the
permeability of a test sample has reached equilibrium by examining a
plot of permeability vs. time for the test. A horizontal plot is
sometimes used as an indicator of equilibrium. However, the soil
being tested may result in a permeability on the order of say 10~7
cm/s. A horizontal plot on this scale may be interpreted as
equilibrium, even when the slope has some small deviation from the
horizontal. At this order of magnitude of permeability, a very
slight deviation in the permeability plot may in fact be an increase
or decrease of several percent. In other words, equilibrium may not
have been reached. Conversely, if the soil type were such that a
value of permeability on the order of 10~3 cm/s resulted, a large
deviation from the horizontal may in fact represent only a small
percent increase or decrease in permeability. In the latter
scenario, equilibrium may have been reached and the engineer may not
realize this due to the visual appearance of the plot.
Die previous paragraph used percent increase or decrease in
permeability as a measure of comparison of the data points from
reading to reading during a permeability test to determine if
equilibrium has been reached. This is similar to the method used by
MoCandless (1988). In this termination criterion, if the value of
permeability does not vary by more than a predetermined percent for a
set number of readings, it is judged that equilibrium has been
reached. The difficulties inherent in this method are how to
determine an acceptable percent change in permeability over the
course of several readings and how to arrive at the number of
readings over which the termination analysis is to be made. For
soils with a low value of permeability, a relatively small percent
change might be selected. The question still remains: "How small of
a percent change is acceptable?11 Even when combining the percent
change over several readings with a visual examination of a plot of
the permeability test data, questions arise as to whether equilibrium
has been reached. Conversely, for a soil with a higher permeability
value, the question becomes: "How large of a percent change is
acceptable?11
11-510
-------
Other Methods Used to Terminate a Permeability Test
Pierce and Witter (1986) use a cxaribined criterion, specifying that at
least one pore volume of permeant must have passed through the sample
and the slope of a plot of log permeability vs. number of cumulative
pore volumes cannot be shown, by the use of a regression analysis, to
differ significantly from zero. Although this criterion is one of
the most promising termination methods to be developed, the use of
linear regression for determining the slope of the plot has several
statistical shortcomings.
Bryant and Bodocsi (1986) suggest the use of an adaptation of Mann's
test for monotone trend and also discuss the use of a Bayesian
analysis. Other termination criteria, used individually or in
combination with other criteria, include: a predetermined number of
pore volumes passing through the sample; a change in permeability by
several orders of magnitude; a predetermined concentration of the
effluent (if a chemical permeant is used); a plot of log permeability
vs. some measure of time becoming horizontal; a value of k greater
than 10~7 cm/s being reached (for a soil to be used as an approved
EPA cover or liner); or the passing of at least two pore volumes of
permeant through a soil sample in conjunction with a horizontal plot
of log permeability vs. some measure of time. Although several of
the criteria used show promise and attempt to overcome some of the
shortcomings of subjective judgment, these criteria do in fact also
exhibit some disadvantages.
The most promising termination criterion appears to be that suggested
by Bryant and Bodocsi (1986). This source has developed the
groundwork for a statistically based termination criterion which
overcomes the statistical disadvantages of the method developed by
Pierce and Witter. The method developed by Bryant and Bodocsi uses
Mann's test for monotone trend to determine if equilibrium has been
reached within predetermined statistical levels of significance and
bounds.
Application of Mann's Test For Monotone Trend
Advantages Over Regression Analysis
Mann's statistic (Bryant and Bodocsi, 1986) is designed to be
sensitive to any increasing trend, and by altering the method
slightly, the statistic is also sensitive to any decreasing trend.
Conversely, linear regression may be relatively insensitive in a case
where permeability is increasing at a decreasing rate, such as
happens when the permeability test is approaching equilibrium.
Mann's statistic is less sensitive to occasional unusual
observations, which occur in most permeability tests. These
occasional unusual observations might unduly affect a slope cxanputed
by least squares regression analysis such as in Pierce and Witter's
method. Finally, Mann's statistic does not require the assumption of
normally distributed within test errors. Psegression analysis
requires the latter (Bodocsi and Bryant, 1986).
11-511
-------
General Example
In order to apply Mann's test for monotone trend to a set of
permeability test data, Bryant and Bodocsi have heuristically adapted
it for use in a sequential manner. For this general example, the
procedure will be described as if a permeability test were being run
for a soil in which the test results would indicate a general
increase in permeability with time as the test approaches a final
equilibrium value. The 4 steps involved in the procedure are
discussed below.
1. The permeability test is permitted to run over a preliminary
period of time to allow at least two pore volumes of flow to pass
through the test sample.
2. Following the preliminary testing period, permeability values
are measured at n equally spaced points in time (t = 1, 2, ..., n).
Mann's test for monotone trend, a nonparametric statistical test,
allows for the use of data which is not normally distributed
(contains less than approximately 30 data points). The use of too
few data points provides insignificant results. Bryant (1986-1987)
recommends the use of 15 data points at the start of the procedure.
Equally spaced test readings should be used. This is a generally
accepted practice in most laboratories, with readings taken at
approximately the same time every day. As long as the time between
test readings does not vary drastically, the use of approximately
equal time intervals should be adequate. For test samples exhibiting
a permeability less than 10~10 cm/s, the pore volumes of flow
permeating a test sample during a reading interval may be difficult
to determine, depending on the experimental apparatus being used and
the environment in which the apparatus is being used. Carson (1988)
indicates that a value of permeability of 1CT12 cm/s may be the
lowest value of permeability which can be obtained with a
conventional permeability testing apparatus. The use of a constant
flow permeameter may eliminate this lower bound. Hie permeability
values are converted to Iog10 permeabilities, denoted as y^, in this
step.
3. Mann's test is used to test a null hypothesis of no trend
(i.e., the plot of yt vs. time is horizontal and equilibrium has been
reached) against the alternative hypothesis of an increasing trend in
(i.e., the value of y^ continues to increase with time) at a
specified level of significance aj_. If the null hypothesis of no
trend is rejected, then the oldest test observation is deleted from
the data set and a new observation is made. Step 3 is repeated until
the null hypothesis of no trend is accepted. Step 4 is then
conducted.
ax represents the probability of a Type I error occurring. To guard
against a Type I error, a^ is chosen to be small. The actual value
of a-ฑ is selected to be 0.01, approximately representing a 99%
confidence level that a Type I error will not occur.
11-512
-------
After the first n data points for analysis are selected, Mann's test
statistic is confuted by calculating the tied ranks and the bivariate
ranks for the 2 x n matrix X representing the row x column matrix of
the first n data points of time and permeability- Ihe matrix X is
xl X2 X3
Yn
This results in a 3 x n matrix F. Row 1 of F contains the tied ranks
R!J of the first row of X; row 2 of F contains the tied ranks R2-j of
the second row of X; row 3 of F contains the bivariate ranks Q-H of
X. -1
The tied rank Ry of an element X^ of a vector is defined as
Rij = 0.5 + u(Xi - Xj) Eq. 3
where
1, if t < 0
u(t) = 0, if t = 0
-1, if t > 0
The bivariate rank Q^j of a pair of elements (X^Y^) is defined as
Qij = 0.75 + U(Xi - Xj)u(Yi - Yj) Eq. 4
where u(t) is defined as in Equation 6.
Kendall's Tau for Step 3, T3, is
T3 = (4 F3j - n2 -3n) / (n(n-l)) Eq. 5
where F3j is the sum of the elements of row 3 of F.
Mann's statistic for Step 3 is found by dividing T3 by the variance
of Kendall's T, ST, which is
ST = 2(2 x n + 5) / (9 x n x (n - 1)) Eq. 6
Mann's statistic for Step 3, denoted by Z3/ is now
Z3 = T^STJI Eq. 7
Zo is now compared to the critical value of Z, Zc, for a level of
significance aฑ. If Z3 is greater than Zc, the null hypothesis is
rejected. The next step is to add a new data point after the next
reading and drop the oldest data point.
4. Step 4 requires that the experimenter specify an upper bound
11-513
-------
on the slope of y^ vs. time plot from horizontal at the termination
of the permeability test. This bound is referred to as B*, which is
greater than zero for a general increase in permeability with time.
In Step 4, the data is adjusted by cxsnputing a value of yt*, which is
Yt* - Yt - B*t Eg- 8
Mann's test is now performed on the adjusted test data, as in Step 3.
If the null hypothesis of no trend for the adjusted data is accepted
at a level of significance a2 against an alternative hypothesis of a
downward trend in the adjusted upper bound data, then the
permeability test is terminated: equilibrium has been reached. If
the null hypothesis is rejected, an additional observation is made,
but the oldest observation in the data set is not deleted. The
sample size is thereby increased by one data point. Return to Step 3
and repeat the procedure until the null hypothesis is accepted in
Step 4.
For acceptance of the null hypothesis for Step 4, Z4 must be less
than Zc at a level if significance a2 When this occurs, the null
hypothesis for Step 4 is accepted. The permeability test can now be
terminated and the permeability value for the sample is that
determined for the terminating observation.
a2 represents the probability of a Type II error. To guard against a
Type II error, a2 ^ chosen to be small. The actual value of a2 is
selected to be 0.05, a level of significance commonly used in
engineering statistical applications (Brubaker and McGuen, 1990) .
B* is a measure of the degree of trend the experimenter judges to be
practically, as opposed to statistically, significant. B represents
an upper bound (for tests exhibiting generally increasing
permeabilities) of the slope of the permeability plot, below which
the experimenter wants the final slope of the permeability plot to
be, in order to be considered for hypothesis testing in Step 4. B*
cannot be selected statistically per se; it can however be selected
practically, using the results of past tests and appropriate levels
of significance aฑ and a2. A goal of this research is to select
practical values of B* to use in Step 4. B* is proposed to be
dependent on soil type.
A typical statistical hypothesis test would terminate testing
immediately upon acceptance of the null hypothesis in Step 3.
