EPA
United States
Environmental Protection
Agency
Office of Research and
Development
Washington, D.C. 20460
EPA/600/R-98/083
October 1998
Technology Evaluation
Report
Mobile Atomic Absorption
Spectrometer for Metals-
Contaminated Soil Characterization
Pace Environmental Laboratories
-------
EPA/600/R-98/083
October 1998
Technology Evaluation
Report
Mobile Atomic Absorption Spectrometer
for Metals-Contaminated Soil
Characterization
Pace Environmental Laboratories
by
Wayne Einfeld
Gary Brown
Environmental characterization and Monitoring Department
Sandia National Laboratories
Albuquerque, New Mexico 87185-01-0
IAGDW89936700-01-0
Project Officers
Eric Koglin
Stephen Billets
Environmental Sciences Division
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Las Vegas, Nevada 89193-3478
-------
Notice
The information in this document has been funded by the U.S. Environmental Protection Agency (EPA) under an
Interagency Agreement (No. DW89936700-01-0) with the U.S. Department of Energy's Sandia National
Laboratories. This technology evaluation was supported by the Consortium for Site Characterization
Technology, a pilot program operating under the EPA Environmental Technology Verification (ETV) Program.
This report has been subjected to Agency peer and administrative review, and it has been approved for
publication as an EPA document. Mention of corporate names, trade names, or commercial products does not
constitute endorsement or recommendation for use of specific products.
-------
Executive Summary
Consortium for Site Characterization Technology
The U.S. Environmental Protection Agency (EPA), through the Environmental Technology Verification
Program, is working to accelerate the acceptance and use of innovative technologies that improve the way the
United States manages its environmental problems. As part of this program, the Consortium for Site
Characterization Technology was established as a pilot program to test and verify field monitoring and site
characterization technologies. The Consortium is a partnership involving the U.S. Environmental Protection
Agency, the Department of Defense, and the Department of Energy.
This report describes the results of a field demonstration conducted at contaminated sites near Butte, Montana, in
which developers of soil characterization technologies were invited to participate. The report presents soil
sample analysis results from a mobile atomic absorption spectroscopy (AAS) system operated by Pace
Environmental Laboratories. This spectroscopic technique was one of four technologies that were used to
analyze soil samples for a number of target elements. Other technologies that were tested include a laser-induced
breakdown spectrometer operated by MelAok, Inc.; a second laser-induced breakdown spectrometer from Los
Alamos National Laboratory; and anodic stripping voltammetry systems fielded by Battelle Pacific Northwest
National Laboratory. The results from these technology demonstrations are published as separate reports.
Technology Classification
The Consortium classifies each candidate technology into one of three development levels on the basis of the
maturity of the technology and its expected time to commercialization. Level 1 designates the least developed
and Level 3 the most developed technologies. The mobile atomic absorption spectrometer system operated by
Pace Environmental Laboratories was classified as a Level 2 technology. The instrumentation system used in the
demonstration is commercially available and thus is a Level 3 technology. However, the field operators of the
instrument were not the instrument developers, and the technology was originally designed to be operated in a
conventional laboratory.
The Consortium has determined that an exhaustive verification of the relatively new and developing Level 1
technologies should not be performed. Level 2 and Level 3 technologies are analyzed in greater detail, with
Level 3 technologies getting the most complete data analysis. The results from Level 1 and Level 2 technologies
are primarily intended for distribution to the technology developers in order to assist them in further instrument
development and refinement.
Demonstration Design
A demonstration of selected Department of Energy-funded technologies was being planned by MSE-HKM, Inc.,
under contract to the Department of Energy. The Consortium chose to augment the planned demonstration by
111
-------
bringing in additional technologies and enhancing the laboratory analysis component of the project. Two sites
contaminated with heavy metals were identified in the Butte, Montana, area for the demonstration. The first site,
Butte/Silver Bow Creek, was contaminated by heavy metals deposited as mill tailings. The second site,
Anaconda Smelter/Mill Creek, was contaminated by dry aerosol deposition of smelter stack emissions. The
surface soils at both sites contained varying concentrations of heavy metals. Soil conditions at each site were
judged to be representative of typical field conditions under which the technology would be expected to operate.
Sixty samples were collected and processed using a preestablished sampling protocol. The soil samples were
dried, homogenized, and split ten ways for distribution and analysis by three analytical laboratories and four
technologies.
The demonstration plan incorporated the use of reference laboratories to analyze metals in the soil samples using
standard EPA laboratory protocols. Laboratory data produced by inductively coupled plasma atomic emission
spectroscopy and direct-aspiration, flame atomic absorption spectroscopy (AAS) methods were validated to
produce a reference set of target metal concentrations in the field soil samples. The reference data set was used
for comparison with analytical results from the demonstration technologies. Quality control samples were also
incorporated into the sample analysis plan to obtain additional performance measures for the laboratory and field
tests.
Demonstration Results
The mobile atomic absorption spectrometer, fielded in this study by Pace Environmental Laboratories, Inc., is a
conventional laboratory benchtop spectrometer that has been installed in a van along with various support
hardware so that on-site soil sample processing and analysis can be carried out. The atomic absorption technique
is based on the principle that ground-state atoms will absorb light at specific wavelengths for each element. A
preanalysis sample digestion step is required to get the elements into solution prior to the absorption analysis
step. The sample solution is aspirated into an air-acetylene flame through which a beam of light is directed. The
method is quantitative since the degree of light absorption can be directly related to the concentration of atoms in
the sample. The direct-aspiration atomic absorption method used in this demonstration is relatively mature since
the instrument is commercially available and is used in many laboratories.
The Pace mobile atomic absorption system was successfully demonstrated alongside three other participating
technologies in this study. All participants set up and operated their instruments during a 1-week period in the
Butte, Montana, area in September 1995. The incorporation of conventional laboratory analysis into the
demonstration plan provided a validated data set that could be used by developers to evaluate the performance of
the technology. A comparison of the field soil sample results from the reference laboratories showed very close
agreement. This observation suggests that the field soil samples distributed to the demonstration participants
were homogeneous in terms of their chemical composition. A high degree of homogeneity facilitates
comparison of the soil analysis results from the demonstrated technologies with those from the reference
laboratories. Data from the reference laboratory and Pace technology data are presented in a variety of forms to
assist in comparing the data sets produced during the demonstration.