Acceptance of the null hypothesis in Step 3 does not necessarily
provide statistically strong evidence that a trend does not exist.
It only implies that such a hypothesis can be maintained.
Step 4 arHซ a check to determine whether the procedure is sensitive
enough to detect trends of a meaningful magnitude. With relatively
noisy data, as is the case for permeability tests, the insignificance
of a hypothesis test for trend does not provide a reasonable
termination criterion. The insignificance of the test must be
combined with a mechanism which ensures achievement of an adequate
11-514
-------
sensitivity against trends of a meaningful magnitude. To guard
against a Type II error, the selection of B* is critical. Step 4
also increases the sample size to increase the sensitivity of future
tests.
In summary, the procedure terminates a permeability test after two
statistical criteria have been met:
1. A hypothesis that the slope of a plot of yt vs. time is equal
to zero can be maintained, and
2. A hypothesis that the slope of a plot of y^ vs. time is equal
to or greater than B* (for tests exhibiting generally increasing
permeabilities) or is equal to or less than B* (for tests exhibiting
generally decreasing permeabilities) is almost certainly false.
The Permeability Plot
A typical permeability plot consists of the dependent variable,
permeability, plotted on the ordinate scale, usually as Iog10' vs.
the independent variable, time, plotted on the abscissa. The measure
of time should be such that an equal interval of time is obtained for
each data point. The two most common measures of time used for
permeability tests and the associated permeability plot are raw time
on test, measured in consistent units of hours, days, etc., and
cumulative pore volumes of flow passing through the sample.
The use of cumulative pore volumes of flow was thought to be superior
to the use of raw time. At the start of a permeability test,
readings are taken at specified periods of time, which are usually
dependent on soil type and the judgment of the experimenter. As a
test progresses, an increased or decreased number of pore volumes of
flow will pass through the sample during the specified time period.
If the time interval between readings is maintained, especially for
soils with low permeabilities, reading the difference in fluid levels
in the standpipes becomes difficult, depending upon the permeability
apparatus being used. This introduces errors into the calculated
value of permeability for a reading interval. If the time interval
is modified, resulting in reading intervals which allow for
approximately similar volumes of permeant to pass through the sample
between readings, a regular interval measure of time is obtained.
Bodocsi and Bowers (1989) and Carson (1988) indicate that the use of
cumulative pore volumes of flow as a measure of time results in a
difficult analysis of graphical plots of log permeability vs. time
when the material has a low value of permeability. Permeability
plots using pore volumes of flow on the abscissa become vertical as
the permeability test progresses, indicating a decrease in
permeability and the associated reduction of volume of permeant
passing through the sample. Raw time is found to be the superior
measure of time for materials with extremely low values of
permeability. Where equipment restrictions do not apply and readings
can be taken at equally spaced time intervals, a time scale using raw
time should be used. The final determination of which scale to use
11-515
-------
on the abscissa rests with the experimenter.
Bodocsi and Bryant (1986) state empirical and practical reasons for
the use of a Iog10 scale on the ordinate. This practice is used in
this paper.
In summary, a plot of log permeability vs. some measure of time is to
be analyzed to determine if the slope of the permeability plot at
equilibrium is essentially zero, within practical and statistical
means.
Parameter Selection
Several parameters are required to perform the various steps
discussed for determining when to terminate a permeability test. The
statistical levels of significance ai and a2 and the upper or lower
bound on the slope of the permeability plot, B*, are used to guide
the experimenter as to whether or not the permeability test should be
terminated. The methodology used to select applicable parameters to
use in the algorithm is discussed.
Methodology
Bodocsi and Bryant (1986) recommend a practical selection of B* based
on the results of past permeability tests and appropriate levels of
significance. This methodology requires data from many permeability
tests and some indication of when the permeability test should have
been terminated.
Bodocsi, et al. (1986), Bowers, et al. (1988), and Carson (1988) ran
a large number of permeability tests. The test data includes data
using water as a permeant to determine the baseline permeability of a
sample and using various chemicals as permeants to determine the
affects of chemical permeants on various grouts. McCandless (1988)
ran several permeability tests to determine the acceptability of
various solidification/stabilization mixes. Eighty five data sets
were selected for analysis in this work. Most of the permeability
plots for the data had an approximate horizontal segment,
representing an apparent termination time.
A panel of five experts was assembled to determine when to terminate
a permeability test, based on the permeability plot for a given data
set. These individuals have a strong experience background in
permeability testing. They have struggled with permeability test
data in the past in trying to determine if a given permeability test
should be stopped.
Each expert reviewed the history of the permeability plot for each of
the 85 data sets, eliminating data points which were thought to be
non-representative of the history of the test (i.e., bad reading,
apparatus problems, sample deterioration, etc.). The expert then
determined when he, as an experimenter, would have terminated the
permeability test. The expert was asked to consider only the portion
11-516
-------
of the plot which would have been available during the actual test.
In other words, "future" readings were not to be considered.
Several panel members were intimately familiar with many of the data
sets. The data sets were given generic names, Data Set 1, Data Set
2, and so on, in an attempt to mask the actual identity of the data.
Those panel members familiar with the data indicated that several of
the plots looked familiar, but an effort was made to use an unbiased
judgment in determining the termination point.
The panel members were asked to indicate which method they used to
select the termination point. Some form of subjective judgment was
used by all panel members.
Final Data Sets and Ranges of Permeability
The data sets supplied to the panel members were modified to
eliminate the data points thought to be non-representative of the
test history. The data files were then set up for use in computer
runs through a FORTRAN computer program which mimics the SAS routine
used by Bryant to calculate the required statistical information.
Five ranges of permeability data were used, based on the apparent
order of magnitude of the termination points selected by the panel.
The panel agreed on the termination point in most cases. In several
instances, the data set was broken up into two separate data sets, to
reflect the selection of one termination point by two or three of the
panel members and the selection of another termination point a large
number of readings away from the former point by the remaining two or
three panel members. The ranges of permeability for which parameters
will be selected are 10~6, 10~7, 10~s, 10"9, and 10~10 cm/s.
Initial Runs To Select Parameters
Several data sets were run to test the significance of a^ and 32 on
the selected termination point. For all computer runs, B* was
allowed to vary for each combination of the levels of significance,
resulting in a total of 500 combinations of a^, 32, and B* for each
data set. The computer program was modified to print a summary
table, listing the combination of the parameters used to indicate the
first point at which the procedure would stop the permeability test.
For the data sets analyzed, B* was the more significant parameter,
resulting in a change in the selected termination point of several
data points for a change in B* of only 0.0001. Clearly, B* is the
more significant parameter, regardless of the selection of the
combination of levels of significance.
Selection of B*
With aj and a2 set at 0.01 and 0.05 respectively, the determination
of B* proceeded as follows. The computer program was modified to
print a summary table indicating the first point at which the null
hypothesis for Step 4 was accepted, along with the value of B* used
for that run. B* was varied from 0.0001 to 0.10. Each data set was
11-517
-------
evaluated, with the values of B* and permeability being tabulated for
the termination point selected by each panel member.
Several data sets gave an indication that the value of B* used in the
analysis is not significant or the values used for B* were not small
enough or large enough to be sensitive to the data set. In other
words, the test was shown to terminate at the same data point for
every value of B* used, or the test was shown to terminate prior to
or long after reaching the termination points selected by the panel.
These data sets will have to be reanalyzed using values of B less
than 0.0001 and greater than 0.10. Therefore, at the time of this
writing, the results should be considered preliminary.
Discussion
A plot of log permeability vs. log B* was generated using the Harvard
Graphics software. This plot illustrates apparent relationships of
log permeability vs. log B*, depending upon the order of magnitude of
permeability. The relationships appear to be more sensitive for the
higher orders of magnitude of permeability.
The apparent relationships between B* and Permeability were analyzed
using simple linear regression with Iog10 transformations on the data
for three orders of magnitude of permeability: 10~7; 10~9; and 10~10
cm/s. The relationships using the data transformations indicate
correlations of 0.528, 0.725, and 0.117, respectively for these
orders of magnitude of permeability. Obviously, other data
transformations will need to used to determine the most significant
relationship.
At this time, there does not appear to be a general relationship
between B* and permeability. The relationship appears to be limited
to distinct ranges of permeability. Future analyses will clarify the
extent and significance of these relationships and develop
mathematical equations to use in the termination procedure to select
an appropriate value of B*. These equations will then be programmed
into the computer routine to intrinsically select B* during the
analyses, while allowing override by the user.
Acknowledgements
The author wishes to thank the USEPA AWBERC Risk Reduction
Engineering Laboratory for their partial support of this research.
The author also wishes to thank Dr. Andrew Bodocsi Dr. Mark Bowers,
and Mr. Richard McCandless of the University of Cincinnati, Mr. David
Carson of USEPA in Cincinnati, and Dr. Earl McCullough of the
University of Wisconsin at Platteville for their assistance in
evaluating permeability plots.
11-518
-------
References
Bodocsi, A., M.T. Bowers, and R.A. Sherer, "Permeability of Grouts
Subjected to Chemicals," In: Proceedings of the Specialty
Conference on Environmental Engineering. Environmental
Engineering Division of The American Society of Civil Engineers,
June 1986, pp. 99-105.
Bodocsi, A., M.T. Bowers, and R.A. Sherer, "Reactivity of Various
Grouts to Hazardous Wastes and Leachates," Final Report, United
States Environmental Protection Agency, Contract No. 68-06-3210,
Work Assignment 13, Cincinnati, Ohio, February, 1988.
Bodocsi, A., and J.L. Bryant, "Precision and Reliability of
Laboratory Permeability Measurements," EPA/600/S2-86-097, United
States Environmental Protection Agency, Cincinnati, Ohio, 1987.
Bowers, M.T., A. Bodocsi, and D.A. Carson, "Reactivity of Various
Grouts to Hazardous Wastes and Leachates - Phase IV," Final
Report, United States Environmental Protection Agency, Contract
Nos. 68-03-3210-13 and 68-03-3379-06, Cincinnati, Ohio, January,
1988.
Brubaker, K.L., and R.H. McGuen, "Level of Significance Selection in
Engineering Analysis," Journal of Professional Issues in
Engineering. American Society of Civil Engineers, Volume 116, No.
4, October 1990, pp. 375-387.
Bryant, J.L., Private discussions with the author, Department of
Quantitative Analysis, University of Cincinnati, Cincinnati, Ohio,
1986 through 1987.
Carpenter, G.W., and R.W. Stephenson, "Permeability Testing in the
Triaxial Cell," ASTM Geotechnical Testing Journal. Volume 9,
Number 1, March, 1986, pp. 3-9.
Carson, D.A., "Hydraulic Conductivity of Modified Cement and Polymer
Based Grouted Soils When Exposed to Hazardous Chemicals", a
thesis presented to the Department of Civil and Environmental
Engineering in partial fulfillment of the requirements for the
degree of Master of Science, University of Cincinnati,
Cincinnati, Ohio, 1988.
Cedergren, H.R., Seepage. Drainage, and Flow Nets. 3rd Edition. John
Wiley & Sons, New York, New York, 1989.
Conrad, D.J., S.A. Shumborski, L.Z. Florence, and A.J. Liem,
"Assessment of the Parameters Affecting The Measurement of
Hydraulic Conductivity for Solidified/Stabilized Wastes," In:
Remedial Action. Treatment, and Disposal of Hazardous Waste.
Proceedings of the 17th Annual REEL Hazardous Waste Research
Symposium. EPA/600/9-91/002, United States Environmental
Protection Agency, Cincinnati, Ohio, April 1991, pp. 543-559.
11-519
-------
Daniel, D.E., D.C. Anderson, and S.S. Boynton, "Fixed-Wall vs.
Flexible-Wall Permeameters," Hydraulic Barriers in Soil and
Rock. ASM Special Technical Publication 874, A.I. Johnson, R.K.
Frobel, N.J. Cayalli, and C.B. Pettersson, Editors, American
Society for Testing Materials, Philadelphia, PA, June, 1985, pp.
107-126.
Evans, J.C., H.-Y. Fang, "Triaxial Equipment for Permeability Testing
with Hazardous and Toxic Permeants,' ASTM Geotechnical Testing
Journal. Volume 9, Number 3, September 1986, pp. 126-132.
Pierce, J.J., and K.A. witter, "Termination Criteria for Clay
Permeability Testing," Journal of Geotechnical Engineering.
American Society of Civil Engineers, Volume 112, No. 9, September
1986.
SAS Institute, Inc., SAS User's Guide; Basics. 1982 Edition. SAS
Institute, Inc., Gary, NC, 1982.
SAS Institute, Inc., SAS User's Guide: Statistics. 1982 Edition. SAS
Institute, Inc., Gary, NC, 1982.
Taylor, D.W., Fundamentals of Soil Mechanics. John Wiley & Sons, New
York, New York, 1948.
Notation
&1 = statistical level of significance for Step 3
a2 = statistical level of significance for Step 4
A = cross-sectional area of soil sample
B* = bound on slope of permeability plot used in Step 4
i = hydraulic gradient
k = coefficient of permeability
n = number of data points used in the test
Q = flow rate of permeant through a soil sample
Qi = bivariate rank of a pair of elements X^,Yj[
Ri = tied rank of an element X^
Sji = variance of Kendall's Tau
T = Kendall's Tau
T3 = Kendall's Tau for Step 3
T4 = Kendall's Tau for Step 4
t = time
y-j. = Iog10 permeability for a time t
yt = adjusted Iog10 permeability for use in Step 4
Z = Mann's Test Statistic
Z3 = Mann's test statistic for Step 3
Z4 = Mann's test statistic for Step 4
ZQ = critical value of Z for use in hypothesis testing
H-520
-------
J 1 Q SAMPLING AND ANALYSIS PLANS TO EVALUATE THE PERFORMANCE OF
LEAD-BASED PAINT ABATEMENT
Benjamin S. Lim. Ph.D., Randy Cramer, Ph.D., Field Studies Branch, John
Schwemberger, M.S., Design and Development Branch, U.S. Environmental
Protection Agency, Washington, D.C. 20460; Bruce Buxton, Ph.D., Steve
Rust, Ph.D., Bob Lordo, Battelle, 2101 Wilson Boulevard, Suite 800,
Arlington, Virginia 22201-3008; Gary Dewalt, Ph.D., James McHugh, CIH,
Midwest Research Institute, 425 Volker Boulevard, Kansas City, Missouri
64110.
Abstract
The U.S. Department of Housing and Urban Development conducted a lead
paint abatement demonstration at 169 houses from five metropolitan areas.
The U.S. Environmental Protection Agency plans to conduct a follow-up
study at these houses to measure levels of lead in dust and soil. Six
types of interior locations will be sampled for dust. Three types of
exterior locations will be sampled for soil. Dust and soil will be
chemically analyzed for lead. In general, soil and dust samples will be
digested by nitric acid and hydrogen peroxide, and analyzed by ICP or
graphite furnace AA. Soil samples will be sieved and dried before
digestion. Dust results will be reported as a loading (pg/square foot)
and a concentration (ptg/g). Soil results will be reported as a
concentration (pg/g).
Introduction
In response to requirements mandated by the Lead-Based Paint Poisoning
Prevention Act, as amended by Section 566 of the Housing and Community
Development Act of 1987, the U.S. Department of Housing and Urban
Development (HUD) carried out a lead paint abatement demonstration
project in FHA re-possessed housing. The demonstration was conducted in
five metropolitan areas across the country. Single family FHA houses in
these cities that were owned by the department were tested for lead-based
paint. Homes that met certain criteria were chosen for the lead paint
abatement project. This HUD project is now virtually completed.
Under an interagency memorandum of understanding, the U.S. Environmental
Protection Agency is providing technical support to HUD on lead-based
paint issues. EPA plans to conduct a follow-up study to the HUD
abatement demonstration in order to measure the levels of lead in
household dust and exterior soil in the years following abatement. The
purpose of the study is to assess the long-term efficacy of the
11-521
-------
abatement methods used in the HUD demonstration project. This
information is needed before the nation embarks on a costly abatement
program.
Study Homes
In the HUD Demonstration, 169 houses were abated for lead paint. Both
interior and exterior housing components were abated. Six different
methods of lead paint abatement were used in the project (encapsulation,
enclosure, heat gun stripping, chemical stripping, mechanical stripping,
and component replacement). The first two methods cover existing lead
paint, the last four remove it. An individual home was likely to be
treated by more than one abatement method.
In order to have enough houses for the statistical analysis in the
follow-up study, houses have been classified as either
encapsulate/enclosure houses or removal houses. Classification was made
on the basis of the square footage abated in the interior of the house
by the encapsulate/enclosure methods (encapsulation and enclosure) and
the removal methods (heat gun stripping, chemical stripping, mechanical
stripping, and component replacement). Interior abatement was chosen for
classification of houses because of an a priori assumption that interior
lead paint abatements have the most impact on interior dust levels.
Interior dust and exterior soil will be collected at each house that has
been re-sold, re-occupied, and recruited for the study. Six interior
locations will be sampled for dust: floors, window sills, window stools,
inside entryways, air ducts, and upholstered furniture/rugs/carpets.
Three exterior locations will be sampled for soil: outside entryways,
along the house foundation, and near the property line. The selection
of locations will be discussed in the next section.
Selection Of Sample Locations
For the follow-up study, two rooms in each house will be selected for
sampling. Rooms will be selected so that the predominant abatement
method used in the room matches the predominant interior abatement method
for the house. In each room, a floor section, a window sill, and a
window stool will be sampled. An air duct will be sampled in each room,
if an air duct is present. In addition, one carpet, rug, or piece of
upholstered furniture will be sampled in each room, pending availability.
Finally, the interior of two entryways will be sampled.
Table 1 summarizes the environmental sampling planned for the study,
including both regular samples (vacuum dust and soil cores) and field
quality control samples (wipe dust, field blanks, and side-by-side
11-522
-------
samples) intended to assess sampling variability and potential sample
contamination. As shown in this table, a total of 23 samples will be
collected from each house during each sampling campaign, with a grand
total of over 6,000 samples being collected in all three sampling
campaigns.
The objectives of the Abatement Performance Study include both assessing
long-term performance of abatement methods and investigating the
contribution to interior dust lead levels from other sources. The role
of each type of sample listed in Table 5 for meeting these objectives is
as follows:
Vacuum dust from floors Provides primary measure of performance
for interior abatement;
Vacuum dust from window sills Provides primary measure of
performance for interior abatement;
Vacuum dust from window stools - Provides measure of performance
for interior abatement, possible measure of performance for
exterior abatement, and possible transport of exterior soil from
outside to inside the house;
Rugs, upholstery, and air ducts Provides measure of source
contribution to interior dust lead levels;
Entryway floor Provides measure of possible transport of
exterior soil from outside to inside the house;
Soil cores Provides primary measure of performance of exterior
abatement, and measure of possible transport of exterior soil lead
into the house;
Wipe dust from floors Provides consistency check against earlier
results from HUD Demonstration and other studies;
Field blanks Provides assessment of potential sample
contamination; and
Side-by-side samples Provides assessment of sampling
variability.