IV
-------
The accuracy of the Pace mobile atomic absorption system was assessed using three methods: (1) Pace results
on quality control samples, (2) mean percent difference estimates for each target element in the field soil samples
relative to reference laboratory data, and (3) regression analysis against reference laboratory data. The latter
gave an overall agreement of ±20 percent or less for most of the nine target elements. Problems were
encountered for selected elements, however. Poor results for arsenic were at least partially attributable to the fact
that the mobile AAS system was not optimized for arsenic determinations. Poor results for chromium were also
encountered and were at least in part attributable to the fact that the level of chromium in the soil samples was at
or near the detection level for the spectroscopic methods employed both by this mobile system and the reference
laboratory.
The overall precision of the mobile system was determined to be in an acceptable range of 20 percent or less by
duplicate analysis of soil sample splits. A comparison of the Pace mobile atomic absorption data with the
reference laboratory data shows generally good agreement between the two data sets. Detailed cost information
on the application and use of this on-site technology is not presented. However, overall costs are expected to be
similar to those encountered in conventional laboratory analyses since the mobile system is essentially a
laboratory system, requiring similar sample processing and operator skills.
-------
Contents
Notice u'
Executive Summary U|
Figures ix
Tables x
Acronyms and Abbreviations xi
Acknowledgments xii
Chapter 1 Introduction 1
Site Characterization Technology Challenge 1
Technology Demonstration Process 1
Technology Identification and Selection 2
Demonstration Planning and Implementation 2
Performance Assessment, Evaluation, and Verification 2
Information Distribution , 3
The Soil-Metals Characterization Demonstration 3
Chapter 2 Technology Description 5
General Description 5
Technology Advantages 5
Technology Limitations 6
Physical Characteristics 6
System Layout 6
Technology Maturity 6
Technology Performance 7
Operational Procedure 7
Quality Control 8
Chapter 3 Demonstration Design and Description 9
VI
-------
Technology Demonstration Objectives 9
Site Selection and Description 9
Site 1 Butte/Silver Bow Creek 10
Site 2 Anaconda Smelter/Mill Creek 13
Sample Collection, Handling, and Distribution 15
Laboratory Selection and Analysis Methodology 16
Columbia Analytical Services 16
MSE Laboratory 17
SNL Environmental Restoration Program Laboratory 17
Demonstration Narrative 17
Deviations from the Demonstration Plan 18
Chapter 4 Laboratory Data Results and Evaluation 19
Laboratory Data Validation Methodology 19
Qualitative Factors 19
Quantitative Factors 19
Laboratory-to-Laboratory Data Comparison 20
Columbia Analytical Services Data 22
General Indicators of CAS Data Quality 22
Quantitative Indicators of CAS Data Quality 22
CAS Performance 26
MSE-HKMData 26
General Indicators of MSE Data Quality 26
Quantitative Indicators of MSE Data Quality 26
MSE Performance 30
Sandia National Laboratories Environmental Restoration Program Laboratory Data 30
General Indicators of SNL Laboratory Data Quality 30
Quantitative Indicators of SNL Laboratory Data Quality 31
SNL Laboratory Performance 31
Laboratory-to-Laboratory Data Comparison 33
VII
-------
Mean Percent Difference 33
Scatter Plots 33
Statistical Bias Testing 37
Intra- and Interlaboratory Variability 38
Reference Laboratory Data Set 39
Chapters Demonstration Results 41
Technology-to-Laboratory Data Comparison Methods 41
Field Observations 41
General Description of Pace AAS Results 41
Quality Control Sample Results 41
Blank Soil Sample Analysis 42
Control Soil Sample Analysis 42
Duplicate Sample Analysis 42
Recovery Analysis 44
Field Soil Sample Analysis Results 44
Comparison of Pace AAS Results with Reference Laboratory Data 49
Mean Percent Difference 49
Correlation Coefficients 49
Statistical Bias Testing 50
Performance Evaluation Conclusions 51
Accuracy 51
Precision 51
General Observations 51
Chapter 6 Developer's Comments 52
References 53
Appendix A: Tabular Data for Pace Flame AAS and Reference Laboratory Field Soil Samples A-l
Vlll
-------
Figures
2-1. Diagram of the mobile laboratory floor plan 7
3-1. Montana regional map showing the Silver Bow and Mill Creek (Anaconda) sampling sites 10
3-2. Local map of the Silver Bow sampling site 11
3-3. Local map of the Mill Creek sampling site 13
4-1. Control soil sample analysis results from CAS 24
4-2. Duplicate soil sample analysis results from CAS 24
4-3. Continuing calibration verification results from CAS 25
4-4. Spiked soil sample recovery results from CAS 25
4-5. Control soil sample results from MSE 28
4-6. Duplicate soil sample results from MSE 28
4-7. Continuing calibration verification results from MSE 29
4-8. Spiked soil sample recovery results from MSE 29
4-9. Control soil sample results from SNL 32
4-10. Continuing calibration verification results from SNL 32
4-11. CAS AAS vs. CAS ICP silver measurements on field replicate soil samples 34
4-12. CAS AAS vs. CAS ICP chromium measurements on field replicate soil samples 34
4-13. CAS AAS vs. CAS ICP iron measurements on field replicate soil samples 35
4-14. MSE ICP vs. CAS ICP silver measurements on field replicate soil samples 35
4-15. MSE ICP vs. CAS ICP chromium measurements on field replicate soil samples 36
4-16. MSE ICP vs. CAS ICP iron measurements on field replicate soil samples 36
5-1. Control soil sample analysis results from the Pace AAS 43
5-2. Duplicate sample analysis results from the Pace AAS 43
5-3. Spike recoveries for Pace AAS 44
5-4. Pace AAS vs. reference laboratory silver 45
5-5. Pace AAS vs. reference laboratory arsenic 45
5-6. Pace AAS vs. reference laboratory copper 46
5-7. Pace AAS vs. reference laboratory cadmium 46
5-8. Pace AAS vs. reference laboratory iron 47
5-9. Pace AAS vs. reference laboratory manganese 47
5-10. Pace AAS vs. reference laboratory lead 48
5-11. Pace AAS vs. reference laboratory zinc 48
IX
-------
Tables
2-1. AAS System Capabilities as Reported by Pace Environmental Laboratories 7
3-1. Demonstration Participants 9
3-2. Typical Heavy Metal Soil Contamination at the Butte/Silver Bow Creek Site 12
3-3. Typical Heavy Metal Soil Contamination at the Anaconda Smelter/Mill Creek Site 14
4-1. Reference Laboratory Blank Soil Sample Results 23
4-2. Serial Dilution Results from MSE 30
4-3. Mean Percent Differences from MSE ICP and CAS AAS Data 33
4-4. Reference Laboratory Linear Regression Results 37
4-5. Wilcoxon Matched Pair Statistical Test Results 38
4-6. Estimates of Intra- and Interlaboratory Sample Variation 39
5-1. Blank Soil Sample Results for Pace AAS 42
5-2. Mean Percent Difference for Pace AAS and Reference Laboratory Data 49
5-3. Linear Regression Parameters for Pace AAS and Reference Laboratory Data 50
5-4. Results from the Wilcoxon Paired Sample Statistical Test 50
-------
Acronyms and Abbreviations
°c
op
AAS
ac
CAS