1-523
-------
TABLB 1. SUMMARY OF ENVIRONMENTAL SAMPLING PLANNED
FOR THE ABATEMENT PERFORMANCE STUDY
Samples Total for Total for
Sample Type Per House One Campaign* Three Campaigns
Regular Samples
1. Vacuum dust
a. Perimeter floor 2 180 540
b. Window sill 2 180 540
c. Window stool 2 180 540
d. Rug/Upholstery 2 180 540
e. Air ducts 2 180 540
f. Entryway floor 2 180 540
2. Soil cores
a. Near foundation 2 180 540
b. Property boundary 2 180 540
c. Entryway 2 180 540
Quality Control Samples
3. Wipe dust
a. Floor 1 90 270
4. Field blanks
a. Vacuum dust 1 90 270
b. Soil cores 1 90 270
5. Side-by-aide samples
a. Vacuum dust floor 1 90 270
b. Soil cores 1 90. 270
Total samples 23 2070 6210
* As Burning an average of 90 houses sampled in each campaign (i.e., 105,
90, and 75 houses in the first, second, and third campaigns,
respectively).
11-524
-------
Interior Dust
It is anticipated that results from the Abatement Performance Study may
be compared with earlier results from the HUD Demonstration and HUD
National Survey. For this reason the sampling and analytical methods for
the Abatement Performance Study have been selected to match as closely
as possible the methods used in these earlier two studies. The sampling
and analytical methods planned for interior dust sampling in the
Abatement Performance Study are summarized in Tables 2 and 3
respectively. Some important points to note in these tables are the
following:
Sampling will be performed in two different rooms of each house
for abated houses this will provide a measure of the variability
in abatement performance within a house, while for control houses
this will provide a measure of the variability in background lead
levels within a house. Rooms in abated houses will be selected
according to the largest square footage abated and the highest
percentage abated by the predominant abatement method for the
house.
Sampling will be performed in each room separately for floors,
window sills, and window stools for abated houses, this will
provide a means to assess differences in the way an abatement
method may perform on different structural components, and for
control houses this will provide a further measure of the within-
house variability of background lead levels.
Sampling will also be performed in each room separately from one
rug or upholstered furniture piece, and one air duct; in cases
where more than one such component is available in a room, the
specific component for sampling will be randomly selected from
those available.
Vacuum sampling, rather than wipe sampling, is the primary method
planned for interior dust as noted earlier, this method allows
for measurement of lead on a concentration basis so that
comparisons among abatement methods, houses, and across time can
be made, controlling for potentially biasing effects due to
variations in the total amount of dust present; vacuum sampling
also allows rugs and upholstered furniture to be sampled.
1-525
-------
TABLE 2. SAMPLING METHODS FOR INTERIOR OUST
OTS Abatement
Performance Study
HUD
Demonstration
HUD
National Survey
Sampling device
Sampling Area
Vacuum (Oast rotary pump,
modified 37-mm mixed cellulose
ester filter cassette)
4 square feet (floors, rugs/
upholstery) Entire area
(window sills, stools)
ro
Samples Collected Total of 12 samples;
One window sill (two rooms)
One window stool (two rooms)
One perimeter floor location
(two rooms)
One front entryway floor
One back entryway floor
Two area rug/upholstered
furniture
Two air ducts
Compositing Will be determined after review
of pilot sampling results
Chubs Thick Baby Wipes
with Aloe (5-3/4x8")
One square foot (floors,
window sills, window
stools)
Three samples per abated
area;
One window sill per abated
area
One window stool per
abated area
One floor per abated area
None
Vacuum (Cast rotary pump, modified
37-mm mixed cellulose ester
filter cassette)
Four square feet (floors)
Entire area (window stools,
sills)
Total of at least 7 samples;
One floor at front (or most
heavily used) entryway
One floor in wet room
One floor in dry room
Each window stool in wet
room
Each window stool in dry
room
Each window sill in wet room
Each window sill in dry room
None
-------
TABLE 3. ANALYTICAL METHODS FOR INTERIOR DUST
OTS Abatement
Performance Study
HUD
Demonstration
HUD
National Survey
en
ฃ3
Sample
preparation
summary
Instrumental
Technique
Est. LOQ
Data reporting
QA/QC Notes:
Filter digested in HNO-,/
H202 Diluted to 25 mL
Graphite furnace atomic
absorption
15 pg/g or 0.15 /jg/sample
ftg/g and fjg/ft2
NIST Buffalo River sediment
(SRM 2704) and Estuarine
sediment (SRM 1646) used
for reference materials.
Wipe ashed at 550-600 C
for 2 hrs. Acid digested
in HNO3/H202 Diluted to 10
mL
Flame atomic absorption
^/g/ sample
No reference material used
Used side-by-side sampling
for duplicates
Filter digested in HNO,/
H2O2 Diluted to 25 mL
Graphite furnace atomic
absorption
0.15 /jg/sample
/jg/sample
In-house spiked soil used
for reference material
-------
For the exterior, two aides of the house will be selected at random.
Foundation samples will be taken one foot from the foundation of the
house on the two sides of the house selected. Foundation samples will
consist of five equally spaced samples along the side of the house. The
five equally spaced samples will be composited into a single soil sample.
On the same two sides of the house selected for foundation samples,
samples will be collected near the property boundary. Two randomized
positions along the property boundary will be chosen. Boundary samples
will consist of a composite of three soil samples collected at the
vertices of an equilateral triangle with a side length of 20 inches.
Finally, soil samples will be collected outside the same entryways for
which interior dust samples were collected. Entryway soil samples will
consist of three soil samples collected at the vertices of an equilateral
triangle with side length of 20 inches.
Exterior Soil
The HUD Demonstration evaluated the abatement of both interior and
exterior painted surfaces, and in fact, for many houses exterior
abatement was the most significant activity performed. Furthermore, the
same abatement method might be expected to perform quite differently on
interior and exterior surfaces. Therefore, the Abatement Performance
Study will evaluate both interior and exterior abatement.
If an abatement method fails to completely control an exterior lead-based
paint hazard, then the resulting effect would most likely be seen as an
increase in soil lead concentrations close to the foundation of the
house. Therefore, exterior soil sampling will provide the primary means
for assessing the performance of exterior abatement. In this assessment,
lead concentrations measured in soil samples taken close to the
foundation will be compared with those measured in samples taken at the
property boundary which are as far as possible from the foundation, and
therefore, primarily affected by only background sources of lead, rather
than lead-based paint abatement. As with interior dust sampling, results
of soil sampling from the Abatement Performance Study will also be
compared with earlier results from the HUD Demonstration and National
Survey. Therefore, the sampling and analytical methods for soil in the
Abatement Performance Study have been selected to closely correspond to
those used in these earlier two studies. Those methods are summarized
in Tables 4 and 5, where the following important points should be noted:
Soil samples will be collected both at the foundation of each house
and at the property boundary for abated houses this will provide
a measure of both soil potentially contaminated by abatement (i.e.,
at the foundation) and soil contaminated mostly by background
sources (i.e., at the property boundary); for control houses this
will provide a measure of the spatial variations in background soil
lead levels.
8
11-528
-------
TABLE 4. SAMPLING METHODS FOR EXTERIOR SOIL
OTS Abatement
Performance Study
HUD
Demonstration
HUD
National Survey
Sampling device
1-inch ID soil recovery probe,
top 0.5 inch of soil is taken.
_ Samples Collected Total of 6 samplest
en
Compositing
One taken 1 ft from foundation
(two opposite sides of unit)
One at property boundary
(two opposite sides of unit)
One at front entryway
One at back entryway
Foundation samples will be a
composite of 5 uniformly-spaced
cores. Boundary and entryway
samples will be a composite of
3 cores spaced 20 inches apart.
0.75 inch ID tube (0.5
square inch surface area),
top 0.5 inch of soil is
taken.
Total of 4 samples (both
SPR 24x1-1/8 soil probe, top 2-3
cm of soil is taken.
Total of 3 samples;
before and after abatement)t One taken 1 ft from foundation
One taken 1 ft from
foundation (all 4 sides
of the unit)
All samples are a composite
of 5 uniformly-spaced cores
along the length of the
wall.
(where exterior XRF occurred,
or if no XRF, then at a wall
selected randomly)
One taken halfway between XRF-
sampled wall and property boundary
One at entryway
All samples are a composite of 3
cores spaced 20 inches apart.
-------
TABLE 5. ANALYTICAL METHODS FOR EXTERIOR SOIL
OTS Abatement
Performance Study
HUD
Demonstration
HUD
National Survey
Sample
preparation
summary
Instrumental
Technique
EST. LOQ
Data reporting
QA/QC Notes:
Sample drying and homogeniza-
tion Digest 0.5 g using
HNO3H2O2 Dilute to 50 mL
Inductively coupled plasma
atomic emission spectrometry
pg/gram dry wt
NIST Buffalo River sediment
(SRM 2704) and Eatuarine
sediment (SRW 1646) used for
reference materials
Oven dry, sieve
Oven dry at 105 C for 24
hrs.
1 gram digested in HNO?
Dilute to 100 mL
Flame atomic absorption
6 pg/g
pg/gram dry wt
Reference material not
specified
0.5 gram digested in HNOi/
H2O2 Diluted to 50 mL
Inductively coupled plasma
atomic emission spectrometry
pg/gram
In-house spiked soil used
for reference material.
10
-------
Samples will be collected from two opposite sides of the house
for abated houses this will provide a measure of the variability
in abatement performance, while for control houses this will
provide another measure of the spatial variations in background
soil lead levels. In selecting sides of the house for sampling,
priority will be given to sides including the largest square
footage abated and the highest percentage abated by the predominant
abatement method for the house.
Samples will be collected immediately outside the front and rear
entryways for both abated and control houses this will provide
a means for assessing possible transport of exterior lead into the
house.
Summary
Sampling and analysis methods described in this manuscript are currently
being tested in a pilot study. There may be changes to the methods
described after the pilot is completed.
11
11-531
-------
"Ml FURTHER EVALUATION OF THE CAGE MODIFICATION TO THE TCLP
Paul White, SAIC, 8400 Westpark Drive, Mclean, VA 22102
ABSTRACT
The purpose of WA 17 was to further evaluate the proposed cage modification to
Method 1311. The new proposal incorporates the use of a stainless steel cage for the
testing of solidified/stabilized waste without prior particle size reduction. Central to this
issue is whether the cage approximates the level of stress that a stabilized waste would
undergo if disposed in a landfill.
To evaluate the utility of the cage, wastes were collected from several waste
generators, including electroplating operations, secondary lead smelters, and creosote wood
preservers. The wastes were stabilized by addition of cement. The stabilized wastes were
tested for compaction strength to provide a standard by which to assess the level of stress
imparted by the cage. It is though that low strength stabilized wastes should be significantly
degraded while high strength formulations should be less degraded. Extractions were
conducted with the cage, and hard plastic bottles to directly compare the level of stress
imparted by each extraction method.
It was found that the bottle and cage were equivalent with respect to the amount of
degradation observed for low strength stabilized wastes i.e. all were degraded completely.
High strength wastes showed that the cage was more aggressive than the bottle and that
waste stability in the TCLP extraction fluid was equally important in predicting the degree
of degradation of the waste. One of the stabilized wastes did show a correlation between
high compaction strength and survivability in the cage. As the compaction strength of the
formulations decreased, the amount of sample degradation increased.
In general, the proposed cage modification would provide a aggressive challenge to the
stability of a stabilized waste without prior size reduction.
11-532
-------
112 COMPARATIVE STUDY OF EPA TCLP AND CALIFORNIA W.E.T
FOR METALS IN DIFFERENT MATRICES.
G.S.Sivia. M.S.Iskander and J.T. Coons,
Hazardous Materials Laboratory. California Department Of Health Services,
2151 Berkeley Way. Berkeley, California 94704.
ABSTRACT;
EPA implemented the Toxicity Characteristic Leaching Procedure (TCLP) to simulate leaching of hazardous
waste and to identify additional characteristics of waste, primarily organic constituents in RCRA waste in
September, 1990. But for hazardous waste characterization particularly for metals, California has a Waste
Extraction Test (WET) which covers seventeen metals including eight regulated by EPA under TCLP.
Comparative studies were carried out to evaluate the TCLP procedure against California W.E.T for leaching
of EPA regulated metals (Ag, As, Ba, Cd, Cr, Pb, Hg, Se). Originally, EPA incinerator ash sample (EPA
Interlaboratory Study XX sample) was used for comparison, and it was found that California WET gave higher
results than the TCLP for the regulated metals extracted.
Later some of the actual hazardous waste soil, sludge and liquid samples received at HML (Hazardous
Materials Laboratory), Berkeley, Ca. from different contaminated sites around California were extracted using
both TCLP and Calif. WET procedures and were analyzed by ICPAES (Inductively Coupled Plasma Atomic
Emission Spectroscopy).
California WET gave consistently higher results for all eight EPA regulated metals in all the matrices tested.
Also California WET seems to be a more aggressive test than TCLP, even if the TCLP results are doubled to
account for the different extraction ratio which is 1-2O for TCLP and 1-10 for California WET. There was no
apparent relationship between soluble metals in TCLP extract as a percentage of W.E.T or as a percentage
of total metals in different samples.
California Waste Extraction Test is also applicable to other metals (Be, Co, Cu, Mo, Ni, Sb, Tl, V, Zn) which
are regulated under Title 22 , California Administrative Code, but not under EPA RCRA regulations. When
results for these metals by the two extraction protocols were compared, Calif. WET came out superior than
TCLP.
In addition, California WET also has advantages over EPA-TCLP procedure because it is simpler in that it does
not require sample digestion after extraction, no pH measurement before extraction and no pre-selection of
extraction solution.
MTROOUCTION:
The most significant risk from the hazardous waste results from the leaching of toxic constituents into
groundwater. The EPA designed Extraction Procedure Toxicity Test (EP Tox.) to simulate the leaching of solid
hazardous waste co-disposed with municipal waste in a sanitary landfill and to asses the potential impact
of the leachate on ground water contamination. But since EP Toxicity test has a limited applicability due to
short list of constituents , EPA proposed a "second generation" extraction procedure TCLP as a replacement
to address the shortcomings of EP Toxicity. The TCLP protocol includes the expanded list of regulated
contaminants from the fourteen listed in the EP Toxicity protocol to a total of fifty-two which includes eight
metals. California has a equivalent extraction test (WET) for soluble metals under its code of regulation "Title
22" which includes seventeen metals. There are many contrasts among these three procedures, which are
listed in the figure 2. Maximum contaminants levels for seventeen metals are listed in figure 3.
This study was designed to compare the extraction efficiencies of telp and Calif, wet test. This comparison
was accomplished in two ways. In the first, the metal extraction effectiveness of the two extraction
procedures was evaluated on a EPA incinerator ash for some metallic contaminants (listed in Table 3) and
I-533
-------
SCHIM additional metal*. The second phase included evaluating the efficiency of the extraction on actual soil
and sludge sample* received at HML. Berkeley, Ca. The effect of sample digestion was also investigated on
extracts from both methods using EPA 3010 digestion procedure (as recommended in tclp protocol).
METHODS:
SAMPLE PRHปARATION:
Incinerator ash sample was very homogeneous, so no sample preparation was done. HML soil samples were
grinded and passed through 10 mesh sieve to get a homogeneous sample before extraction. The liquid
samples (containing <0.5% nonfilterable solids) were filtered through appropriate filter papers ( 0.45 urn for
wet & 0.6 urn for tclp) while sludge samples were filtered through appropriate filters; filtrate was saved and
solid part was extracted with proper extraction fluid, filtered and combined with original filtrate before
analysis.
TCLP Procedure :
Incinerator ash sample received as a part of EPA Interiaboratory study xx was used in the preliminary
investigation. TCLP protocol was followed as outlined in tdp flowchart (Figure 1). In order to select the
proper extraction fluid for tdp . sample pH was determined . The sample pH in reagent water was 5.69, but
after adding 3.5 ml of 1.0 N Hd to the sample solution, heating to 50 C for 10 minutes and cooling, the pH
came down to < 5.0 (pH 4.70). so extraction fluid #1 was used. Extraction fluid #1 is a acetate buffer
which is prepared by adding 5.7 ml of glacial acetic add to about 900 ml deionized water, then adding 64.3
ml of 1.0 N NaoH. and diluting to a volume of 1 liter. The pH of fluid was 4.93.
25 gm of ash sample was extracted in triplicate for 18 hrs. over a rotary extractor at 30 r.p.m. with 500ml
of extraction fluid. The sample* were filtered through 0.6 um glass microfiber filters (14.2 cm) under pressure
with nitrogen (in Millipore Hazardous Waste Filtration System OM 100). The filtrate for each replicate was
divided in to two portions; one portion was analyzed as such , while the other portion was digested using
EPA 3010 digestion procedure (SW 846. 3rd edition ,1986). Both the extracts were analyzed for soluble
metals with ICPAES (Inductively Coupled Plasma Atomic Emission Spectroscopy).
25-50 gms of HML oil and sludge samples, and 10O ml of liquid sample were used for TCLP extraction. The
above TCLP protocol was followed.
;(WCTJ:
California wet does not need pH test of the sample before extraction. Wet extraction fluid (dtrate buffer) was
prepared by adding 42.0 gm of monohydrate citric add in 950 ml of deionized water and then adjusting the
pH to 5.0 by adding 50 % NaoH and making the volume to 1 liter. Prior to use in the extraction step, buffer
was deoxygenated by purging with nitrogen gas. 25-50 gms of incinerator ash and HML samples were
extracted with 250-500 ml of citrate buffer on a regular mechanical shaker for 48 hrs. After extraction the
fluid was filtered through 0.45 um filter paper . The filtrate was divided into two portions; one part was
analyzed as such while other part was digested using EPA 3010 digestion procedure. Both these extracts
were analyzed for soluble metals by ICPAES.
TOTAL META' ft
Incinerator ash and all other samples were also digested using EPA 3050 digestion procedure (SW 846. 3rd
edition, 1986) and analyzed for total metals for comparison purposes. Yttrium was used as internal standard
in aB samples and standards to compensate for viscosity differences in different matrices before analysis.
Icp was used for analysis of all the samples except Hg.
HP ANALYSIS;
Hg analysis on soil and sludge (total Hg) and undigested portions of TCLP and WET extracts ( soluble Hg)
was done by Cold Vapor technique ( EPA 7470. SW 846. 3rd edition. 1986 ).
QUALITY CONTROL:
Method blank, method spike , matrix spike duplicates, qc check sample, and EPA reference standards (ICAP-
19, ICAP-7. WP-287) were analyzed with each set of samples as a quality control check on analysis. Percent
red, rpd, and % recovery were calculated as a means of determining the precision and accuracy of the data.