CCV
CLP
cm
cm2
cm3
DoD
DOE
EPA
g
ICP
ICP-AES
kg
LANL
LCL
LIBS
m
MCHD
MCLD
MCMD
mg
mm/m
MPD
NERL
RPD
SBHD
SBLD
SBMD
SNL
UCL
V
W
degrees centigrade
degrees Fahrenheit
atomic absorption spectroscopy
alternating current
Columbia Analytical Services
continuing calibration verification
Contract Laboratory Program
centimeter
square centimeters
cubic centimeters
Department of Defense
Department of Energy
Environmental Protection Agency
gram
inductively coupled plasma
inductively coupled plasma atomic emission spectroscopy
kilogram
Los Alamos National Laboratory
lower 95 percent confidence limit
laser-induced breakdown spectrometer
meter
Mill Creek-high demonstration
Mill Creek-low demonstration
Mill Creek-medium demonstration
milligram
millimeters per meter
mean percent difference
National Exposure Research Laboratory
relative percent difference
Silver Bow-high demonstration
Silver Bow-low demonstration
Silver Bow-medium demonstration
Sandia National Laboratories
upper 95 percent confidence limit
volt
watt
XI
-------
Acknowledgments
The demonstration of contaminated soils measurement technology benefited from the contributions of numerous
personnel from the U.S. Environmental Protection Agency, the Department of Energy, and the Department of
Defense and their contract organizations.
Participants in the demonstration planning, execution, and analysis were:
EPA National Exposure Research Laboratory, Environmental Science Division
Stephen Billets, Eric Koglin, Gary Robertson
EPA Office of Solid Waste and Emergency Response
Oliver Fordham, Jr.; Howard Fribush
Sandia National Laboratories
Gary Brown, Wayne Einfeld, Michael Hightower, Art Verardo, Susan Bender, Robert Helgesen
MSE-HKM, Inc.
Jay McCloskey, Frank Cook
Columbia Analytical Services
Linda Huckestein
Ames Laboratory
Glenn Bastiaans, Marv Anderson
Naval Research Laboratory
John Moon
Los Alamos National Laboratory
David Cremers
Pace Environmental Laboratories, Inc.
Jim Archer
Battelle Pacific North west National Laboratory
Khris Olsen
MelAok Instruments, Inc.
Hayward Melville
XII
-------
Chapter 1
Introduction
Site Characterization Technology Challenge
Rapid, reliable, and cost-effective field analysis and screening technologies are needed to assist in the complex
task of characterizing and monitoring hazardous and chemical waste sites. Environmental regulators and site
managers often are reluctant to use new technologies that have not been validated in an objective U.S.
Environmental Protection Agency (EPA)-sanctioned testing program or through a similar process that facilitates
acceptance. Until the performance of field characterization technologies can be verified through objective
evaluations, users will remain skeptical of innovative technologies, despite the promise of better, less expensive,
and faster environmental analyses.
The Consortium for Site Characterization Technology was established as a pilot program under the
Environmental Technology Innovation, Commercialization and Enhancement Program, as outlined in 1993 by
President Clinton's Environmental Technology Initiative, to specifically address many of these concerns. The
Consortium is a partnership among the EPA, the Department of Energy (DOE), and the Department of Defense
(DoD). The mission of the Consortium is to identify, demonstrate, and assess innovative field instruments. It
also disseminates information about technology performance to developers, environmental remediation site
managers, consulting engineers, and regulators. As a partnership, the Consortium offers valuable expertise to
support the demonstration of new and emerging technologies. Through its organizational structure, it provides a
formal mechanism for independent assessment, evaluation, and verification of emerging field analytical site
characterization technologies.
Technology Demonstration Process
The Consortium provides technology developers a clearly defined performance assessment, evaluation, and
verification pathway for EPA acceptance. The pathway is outlined in the four components of the Consortium's
evaluation and verification process:
• Technology identification and selection
• Demonstration planning and implementation
• Performance assessment, evaluation, and verification
• Information distribution
Each component is discussed in detail in the following paragraphs.
-------
Technology Identification and Selection
The first step of the process is a determination of technology needs. Because a wide range of field
characterization and monitoring needs exists, the Consortium must prioritize a technology's suitability for
demonstration. Priority is based on the environmental and fiscal impact of the technology and on the likelihood
that its acceptance and use will provide cost-effective and efficient environmental solutions. Surveys of EPA,
DOE, DoD, state, local, and tribal agencies and industry are carried out to identify candidate technologies that
could meet the needs of the environmental characterization community.
Beyond the initial identification, a critical aspect of technology selection is an assessment of the technology's
field deployment readiness. Commercialized instruments, or those ready for production, that have a history of
successful laboratory or field operation are prime candidates for the demonstration process. Early prototypes,
evolving technologies, or laboratory instruments requiring extensive testing and modification prior to field
deployment are less desirable as demonstration candidates. The candidate technology must meet criteria for one
of three levels of maturity:
Level 1 - Demonstrated in a laboratory environment and ready for initial field trials
Level 2 - Demonstrated in a laboratory environment and in limited field trials
Level 3 - Demonstrated extensively in the laboratory and in field trials and commercially available
Assessment of the readiness of candidate technologies for field demonstration is based on the following criteria:
• Field portability or transportability
• Applicability to numerous environmentallyaffected sites
• Potential for solving problems inherent in current analytical methods
• Per sample cost factors
• Potential improvements in data quality, sample preparation, or analysis time
• Ease of use
Demonstration Planning and Implementation
A technology demonstration plan is prepared according to guidelines provided by the Consortium. This plan
includes a technology description, an experimental design, a sampling and analysis plan, a quality assurance
project plan, and a health and safety plan. These plans are designed to enable an objective test of technology
performance. The demonstration plan also calls for the generation of a validated reference laboratory data set
with which the field technology can be compared. Following approval by the EPA and acceptance by the
technology developers, the demonstration plan is implemented at appropriate field locations. The Consortium
provides technical support to the technology developer during plan preparation and execution and also audits the
data collection process.
Performance Assessment, Evaluation, and Verification
In this component of the demonstration process, the technology analytical results are compared with a reference
laboratory data set. The principal product of this phase of the project is a technology report, prepared by an
independent party known as the verification organization. The report documents demonstration results and
provides an assessment of the technology's performance. The degree of data analysis in the report is determined
-------
by the level of maturity of the technology under evaluation; the more mature technologies receive more detailed
analysis.
Level 1 demonstrations are intended to provide the technology developer with access to a controlled field
demonstration in which the system can be tested. A detailed evaluation of system performance is left to the
developer using the validated reference data set obtained during the demonstration. Level 2 technology
performance is evaluated by the Consortium on a limited basis. The most extensive evaluation is done for
Level 3 technologies. In this case, the capabilities of the technology are evaluated by the Consortium, and a
formal verification statement documenting the technology's performance is issued by the EPA.
Information Distribution
Innovative technology evaluation reports from these demonstrations are peer reviewed and approved for
distribution by the EPA. The Consortium has developed an information distribution strategy to ensure that these
documents are readily available to interested parties. This strategy includes access to information via the World
Wide Web through a program supported by the Superfund Technology Innovation Office.
The Soil-Metals Characterization Demonstration
The objectives of the metals-contaminated soil characterization technology demonstration were twofold:
1. provide an opportunity for technology developers to analyze soil samples under a documented and
scientifically sound experimental plan and
2. provide a validated soil analysis data set from conventional analytical laboratories using prescribed EPA
laboratory analysis methods with which technology developers could compare their results
The process used for technology selection involved the publication of a notice of intent to conduct a technology
demonstration, which was accompanied by solicitation of applications from interested parties. Usually, the
Consortium selects applicants based on the readiness of the technology for field demonstration and on its
applicability at environmentally affected sites as determined by the level of regional and national interest in the
specific technology.
For this demonstration, the Consortium joined a project funded by the Department of Energy in which several
technologies had already been selected for demonstration. The Consortium formalized the demonstration plan
development, brought additional technologies to the demonstration, and enhanced the analytical laboratory
component of the project.
Contractual arrangements were established with several chemistry laboratories to conduct soil analyses by
conventional methodologies. Included in these arrangements was a plan to carry out a preliminary site
assessment that involved limited sampling and analysis of soils from the area selected for the demonstrations.
These preliminary data were used to further develop the site sampling and analysis plan, prior to the actual
demonstration.
The following chapters of this report present the details of the demonstration project. Chapter 2 describes the
Pace atomic absorption spectrometer. Chapter 3 describes the site selection, soil sampling, laboratory selection,
and analysis methodology. The technical approach taken in evaluation and validation of laboratory data is also
outlined in Chapter 3. Chapter 4 gives a detailed analysis of the laboratory data validation process and describes
-------
how a reference laboratory data set was determined. Chapter 5 gives results and an analysis of the performance
of the Pace system. Chapter 6 contains developer's comments regarding the demonstration.
-------
Chapter 2
Technology Description
General Description
Flame atomic absorption spectroscopy (AAS) is a mature technology that is used to measure elemental species in
a variety of sample media. Atomic absorption occurs when an atom in a particular energy state absorbs photons
of incident light at a specific wavelength as the atom transitions to a higher energy electronic state. An aerosol
nebulizer in the atomic absorption spectrometer is used to atomize a sample fluid containing the element of
interest into an air-acetylene flame. The thermal energy in the flame destroys all chemical bonds in the aspirated
sample, changing all species to the atomic state. A light beam from a hollow cathode lamp, specific for each
element, is directed through the flame, onto a monochromator, and ultimately to a detector. The monochromator/
detector measures the degree of light absorption from the beam at a specific wavelength band while the solution
is aspirated into the flame. The extent of light absorption is directly related to the concentration of the element of
interest in the liquid sample that is aspirated into the flame.
The direct-aspiration flame AAS method is much like inductively coupled plasma atomic emission spectroscopy
(ICP AES) in that the sample must be in solution prior to analysis. Standard methods usually specify acid
digestion of the sample using a mixture of nitric and hydrochloric acids. Sample digestion can also use nitric
acid and a microwave oven. Only those species that are solubilized in the digestion process can be detected by
the method. Those species that are strongly bound to insoluble components in the sample may not be reduced to
the atomic state in the flame aspiration phase and thus would not be detected by the instrument.
Technology Advantages
According to information provided by Pace Environmental Laboratories, the mobile AAS analytical technique
offers several advantages over conventional laboratory-based atomic emission or absorption spectroscopic
methods. The most important advantages are given below:
• Laboratory hardware equivalence—Laboratory-proventechnology in a mobile environment
• Laboratory analysis method equivalence—EPA SW-846 Method 3050A, Acid Digestion of Sediments,
Sludges, and Soils, and SW-846 Method 7000 A, Atomic Absorption Methods
• Analysis costs—Cost per sample in the mobile laboratory compare favorably with those quoted by a
conventional laboratory
• Instrument precision and accuracy—Accuracy and precision comparable to conventional laboratory
methods
• Response time—Results are available in near real time. Shipping and reporting delays are avoided
-------
Technology Limitations
Several of the more important technology limitations are listed below:
• Analysistime—A throughput time of one soil sample per hour is typical
• Instrument cost—The estimated total capital cost is $ 172,000 for the flame absorption spectrometer and
accompanyingvehicle. Vehicle acquisition and modification costs result in considerably higher costs
compared with a conventional laboratory system
• Instrument mobility—Because of the requirement of a relatively large vehicle to house the spectrometerand
accompanying facilities, areas with poor or limited access pose a problem for on-site analysis
• Analytical versatility—The system is configured for sample introduction by conventional flame aspiration
only and as such is not optimized for low-level arsenic analysis
Physical Characteristics
The Pace Environmental Laboratories mobile AAS system is a Perkin-Elmer Model 3110 atomic absorption
spectrometer that was not modified in any way prior to installation in an 8- x 16-foot trailer. The trailer was
modified to include an exhaust vent above the spectrometer's flame aspiration unit and a fume hood for acid
digestion of soil samples. The overall size of the instrument is 0.7 m (wide) x 0.4 m (high) x 0.6 m (deep). Its
weight is 55 kg, and although the instrument is a comparatively small benchtop unit, it is not portable. The
instrument requires about 300 W of 115 V ac power to operate. About 3 hours were required for initial
instrument setup and about 1 hour setup time each day prior to sample analysis. Operators included two trained
technicians and a staff chemist. All three individuals were involved in the analysis process, principally because
of the considerable time requirements associated with sample preparation and digestion prior to analysis on the
spectrometer.