11-534
-------
RESULTS:
The results of Incinerator ash sample for total, and soluble metals in wet and tclp extracts are summarized
in figure 4. The mean and % rsd of three replicates are listed . The concentration of total metals in ash
sample varies widely from less than detection limit for selenium to 55000 mg/kg for Calcium. Although there
is 10 mg/kg of Ag in ash sample, but none is extracted in tclp or wet extract. Arsenic and Chromium are
present at about the same level (69.4 and 62.6 mg/kg respectively) as total metals in the ash sample, but
As is not extracted at all and Cr is only 0.2% in TCLP extract, as compared to 5.2% As and 1.3% Cr in wet
extract as a %age of total metal content. In general, soluble metals extracted by TCLP as a percentage of
total metals varied from non-detected (As. Se, Ag) to maximum of 4.8% for Cd ( Figure.7a), where as wet
extracted quite a higher concentration of As (5.2%), Pb (6.5%) and Cd (7.3%). This data indicates that WET
extracted consistently higher amounts of soluble metals as compared to tclp for all the metals analyzed in
this experiment. Of particular interest are As and Pb which were extracted in significantly higher amounts
by wet method as compared to tclp. Using soil and sludge matrices besides Incinerator ash gave similar
results. Results of HMUM542 (sludge) and HML# 1543 (soil) samples are discussed here. Both these
samples came from Empire Mine , Grass Valley. Ca. and were high in total As and Pb, and also had Ba and
Cd above STLC ( soluble threshold limit concentration). The pH of the samples was between 5-6. Both these
samples were digested for total metals as well as extracted by tclp and wet methods for soluble metals. The
data has been graphically presented (Figures 8,9). In order to fit the data to scale, bars for total metal
represents only 10% of total metals concentration (Figure 8) while 5% for As and Pb . and 1% for Ba and
Cd (Figure 9). Interpretation of this data showed the same trend that wet test gave quite higher results than
tclp for both soil and sludge samples for As, Ba, Cd, and Pb. Total Ag, Cr, and Se were less than detection
limit and consequently none was extracted in tclp and wet extracts. Both these samples were high in total
as well as soluble Hg than threshold limits (figure 3). Total, wet, and tclp concentrations of Hg in sludge
sample was 32.9, 2.65, and 0.53 mg/kg while in soil 95.0, 3.21, and 0.39 mg/kg respectively (Figure 8a).
In addition, studies were carried out to find out if digestion after extraction makes any differences in the
recovery of metals. A set of soil samples received at HML from a contaminated shipyard in San Francisco,
and another set of sludge samples from metal recycler. Short Scrap Iron and Metal Inc. Redding were used.
These samples were analyzed for total and soluble metals. A portion of the extracts from both the TCLP and
wet extracts was taken and digested with EPA 3010 ( as recommended in tclp protocol) and other half
portion was analyzed as such. The results were compared of both digested and undigested extracts for both
tdp and wet methods (Figure 10,111.The data indicate that there is no differences in digested and undigested
extracts for recovery of any of these metals tested. Although digestion step for the extract is only a part of
tdp protocol, but was tried in the wet method too for these samples . Out of four soil and two sludge
samples tested in this experiment , none of the metals showed a significant difference in digested and
undigested recoveries.
TCLP results when multiplied with a factor of two in order to take into account the dilution factor of(1-20)
compared to WET (which is 1-10) in a soil sample from Orange County Steel and Salvage, Anaheim, Ca.
were still lower than wet (FigureTb). Similarly some of the soil and sludge samples received at HML from
Empire Mine, Grass Valley, Ca. were high in total Pb and As (Figure 8,9), but soluble Pb (wet) extracted in
sludge sample was about 8.3 % of total where as in soil it was about 5% , although total Pb was about
6000 mg/kg in both the sample matrices. Higher cone, of soluble Pb (wet) in sludge sample may be due to
the fact that part of sludge sample was liquid which had more soluble Pb and was recombined with the
extract after the solid portion was extracted with appropriate extraction fluid. Arsenic shows the same
pattern, since total As ranged from 1630-1750 mg/kg in sludge and soil, respectively (Figure 8a), but soluble
As (wet) in sludge was 1.2% and 0.7% in soil as percentage of total As. In essence the data shows that
both As and Pb as soluble metals ( by wet method) are present in higher concentration in sludge sample than
in soil, but percentage wise Pb is extracted more than As by the same method in the same matrices.
Although same relationship is true in tdp extraction procedure for sludge and soil i.e Pb is extracted more
by tdp in sludge than soil, but As is higher in soil sample than sludge though total As amount is the same
in both matrices. There seems to be no consistent relationship of wet and tclp soluble metals extracted as
percentage of total metals. The reasons for higher recovery in wet than tdp seems to be associated with
type of buffer and extraction time. Citrate buffer used in wet is more aggressive than acetate buffer and also
longer extraction time of 48 hrs in wet than 18 hrs. in tclp might be a factor in solubilizing more metals.
For liquids (containing <0.5 % nonfilterable solids), a HML sample (F1783) which was high in silver did not
show any significant difference between tdp and wet soluble silver (Figure 8a), since the sample was not
subjected to extraction and was only filtered through specified filter papers in each method.
QUALITY CONTROL:
A comprehensive QC guidelines were followed to validate the data for predsion and accuracy (figure 5,6).
All the QC samples show very good precision and accuracy. Method blank results donot show any
contamination
11-535
-------
duplicate matrix spikes where each matrix was spiked before digestion or extraction. In method spikes,
reagents or extraction fluids were spiked before extraction or digestion. Method spikes recoveries for tclp
and wet varies from 80-95 % except for Ag . Relative percent difference (RPD) for duplicate method spikes
in both extracts range from 0.2-13 %. except for Ag in tdp which is 24.8 (figure 5). For total metal analysis,
RPD for matrix spike duplicates on incinerator ash sample is under 13 % except for Ag and Ba, and %
recoveries of matrix spikes are in high eighties. Low Ag recovery may be due to addition of Hd in method
EPA-3050. Same may be the reason for bad precision (high rpd) in matrix spike duplicates (total metals) for
Ag. But these deviations do not affect our results because total Ag was below detection limit in all soil and
sludge samples and was present at such a low level in ash sample that it was not extracted in tdp and wet
extracts. Low matrix spike recovery for Ba in total metal determination may be due to the precipitation of
Ba as Baso4 in the ash sample, mhouse HML soil qc check sample with known values was also digested and
analyzed and % recoveries varied from 98-114%. To check the accuracy of instrumental analysis of the
samples, EPA reference standards (lcap-19, lcap-7) were analyzed along with the samples and percent
recoveries ranged from 99-1 10% (figure 5). Matrix spikes recoveries of Pfa in total and wet extract does not
showup due to high concentration of Pb present in these samples, and also cone, beyond the calibration
curve (100 ppm) of Icp instrument.
Tdp and wet extracts were also post spiksd at 10mg/kg and 4mg/kg level. Percent recoveries for pre-spikes
in wet ranged from 42-76% while in post spike varied 74-121%. Tdp pre-spike recoveries ranged from 1 1-
83% while post spikes were 76-107%. In general, post-spikes recoveries were good for both extraction
methods, but pre-spikes recoveries were better In wet extracts than tdp. Precision was good in both tdp
(under 15%) and wet (under 6%) (figure 6). Both the precision and accuracy were better in Calif, wet than
tdp. For Hg analysis, method spike for wet and tdp gave recoveries 71.6 and 128%, respectively, when
spiked at 1.0 mgykg. EPA WP-287 reference standard (T.V. 0.1 mg/kg) gave 108 % recovery when analyzed
by cold vapor along with soil and sludge samples (Rgure 8a).
CONCLUSION:
Differences exist between tdp and wet methods in terms of solubilizing metals in different matrices of soil
.sludge, and incinerator ash. Wet gave higher results for all EPA regulated metals and some additional metals
tested. Wet results for metals were still higher even when tdp results were multiplied with a factor of two
to account for difference in dilution factor for both the methods. Also wet procedure is simpler than TCLP,
that H does not require no pre-selection of extraction fluid, no pH determination of sample, and also no after
digestion of extract and thus saves lots of total analysis time for routine samples. Although digestion of
samples after extraction is part of tdp method, but in the samples tested it did not make any significant
difference in soluble metals recovered whether extract was digested or not, both in tdp and wet methods.
There is no apparent relationship between tdp or wet in soluble metals extracted as a percentage of total
metals in different matrices.
The authors want to thank Emery G. Lee, Public Health Chemist at HML for analyzing the extracts for Hg
analysis by cold vapor technique and for help in preparation of certain slides by photographing the graphs
from the computer screen.
1. Test Methods for Evaluating Solid Wastes: Physical / Chemical Methods, U.S.Environmental protection
Agency, Office of Solid Waste, Washington, DC, SW 846. Vol. 1A, 3rd edition, Sept. 1986.
2. Federal Register. - Extraction Procedure Toxidty Characteristics" May 19,1980. 45, 33063-33285
3. Federal Register. " Toxidty Characteristic Leaching Procedure " November 7,1986, 51., 40572-40654
4. Bricka. R.Mark. Teresa T. Holmes and M. John Cullinane Jr. 1988. A Comparative Evaluation Of The
USEPA TCLP and EP Extraction Procedures. US Army Engineer Waterways Experiment Station, Vicksburg,
Mississippi 39180.
5. CaBfomia Code of Regulations. Title 22, Vol.29, Article 11, Sections 66699, 66700, Environment
Health, p679-681. published by Bardays Law Publishers, 400 Oyster Point Bl, P.O.Box 3066, South San
Francisco, CA.S4080.
11-536
-------
TCLP Flowchart for Metals
Wet Waste Sample *
<0.5% non-filterable
Solids
Representative
Waste Sample
dry waste
Liquid/Solid
Separation:
0.6 to 0.8
Glass Fiber
Filtration
TCLP
"discard
solid
sample
> Sludge
Contains >0.5%
non-filterable
solids
solid
liquid
Reduce Particle Size
if >9.5mm in narrowest
dimension or surface
area < 3.1 cm2
Liquid/Solid
Separation:
0.6 to 0.8 ;x
Glass fiber
Filtration
extract
Pre Screening
to Select Extraction
Fluid
liquid
Store at 4ฐ C
Liquid Solid
Separation:
0.6 to 0.8 jra
Glass Fiber
Filtration
TCLP
-^-discard
solid
liquid
extract
-> Analytical
I Methods
FIGURE 1
I-537
-------
Major Differences Among
the three Extraction Procedures
WJE.T.
1. One set extracting
solution. Citrate
Buffer pH 5.0
2. Sample to extraction
fluid ratio is 1:10
TCLP
Extraction fluid selection
depends on sample pH:
a. Acetate buffer pH
4.93 ฑ 0.05
b. Acetic acid solution
pH 2.88 ฑ 0.05
Sample to extraction fluid
ratio is 1:20
E.P.TOX
One extraction solution:
distilled deionized H2O +
0.5 N acetic acid to pH
5.0 ฑ 0.2
Sample to extraction fluid
ratio is 1:20
3. Does not specify
extraction vessel
design
4. Requires use of 0.45
nm membrane filter
for extract after
extraction
5. Uses mechanical
shaker for extraction
6. Extraction period of
48 hours
7. No monitoring of pH
required during
extraction
TCLP requires extraction
bottles made of glass,
polypropylene, high
density polyethylene for
non-volatiles
TCLP requires use of 0.6
to 0.8 /im glass fiber filter
Protocol does not specify
reaction vessel design
Requires rotary agitation
in end over end fashion
at 30 ฑ 2 r.p.m.
18 ฑ 2 hours
No monitoring of pH
required during
extraction
Requires use of 0.45
cellulose triacetate filters
Allows either a
blade/stirred open vessel
or a rotary end over end
agitator
24 hours
Requires monitoring and
adjustment of pH to 5.0
during extraction
8. Does not require
acid digestion after
extraction for metals
Requires acid digestion
after extraction for metals
other than mercury
Requires acid digestion
of extract for metals
other than mercury
FIGURE 2
M-538
-------
Maximum Concentration of Metallic
Contaminants for Characteristic of EP Toxicity,
TCLP, and California W.E.T.
Contaminant
Arsenic
Barium
Cadmium
Chromium
Lead
Mercury
Selenium
Silver
Antimony
Beryllium
Cobalt
Copper
Molybdenum
Nickel
Thallium
Vanadium
Zinc
Maximum Concentration
mg/L
5.0
100.0
1 .0
5.0
5.0
0.2
1.0
5.0
California Wet Only
15.0
0.75
80.0
25.0
350.0
20.0
7.0
24.0
250.0
FIGURE 3
I-539
-------
Department of Health Services
Hazardous Materials Laboratory
Inorganic Section
California W.E.T. vs T.C.L.P. Comparison Study
EPA Incinerator Ash
Summary of Results
Total Metals
(mg/kg)
Ag
As
Ba
Ca
Cd
Cr
Mg
Ni
Pb
Se
Zn
A
9.03
73.8
404
54900
368
58.2
6890
27.2
7290
<3.0
23500
B
8.73
69.8
401
55700
372
64.0
7290
31.9
7390
<3.0
24400
C
13.0
64.6
316
54500
392
65.6
7150
28.6
7320
<3.0
24300
Mean
10.3
69.4
374
55000
377
62.6
7110
29.23
7330
<3.0
24100
RSD
23.2
6.6
13.4
1.11
3.4
6.22
2.85
8.25
0.70
0.00
2.05
Soluble Metals by
TCLP Extraction (mg/L)
Soluble Metals by
California W.E.T. (mg/L)
A
<0.01
<0.03
0.18
604
17.8
0.14
89.8
0.17
40.0
<0.06
402
B
<0.01
<0.03
0.17
593
18.4
0.11
88.7
0.20
37.7
<0.06
407
C
<0.01
<0.03
0.19
597
18.5
0.12
90.0
0.18
37.0
<0.06
411
Mean
<0.01
<0.03
0.18
598
18.2
0.12
89.5
0.18
38.2
<0.06
407
RSD
0.00
0.00
5.56
0.93
2.08
12.4
0.78
8.33
4.11
0.00
1.11
A
<0.01
3.47
1.02
3000
26.3
0.80
241
0.52
496
<0.06
1410
B
<0.01
3.62
0.96
3010
27.7
0.84
241
0.49
472
<0.06
1420
C
<0.01
3.69
1.03
2973
28.1
0.82
238
0.49
457
<0.06
1390
Mean
<0.01
3.59
1.00
2990
27.4
0.82
240
0.50
475
<0.06
1410
RSD
0.00
3.13
3.77
0.64
3.45
2.44
0.72
3.46
4.14
0.00
1.09
Gurmail S. Sivia
September, 1990
FIGURE 4
-------
State of California
Department of Health Services
Hazardous Materials Laboratory
Inorganic Sectio
'buality Control for yV.E.T. vs TCLP Study
II
Ag-Silver
As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Ni-Nickel
Pb-Lead
Se-Selenium
Zn-Zinc
Method ''
Blank
<0.01
<0.03
0.01
0.02
<0.06
<0.02
0.08
<0.06
0.04
EPA "
ICAP-19
found
0.99
1.07
1.04
1.02
1.10
1.06
1.10
true
1.00
1.00
1.00
1.00
1.00
1.00
1.00
%
Recovery
99.0
107
104
102
110
106
, 110
EPA "
ICAP-7
found
1.00
1.02
true
1.00
1.00
%
Recovery
100
102
HML ''
Soil QC
found
51.3
47.1
41.7
24.1
49.5
29.7
49.9
54.1
61.1
true
50.9
45.3
39.8
24.3
43.3
30.1
50.7
52.4
55.0
%
Recovery
101
104
104
99.2
114
98.7
98.4
103
111
Spiked Duplicates for Total Metals determinations
Ag-Silver
As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Pb-Lead
Se-Selenium
Unspiked
(mean)
0.21
1.39
7.48
7.54
1.25
147
<0.06
Spike
A
0.53
9.64
11.4
15.5
9.46
*
8.33
Spike
B
0.19
10.5
5.61
16.4
10.1
*
9.47
RPD
94.4
8.54
68.1
5.64
6.54
*
12.8
Spike
added
10
10
10
10
10
10
%Recovery
Mean
1.5
86.8
10.3
84.1
85.3
89.0
Method Spiked Duplicates
% Recovery
(means of 2)
TCLP W.E.T
18.2
95.0
92.0
93.0
88.7
82.7
88.6
67.8
90.8
80.0
80.0
84.1
79.1
92.4
R.P.D.
TCLP W.E.T.
24.8
3.41
4.22
4.86
6.61
9.08
12.9
7.82
0.54
2.26
2.36
0.24
6.75
1.28
Units are mg/L or mg/kg
Sample is EPA Incinerator Ash
FIGURE 5
-------
Stati of California
Department of Health Servioet
Hazardous Materials Laboratory
Inorganic Section
California W.E.T.
TCLP vs California W.E.T. Study
Quality Assurance
Method
Blank
Ag
As
Ba
Cd
Cr
Pb
Se
0.01
0.04
0.01
0.01
0.06
0.02
0.06
Duplicate Spiked Samples
Unspiked
mean
<0.01
3.59
1.00
27.4
0.82
475
<0.06
Spike
A
<0.01
10.2
5.02
32.8
7.58
470
7.45
spike
<0.01
10.7
5.33
33.3
7.73
473
7.79
RPD
0.00
4.78
5.99
1.51
1.96
0.63
4.46
Spike
added
10
10
10
10
10
*
10
% Recoveries
A B mean
0.00
66.1
40.2
54.0
67.6
*
74.5
0.00
71.1
43.3
59.0
69.1
*
77.9
0.00
68.6
41.8
56.5
68.4
76.2
Post Spike
Spike
Result
7.40
13.9
11.1
35.9
10.3
*
12.1
Spike
added
10
10
10
10
10
10
%
Recovery
74.0
102
101
96.0
95.2
121
TCLP
* beyond calibration of ICP
Duplicate Spiked Samples
Ag
AS
Ba
Cd
Cr
Pb
Se
Unspiked
mean
<0.01
<0.03
0.18
18.2
0.12
38.2
<0.06
Spike
A
<0.01
1.12
1.13
26.8
2.67
43.2
2.70
Spike
B
<0.01
1.06
0.97
26.1
2.48
42.8
2.48
RPD
0.00
5.5
15.2
2.6
7.4
0.9
8.5
Spike
Added
10
10
10
10
10
10
10
% Recoveries
A B Mean
0.1
11.2
9.50
86.0
25.5
50.0
26.4
0.1
10.6
7.90
79.0
23.6
46.0
24.2
0.1
10.9
8.70
82.5
24.6
48.0
25.3
Post Spike
Spiked
3.05
4.02
3.98
21.5
3.95
42.3
4.29
Spike
Added
4.00
4.00
4.00
4.00
4.00
4.00
4.00
%
Recovery
76.3
101
95.2
96.8
95.8
102
107
Units are mg/L or mg/kg
Semple Is EPA Incinerator Ash
FIGURE 6
-------
to
u
o.
o
California Wet and TCLP Results
as % of Total (EPA incinerator ash)
o 10.00-1
7.5O-
5.00-
2.50
o.oo
Aa As Ba Cd Cr Pb Se
TCLP
WET
WET and TCLP X 2
Comparison of Results
20 -i
Soil Sample from
Orange County Steel A: Salvage
Anaheim, CA
Cd Cr
EUmenl
b.
IฑJ WET
TCLP X 2
Pb
a.
FIGURE 7
-------
Total, WET, TCLP
Sludge sample F1542
in
f
Ul
700
600
500
400
300
200
100
/
Site: Empire Mine
Grass Valley, CA
As-Arsenic Pb-Lซad
Elซmซnt
10% of Tofal
WET
TCLP
I
Ol
-f
Dl
ii
a:
Total,WET, TCLP
Sludge sample F1542
20
10
Site: Empire Mine
Grass Valley, CA
(X50)
Ba-Barlum Cd-Cadmlum
Elซmซnf
10% of Total
W.E.T.