Ancillary equipment and supplies include an electric generator, gravimetric balance, glassware, fume hood, hot
plates, and various chemicals used in sample digestion. Minimal chemical waste is produced during instrument
operation; however, acid waste is produced during the sample digestion phase of the analysis. Minimal routine
maintenance is required during normal instrument usage. Daily calibrations are performed on the instrument,
with accompanying periodic checks on calibration drift during the sample analysis sequence.
Because this mobile AAS is a commercially available instrument, data processing capabilities are well developed
and automated. Sample throughput is in the range of one to two samples per hour. This estimate is based on 10
hours per day of operational time, with continuous sample preparation, digestion, and analysis for nine target
elements.
System Layout
A diagram of the mobile laboratory layout in the trailer is shown in Figure 2-1.
Technology Maturity
The Pace AAS system is a commercially available laboratory instrument that was mounted in a trailer.
Consequently, the technology can be regarded as mature. Under the guidelines of the Consortium, commercially
available instruments are normally classed as Level 3 in terms of their performance evaluation. This particular
demonstration was given a Level 2 classification because the operators were not affiliated with Perkin-Elmer, the
-------
16'"
Gas |
Cylinder
Storage
Instrumentation Counter Space
Generator
8' Doorway
Fume Hood
Generator
Sink
Refrigerator
Doorway
Data
Acquisition/
Workspace
Figure 2-1. Diagram of the mobile laboratory floor plan.
instrument developer, and because the instrument was designed for use in a fixed laboratory. The instrument
may be considered field mobile rather than field portable.
Technology Performance
The analytical capabilities of the Pace AAS system as reported by Pace Environmental Laboratories are
presented in Table 2-1. Minimum detection levels and accuracy and precision data are given in the table for the
target elements in this demonstration.
Table 2-1. AAS System Capabilities as Reported by Pace Environmental Laboratories
Element
Silver (Ag)
Arsenic (As)
Cadmium (Cd)
Chromium (Cr)
Copper (Cu)
Iron (Fe)
Manganese (Mn)
Lead (Pb)
Zinc (Zn)
Minimum Detection
Level (mg/kg)
0.5
12.5
0.25
2.5
1.0
1.5
0.5
5.0
0.25
Upper Cone. Limit
(mg/kg)
200
5000
100
250
250
300
100
1000
50
Accuracy (%)
±10
±10
±10
±10
±10
±10
±10
±10
±10
Precision (%)
±20
±20
±20
±20
±20
±20
±20
±20
±20
Operational Procedure
All soil samples were digested in accordance with EPA SW-846 Method 3050A, Acid Digestion of Sediments,
Sludges, and Soils. Atomic absorption measurements were carried out using EPA SW-846 Method 7000A. The
specific methods for each of the target elements are given below:
-------
Arsenic None (A nonstandard direct aspiration method was used)
Cadmium Method 7130: cadmium (atomic absorption, direct aspiration)
Chromium Method 7190: chromium (atomic absorption, direct aspiration)
Copper Method 7210: copper (atomic absorption, direct aspiration)
Lead Method 7420: lead (atomic absorption, direct aspiration)
Iron Method 7380: iron (atomic absorption, direct aspiration)
Manganese Method 7460: manganese (atomic absorption, direct aspiration)
Silver Method 7760A: silver (atomic absorption, direct aspiration)
Zinc Method 7950: zinc (atomic absorption, direct aspiration)
Quality Control
The analytical procedure incorporated a number of quality control samples. Initial and continuing calibrations
were run in the same manner as that for the reference laboratories (described in Chapter 4). Method blanks were
also prepared and periodically run throughout the analysis sequence. One out of every 20 of the soil samples was
split and the split spiked with a known amount of each target element. Element recoveries were then determined
in the same manner as that for the reference laboratories (described in Chapter 4). Additional information
concerning the quality control sample results is given in Chapter 5.
-------
Chapter 3
Demonstration Design and Description
Technology Demonstration Objectives
The primary objective of this demonstration was to prepare and execute a scientifically sound test protocol for
the collection and analysis of data from metals-contaminated soil samples as determined by candidate
technologies. To assist the technology developers in evaluating the data collected from their instruments, the
Consortium conducted a parallel analysis of replicate soil samples by conventional laboratory methods.
Table 3-1 lists the demonstration participants and their accompanying technologies.
Table 3-1. Demonstration Participants
Participant
Los Alamos National Laboratory
MelAok Instruments, Inc.
Pace Environmental Laboratories, Inc.3
Battelle Pacific Northwest National Laboratory
MSE-HKM, Inc.
Sandia National Laboratories Environmental
Restoration Program Laboratory
Columbia Analytical Services, Inc.
Technology/ReferenceLaboratory
Laser-induced breakdown spectrometer (LIBS) (technology)
Laser-induced breakdown spectrometer (technology)
Flame atomic absorption spectroscopy (technology)
Anodic stripping voltammetry (technology)
Inductively coupled plasma atomic emission spectroscopy
(reference laboratory)
Inductively coupled plasma atomic emission spectroscopy
(reference laboratory)
Inductively coupled plasma emission spectroscopy and
flame atomic absorption spectroscopy (reference laboratory)
"Point of contact: Jim Archer (612) 525-3475.
The technologies demonstrated, with one exception, were at the low end of the maturity curve. Consequently, a
rigorous technology assessment was not performed on these systems. The soil analysis data from the analytical
laboratories were validated and provided to the developers along with their own data for use in further
development and refinement of their instruments.
Site Selection and Description
To properly assess a field screening technology, a suitable site with soil contaminated by metals was required.
Early in the project, a demonstration plan was developed that presented the following criteria to assist in site
selection.
The site soils must contain a wide concentration range of the heavy metals arsenic, cadmium, chromium,
copper, iron, manganese, lead, silver, and zinc.
-------
• The site must have at least two sample col lection areas with significantly different soil types.
• The heavy metal concentration levels in the soil must be reasonably well characterized and documented.
• The site must be readily accessible for conductingtechnology demonstrations.
The DOE Characterization Monitoring and Sensor Technology Cross-Cut Program had funded a demonstration
project through the Western Environmental Technology Office in Butte, Montana, at a metals-contaminated soil
site. The project had been awarded to MSE-HKM, Inc., an on-site contractor (hereafter referred to as MSE).
Consortium members, including the EPA Environmental Sciences Division of the National Exposure Research
Laboratory (NERL) and Sandia National Laboratories (SNL), chose to augment this demonstration by soliciting
additional technologies for demonstration and by providing additional laboratory analysis of the soil samples
used in the demonstration. During the preparation of the demonstration plan, two sites, Butte/Silver Bow Creek
and Anaconda Smelter/Mill Creek, were selected for the study. Figure 3-1 shows the general location of the sites.
Site 1 Butte/Silver Bow Creek
Location
The Butte/Silver Bow Creek site extends from the west side of Butte, Montana, along Silver Bow Creek to the
confluence of Sand Creek and Silver Bow Creek. The site is contaminated by heavy metals from historic and
modern mining and mill tailings deposits. Figure 3-2 shows the Butte/Silver Bow Creek collection site.
Upper Clark Fork
Super-fund Sites
Silver Bow Creek,'
.' Butte Area Site
Figure 3-1. Montana regional map showing the Silver Bow and Mill
Creek (Anaconda) sampling sites.
10
-------
SBH - High metal* concentration
SUM - Medium meteli concentration
SBL - Low metals concentration
Figure 3-2. Local map of the Silver Bow sampling site.
History
Mining activities in the Butte area started with a group of small gold, silver, and copper mining operations. Butte
became an important mining district in the late 1800s as the size and number of mines grew. With the growth of
ore extraction activities came the need for easy access to ore processing facilities. Consequently, many mills and
smelters were constructed in the region to concentrate and purify ores from the underground mines. Waste
materials from the mineral extraction process, known as tailings, were impounded in ponds and were eventually
discharged into Silver Bow Creek.
Approximately 230 km of stream and riparian habitat have been affected by these local operations. The region of
contamination begins in Butte and extends westward along Silver Bow Creek to the Milltown Reservoir.
Significant mill tailings deposits are found along the creek as well as dispersed over the Silver Bow Creek flood
plain, resulting in a large area of contaminated soil.
During the 1960s and 1970s, mining activities gradually shifted from underground to open-pit mining. In 1982,
the Anaconda Minerals Company discontinued underground mining in Butte. In the same year, the EPA started
site contamination investigations in the area. By the early 1990s, mining operations had ceased and remediation
efforts were implemented.
Characteristics
The Butte/Silver Bow Creek sample area encompasses approximately 5.5 km of Silver Bow Creek. The
principal groundwater-bearing structure is a shallow alluvial aquifer composed of coarse-grained fan and
floodplain deposits. Bedrock formations are found at approximately 1 to 10 m below the surface. The deposits
are moderately permeable and are hydraulically connected to the perennial Silver Bow Creek surface stream.
Because the Silver Bow Creek is an eroding bedrock valley, the erosion slopes are narrow and near the stream.
11
-------
A relatively high surface stream gradient of 3.2 mm/m produces a high-energy stream characterized by a straight
stream channel and narrow floodplain.
Mill tailings deposits at the Butte/Silver Bow Creek site have produced widespread soil contamination. The
contaminated areas are continuous and confined to the narrow floodplain surrounding Silver Bow Creek.
Preliminary characterization efforts, conducted during the site selection process, revealed that heavy metals
deposits are most concentrated in the top 15 to 50 cm of the soil to a maximum depth of 1.2 m (MSE, 1996). A
soil analysis to assess the degree of mill tailings contamination of the local soils was carried out by MSE.
Surface soil analysis results for three sampling locations showing the range of contaminant metal concentrations
are summarized in Table 3-2.
Table 3-2. Typical Heavy Metal Soil Contamination at the Butte/Silver Bow
Creek Site
Metal
Aluminum (Al)
Arsenic (As)
Cadmium (Cd)
Chromium (Cr)
Copper (Cu)
Iron (Fe)
Lead (Pb)
Manganese (Mn)
Silver (Ag)
Zinc (Zn)
Soil Concentration (mg/kg)
Sample 1
6,780
1,200
41.1
7.23
2,150
31,800
2,110
2,490
90.4
12,300
Sample 2
2,990
297
11
6.25
1,350
16,500
681
1,160
15.9
2,710
Samples
9,480
174
0.46
13.5
315
12,200
182
2,170
231
321
Note: Data from a preliminary soil assessment by MSE-HKM, Inc. See MSE, 1996.
Sampling Location Details
The first of three sample areas was selected at a location approximately 45 m north of the Silver Bow Creek bed
in the creek floodplain. The predemonstration samples from this area generally showed the highest
concentrations of contaminant metals of all predemonstration samples. Consequently, this site was designated
"SBHD" (Silver Bow-high demonstration).1 A 27-m, northwest-to-southeast transect of the SBHD sample area
was divided into ten 400-cm2 sample plots equally spaced at 3-m intervals along the transect. Each plot was
designated with the SBHD identifier followed by a plot number ranging from 1 to 10, with the number increasing
from northwest to southeast.
A second sample area was located stream-side, within the Silver Bow Creek bed, and was designated area
"SBMD" (Silver Bow-medium demonstration). A 27-m, northwest-to-southeast transect running along the
streamside of the SBMD sample area was divided into ten 400-cm2 sample plots, equally spaced at 3-m
intervals. Each plot was designated with the SBMD identifier followed by a plot number ranging from 1 to 10,
with the number increasing from northwest to southeast.
A third sample area was located on a hilltop overlooking the SBHD and SBMD sites approximately 120 m from
the stream side and was designated area "SBLD" (Silver Bow-low demonstration). A 27-m, northwest-to-
The naming convention uses high, mid, and low as a matter of convenience. These designations do not always correspond
to the metal concentrations encountered in the samples.
12
-------
southeast transect running along the hill top of the SBLD sample area was divided into ten 400-cm2 sample plots,
equally spaced at 3-m intervals. Each plot was designated with the SBLD identifier followed by a plot number
ranging from 1 to 10, with the plot number increasing from northwest to southeast.