TCLP
FIGURE 8
-------
s
en
Comparison Of Total, Tclp, And California Wet
Extracts (mg/kg)
HML NUMBER
SAMPLE TYPE
AS-ARSENIC
BA-BARIUM
CD-CADMIUM
CR-CHROMIUM
PB-LEAD
SE-SELENIUM
AG-SILVER
Hg-Mercury
F1542
SLUDGE
TOTAL
1630
44.4
95.1
< 9.40
5760
< 25.5
< 3.90
32.9
WET
19.6
1.33
0.05
< 0.19
478
< 0.51
< 0.08
2.65
TCLP
< 0.19
0.22
< 0.03
< 0.19
64.2
< 0.51
< 0.08
0.53
F1543
SOIL
TOTAL
1750
105
121
9.40
6190
25.5
3.90
95.0
WET
11.8
0.89
0.37
< 0.19
311
< 0.51
< 0.08
3.21
TCLP
0.76
< 0.13
0.18
< 0.19
25.7
< 0.51
< 0.08
0.39
EPA WP-287
Hg-Standard
Quality Control ( Hg-Analysis )
True Value Result % Recovery
Soluble Silver by "WET" and "TCLP"
(filtered through .45 or .6 - .8 micron)
0.100
0.500
0.108
0.470
Mtd-Spike(Wet) Spiked at i.o mg/kg
Mtd-Spike(Tclp) Spiked at 1.0 mg/kg
108
94
71.6
128
4.00
3.20
2.40
1.60
0.80
0.00
/
/
Liquid Samples
Site: Sierra Medical.
Fresno
v
/
V
/
Em WET
KSSSSj TCLP
FI7S3
SompU Numbtr
FIGURE 8A
-------
Total, WET, TCLP
Soil sample F1543
in
rt
-V.
OI
E
400
300
200
Site: Empire Mine
Grass Vajlev,-,
As-Arsenic
Pb-Lซad
Eltmtnl
5% of Total
WET
TCLP
Total, WET, TCLP
Soil sample F1543
2
at
E
B
r.
,
Site: Empire Mine
Grass Valley, CA
BaBarium Cd-Cadmlum
EUmปnt
1% of Total
l:::::x::l WET
FIGURE 9
-------
California Department of Health Services
Hazardous Materials Laboratory
Inorganic Section
Comparison Between Digested and Undigested
TCLP Extracts
HML NUMBER
As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Pb-Lead
Se-Selenium
Ag-Silver
Triple A Hunter's Point, San Francisco (Soil samples)
: C863 C864 C866 C871
D
< 0.19
3.88
< 0.03
< 0.19
1.40
< 0.51
< 0.08
UD
< 0.19
3.82
< 0.03
< 0.19
1.33
< 0.51
< 0.08
D
< 0.19
1.68
0.03
< 0.19
5.37
< 0.51
< 0.08
UD
< 0.19
1.66
0.04
< 0.19
5.31
< 0.51
< 0.08
D
< 0.19
1.33
< 0.04
< 0.19
2.88
< 0.51
< 0.08
UD
< 0.19
1.32
< 0.03
< 0.19
2.82
< 0.51
< 0.08
D
< 0.19
1.07
< 0.03
< 0.19
1.27
< 0.51
< 0.08
UD
< 0.19
1.04
< 0.03
< 0.19
1.20
< 0.51
< 0.08
8
Short Scrap Iron & Metal, Inc., Redding
(Sludge
HML NUMBER :
As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Pb-Lead
Se-Selenium
Ag-Silver
F2541
D
< 0.19 <
1.67
0.18
< 0.19 <
1.28
< 0.51 <
< 0.08 <
UD
0.19
1.68
0.18
0.19
1.21
0.51
0.08
samples)
F2542
D
< 0.19 <
4.26
0.11
< 0.19 <
0.56
< 0.51 <
< 0.08 <
UD
0.19
4.26
0.11
0.19
0.55
0.51
0.08
Notes: D = Digested, UD = Undigested. Mean of two replicates reported,
FIGURE 10
-------
California Department of Health Services
Hazardous Materials Laboratory
Inorganic Section
Comparison Of Digested And Undigested
California Wet Extracts
HML Number
As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Pb-Lead
Se- Selenium
Ag-Silver
Triple A Hunter's Point, San Francisco
( Soil samples )
: C863 C864 C866
D
< 0.19
14.9
0.07
2.88
15.1
< 0.51
< 0.08
UD
< 0.19
14.9
0.08
2.92
15.1
< 0.51
< 0.08
D
< 0.19
7.87
0.08
3.53
21.07
< 0.51
< 0.08
UD
< 0.19
7.65
0.08
3.40
21.0
< 0.51
< 0.08
D
< 0.19
3.24
0.07
1.03
26.2
< 0.51
< 0.08
UD
0.20
3.22
0.07
1.01
26.5
0.51
0.08
C871
D
< 0.19
3.18
0.06
1.04
7.83
< 0.51
< 0.08
UD
< 0.19
3.15
0.07
1.05
7.95
< 0.51
< 0.08
Short
HML Number :
As-Arsenic
Ba-Barium
Cd-Cadmium
Cr-Chromium
Pb-Lead
Se-Selenium
Ag-Silver
Scrap Iron
And Metal
(Sludge
F2541
D
< 0.19
5.62
0.70
0.48
27.5
< 0.51
< 0.08
UD
< 0.19
5.58
0.70
0.47
27.2
< 0.51
< 0.08
Inc. , Redding
samples)
F2542
D
< 0.19 <
24.7
0.75
0.55
17.3
< 0.51 <
< 0.08 <
UD
0.19
24.7
0.78
0.53
17.4
0.51
0.08
Notes: D = Digested, UD = Undigested. Mean of three replicates reported.
FIGURE 11
-------
AUTHOR INDEX
-------
AUTHOR INDEX
Author
Paper
Number
Author
Paper
Number
Author
Paper
Number
Abdel-Hamid, M.
Actor D..
Alchowiak,J
Allison, J
13
....1
19
24
*ป{) 1
Amide, E.N. ---------------- 63
Amin,J..._ _______________________________ 103
Anderau, C. .............................................28
Anderson, D. A........... ____ .... ----- .......10
Anderson, D. R. --------------------------- 68
Ashraf-Khorassani, M ---- 71
Atwood,R.A..
Austiff,G.A...
Baker, R.D....
Bates, C
Barone, G..ซ...ซ
Bath,R.J
Baughman, K. W..
Beaty,R.D
Beckert, W. F.
Bellar,T.A
Bencivengo, D. J
Benedicto, J-
Berges,J.A..
Betowski, L. D.....
Blair, P.O.,
..100
..-95
.-52
88
______ 98
----- 101
________ 29
...49,56
------ 44
Bloemen, H. J. Th. ..
BoUman, M....^.........^
Boyer,D.S..
Breen, J. J.ซ~~.
Broadhead, M..
Brown, J. R. ....
Brown, R. D.....
Bruce, M. L
Bubnis, B
Buote, B..
Burnetti,J...
Butler, L.C
Buxton,B
Bychowski, J. T..
Calvi,J.P.
Carter, J..
>*ป** 4D
.49
11,63
ป53
.....87
...104
....35
Chiang, T.C.H....
Chong, P..
Coakley.W...
Colby, B.N..
Coons, J. T.
Cornell, J. L..
Cotter, R.
Craig, C. A
Cramer, R..
Cunningham,!...
Davis, C.B.
Denoyer, E. R..
Dewalt,G
Dfllard,J.W...
Dodhiwala, N...
Doeffinger, J. ...
Dogruel, D. ~~.
-71
37
>**** ปu O
-------
Author
Paper
Number
Author
Paper
Number
Author
Paper
Number
Naser,Z
Newberry.W.R.....
Nwosu, J................
O'Donnell, A. D
Offutt, C. K. . ....
Olbrot, R. M
Paessun, M. A................
Pan, J. C
Panholzer, F...................
**in% J* *
Persson, J.-A. ...
Petersen, J. ........
Pickering, M.V..
........14
35
........68
90
........................82
.... . 35
g
59
....................... 79
..108
77
i...ซ / /
.....51
.....77
k*ปM<3v
,...57,60,81,100
72
Pohcowicz, C
Pospisil, P. A.
Prashar, S ,
Price,J.
Ray, L.
Rettberg, T. M.
Kicimrosoiiy j^* /v. .ซy
Richter,B. E 72
Robertson, G. L....................11,16,26,63
...34
...87
ปซซt**v4
71
Rosselli, A. C
JCotlini&ii* Wง *<******ป** Iu5
Rubin, R................................................ซ86
Rust, S. .....................................110
Ryan, J. F............ป...........................32,48
MyilUy fป W* ......^4O
Rynaski, A 72
Schaleger, L
Schalk,A.
uCOOIIClCl) iTป Kซ ******ป*<
odineif F 't t*. * ...ป ...........aปi
Schwemberger, J. 110
.65
.79
,91
.75
......................87
...................... / /
......... /
....112
,...105
Seeley, R. C...
Settle, F. A
Shah, N. K
Simes, G. F.
Sivia, G. S
Sleevi, P. ......
Smith, D. .................................................61
Spafford, R. B.
Spear, R. D......
Spurlin, S. R. ....
Stainken, D. M.
Stanton, L ,
SteIz,W.G.
Stephens, M. W.
Stock, M...
Syhre, D.
Tatro, M. E. ...
Taylor, C.
.......42
Thomas, R.
Thomas, R. J.
Tilbury, M. D.
Towa, B
Troast,R.
Tucker, E. S.
Turman, K. ................
Unwin,j. P. .................
..............81
37
VanKley,H.,
Vandermark, T. L.
Vanderveer, E. P..
Vargo,C
Vasavada, S
Venna, V. L.
Villalpbos, K
Voice, T. C ,
Vonk,N
Voyksner, R. D. ...
Wakakuwa, J. ......
Walter, P.
Weesner,F.J
Weichert, B. A.....
Weitz, S.
Wentworth, N. W.
Weston, A. F.
Wetzel,D.L
White, P.
Wilborn, D
Williams, B. .....
Williams, L
Winslow, M. G..
Wittwer, T. .....
Woolfenden, E. .....
Worthington,J.C.
Xiques, D. R..........
Yagley,T.J.
Zweidinger, R. A. .
...81
.....7
..38
69
65
............ 78
.............48
...........104
53,70
...5,21,22
/ /
62
------- |