Site 2 Anaconda Smelter/Mill Creek
Location
The Anaconda Smelter/Mill Creek sample area, as shown in Figure 3-3, covers approximately 16 km2 between
Anaconda and Opportunity, Montana. The site is located approximately 40 km west of Butte and near the
Anaconda smelter. It is bounded by state highway 1 to the north and state highway 241 to the west. Flue dust
produced by 100 years of smelter operation has contaminated the site, with heavy metals by the process of aerosol
deposition.
MCH - High metili concentration
MCM - Midium mull connntrttlon
MCL - Low nwtili cooc«nu«tion
-I
Figure 3-3. Local map of the Mill Creek sampling site.
History
The first copper smelting facilities to process ore from Butte area mining operations were in the Anaconda
Smelter/Mill Creek area. The site consists of two facilities, the Upper Works, started in 1884, and the Lower
Works, started in 1888. A silver ore refinery was also located between the copper smelting complexes. Smelter
flue dust containing high levels of metals such as copper, arsenic, cadmium, and lead was produced as a by-
product of the Anaconda smelting activities. Until 1976, flue dust generated by reverberatory furnaces was
reprocessed for arsenic recovery. After 1976, the reverberatory furnaces were replaced by an electric furnace,
and flue dust was collected by a pollution control system.
From 1976 through 1992, nine dust piles with a total volume of approximately 350,000 m3 were deposited on the
hills around the smelter. From 1985 through 1992, wind scouring of the dust piles was controlled by surfactant
13
-------
application. Since 1992, however, considerable amounts of the flue dust have been resuspended and deposited
downwind from the smelter stack and dust piles.
Characteristics
The Anaconda Smelter/Mill Creek sample area is immediately adjacent to the Anaconda Smelter site. The area
consists of a thick layer of moderately permeable, coarse-grained, floodplain deposits over bedrock. Mill Creek
and the sample collection area lie in a structurally broad valley with an accompanying wide floodplain. Mill
Creek is also a tributary of Silver Bow Creek.
Deposition of smelter flue dust at the Anaconda Smelter/Mill Creek site has produced widespread soil
contamination with metals across the entire floodplain. Arsenic, cadmium, and lead are most concentrated in the
top 15 cm of the soil. Cadmium and lead concentrations decrease more rapidly with depth than does arsenic
concentration. Typical analysis results from three surface soil samples taken in the Mill Creek area are presented
in Table 3-3, as measured during the predemonstration site assessment carried out by MSB.
Table 3-3. Typical Heavy Metal Soil Contamination at the
Anaconda Smelter/Mill Creek Site
Element
Al
As
Cd
Cr
Cu
Fe
Pb
Mn
Ag
Zn
Concentration (mg/kg)
Sample 1
5,150
1,170
7.9
10.3
1,320
17,400
515
305
10.3
689
Sample 2
3,450
887
4.66
6.71
573
13,800
400
146
5.03
577
Sample 3
3,640
617
2.92
6.52
506
16,300
277
106
4.63
414
Note: Data from a preliminary soil assessment by MSE Inc.. See MSB, 1996.
Sampling Location Details
The first Mill Creek sampling location was approximately 115m southwest of the highway 1 and highway 241
intersection, and was designated area "MCHD" (Mill Creek-high demonstration).2 A 27-m, southwest-to-
northeast transect of the MCHD sample area was divided into ten 400-cm2 sample plots, equally spaced at 3-m
intervals along the transect. Each plot was designated with the MCHD identifier followed by a plot number
ranging from 1 to 10, with the plot number increasing from southwest to northeast.
A second sample area was located approximately 180 m southwest of the intersection of highway 1 and highway
241, and was designated area "MCMD" (Mill Creek-medium demonstration). A 27-m, west-to-east transect of
the MCMD sample area was divided into ten 400-cm2 sample plots, equally spaced at 3-m intervals. Each plot
was designated using the MCMD identifier followed by a plot number ranging from 1 to 10, with the plot
number increasing from west to east.
The naming convention uses high, mid, and low as a matter of convenience. These designations do not always correspond
to the metal concentrations encountered in the samples.
14
-------
The third Mill Creek sample area was located approximately 115m south of the intersection of highway 1 and
highway 241 and was designated area "MCLD" (Mill Creek-low demonstration). A 27-m, west-to-east transect
of the MCLD sample area was divided into ten 400-cm2 sample plots equally spaced at 3-m intervals. Each plot
was designated using the MCLD identifier followed by a plot number ranging from 1 to 10, with the number
increasing from west to east.
Sample Collection, Handling, and Distribution
Sampling Methods
Ten samples were taken from each of three locations at two sites for a total sample size of 60. The soil in each
400-cm2 sample plot was removed with a clean stainless steel hand trowel to a depth of 2.5 cm, passed through a
No. 10 mesh sieve, homogenized by five passes through a 14-channel riffle splitter, and placed in 1,000-cm3
labeled glass containers. Each 1,000-cm3 sample contained approximately 2.5 kg of soil. Sample collection
proceeded from levels of low metals concentration to high concentration. All sampling equipment was
decontaminated by a detergent wash and double rinse with deionized water between use at each sampling
location.
Sample Handling
All soil samples were taken to MSE, Inc., where they were dried for 12 hours at 105 °C in an oven. After drying,
each soil sample was split ten ways. Each split contained an estimated 150 g of soil and was placed in a labeled
container. Splits were distributed to analytical laboratories, various technology demonstrators, and archives.
Soil sample collection, homogenization, drying, and splitting were carried out during the week of September 18,
1995, by SNL and MSE laboratory personnel prior to the technology demonstration. Samples were stored in
locked coolers at room temperature until distribution.
Sample Distribution
The distribution of the ten sample splits is shown in Table 3-4. The sample numbering convention was in the
format: AABB-NN-nnn, where
AA = Site (SB or MC)
BB = Transect (HD,MD or LD)
NN = Plot No. (01-10)
nnn = Split No. (001-010)
With the exception of Columbia Analytical Services (CAS) and Los Alamos National Laboratory (LANL), each
analytical laboratory and technology demonstrator received a total of 64 samples (60 field soil samples plus 2
blank and 2 control samples). LANL received two sets of splits for a total of 124 samples and CAS received a
total of 32 samples (the 30 field samples plus 1 blank and 1 control sample), because only half of the field soil
samples were selected for analysis at this laboratory.
In addition to soil from the site, each laboratory and technology demonstrator received several quality control
samples. Included in this set were two blank soil samples and two control soil samples prepared and analyzed by
Environmental Resource Associates, Arvada, Colorado, a soils analysis quality control laboratory. These blank
and control samples consisted of topsoil that was dried, ground, sieved, and spiked with various metals (in the
case of the control sample). The soil was then thoroughly homogenized and split into samples that were
15
-------
Table 3-4. Distribution of Field Soil Sample Splits
Soil Sample
Split No.
01
02
03
04
05
06
07
08
09
10
Recipient Technology/ReferenceLaboratory
Los Alamos National Laboratory LIBS (technology)
Los Alamos National Laboratory LIBS (technology)3
MelAok Instruments, Inc. LIBS (technology)
Battelle Pacific Northwest National Laboratory-anodicstripping voltammetry (technology)
Pace Environmental Laboratories, Inc.-flame atomic absorption spectroscopy (technology)
MSE-HKM, Inc. (reference laboratory)
Sandia National Laboratories (reference laboratory)
Columbia Analytical Services, Inc. (reference laboratory)
Sandia National Laboratories- archive
Sandia National Laboratories- archive
a Originally, two similar laser-induced breakdown spectroscopy systems were to be fielded by Los Alamos researchers, with each requiring a
sample split. As a result of logistical difficulties, only one system was actually brought to the site and used in the demonstration.
subjected to a round-robin analysis at qualified laboratories. The results from 20 or more analyses of the soil
batch were used to define a mean value for each element along with a 95 percent confidence interval (mean value
± 2 x standard deviation).
Each laboratory and developer of a demonstration technology was also instructed to produce matrix duplicates of
at least two of the field soil samples so that a measure of analytical precision could be obtained. In the interest of
having a diverse but manageable list of target elements, nine metals were selected for analysis by all participants:
arsenic (As), cadmium (Cd), chromium (Cr), copper (Cu), iron (Fe), lead (Pb), manganese (Mn), silver (Ag), and
zinc (Zn).
Laboratory Selection and Analysis Methodology
Columbia Analytical Services
Analysis of soil samples was carried out at Columbia Analytical Services, Inc., in Kelso, Washington, along with
analysis of several quality control samples. Analysis was carried out at this EPA Contract Laboratory Program
(CLP) laboratory to provide a soil analysis data set that could be used as a cross check with the more
comprehensive soil sample analysis carried out at the MSE laboratory. As a result of program cost constraints,
analysis at the CAS laboratory was limited to half (30) of the 60 field soil samples collected during the
demonstration.
Soil samples were digested using EPA SW-846 Method 3050A: Acid Digestion of Sediments, Sludges, and Soils.
Columbia Analytical Services analyzed all 32 control and field soil samples by inductively coupled plasma
atomic emission spectroscopy using EPA SW-846 Method 6010A.
The laboratory also generated its own duplicates of the 32 soil, control, and blank soil sample digestates and
conducted a second analysis by atomic absorption spectroscopy (AAS) using EPA SW-846 Method 7000A. The
specific methods employed in the analysis included flame aspiration and graphite furnace. They are listed below
for each of the target elements.
16
-------
Ag (silver) Method 7760A: silver (atomic absorption, direct aspiration)
As (arsenic) Method 7060A: arsenic (atomic absorption, furnace technique)
Cd (cadmium) Method 7131 A: cadmium (atomic absorption, furnace technique)
Cr (chromium) Method 7190: chromium (atomic absorption, direct aspiration)
Cu (copper) Method 7210: copper (atomic absorption, direct aspiration)
Fe (iron) Method 7380: iron (atomic absorption, direct aspiration)
Mn (manganese) Method 7460: manganese (atomic absorption, direct aspiration)
Pb (lead) Method 7420: lead (atomic absorption, direct aspiration)
Zn (zinc) Method 7950: zinc (atomic absorption, direct aspiration)
A matrix duplicate sample was also made of original sample number MCLD-1-008. This duplicate was digested
and analyzed by ICP and AAS methods to give a measure of overall laboratory analytical precision on matrix
samples.
MSE Laboratory
The MSE laboratory, located near the sampling site in Butte, Montana, did the preassessment soil sampling and
analysis. It also performed, in collaboration with SNL, the actual demonstration soil sampling, processing, and
distribution. The MSE laboratory carried out a complete analysis of all demonstration soil and quality control
samples. Although MSE is not a CLP laboratory, it used standard EPA SW-846 methodology in its analyses.
The laboratory adheres to quality control procedures specified in the standard EPA analysis protocols used for
soils analysis and operates under a written quality assurance plan.
Sixty soil samples plus 2 control soil samples and 2 blank soil samples were digested using EPA SW-846
Method 3050A: Acid Digestion of Sediments, Sludges, and Soils. All 64 samples were analyzed by ICP using
EPA protocol SW-846 Method 6010A. Matrix duplicates were also made of 4 samples. These underwent
digestion and analysis by ICP so that a measure of method precision could be obtained for this particular soil
matrix.
SNL Environmental Restoration Program Laboratory
The SNL Environmental Restoration Laboratory was selected as an additional reference laboratory. This
laboratory primarily provides rapid screening data which are used in conjunction with conventional CLP-type
analysis for the Sandia internal environmental restoration program. A laboratory quality assurance/control plan
was under development during this study. Data from this laboratory were obtained with a mobile inductively
coupled atomic emission spectroscopy system. The unit is a conventional benchtop ICP system that has been
adapted for field use. The instrument exhibits higher detection limits and more calibration drift than benchtop
units normally used in the laboratory.
Soil samples were digested at the SNL laboratory in a slightly different manner than that used at the other two
laboratories. This laboratory used a microwave-assisted acid digestion method formally designated SW-846
Method 3051: Microwave Assisted Acid Digestion of Sediment, Sludges, Soils, and Oils. The SNL laboratory
analyzed all 64 soil and quality control samples by ICP using EPA protocol SW-846 Method 601OA.
Demonstration Narrative
Predemonstration soil samples were collected during the week of August 21, 1995. These samples were used by
the participants in instrument setup and calibration. The actual demonstration soil samples were collected
17
-------
September 18-22, about 1 week prior to the technology demonstration. Sample processing and packaging was
completed on September 24. Participants in the demonstration were on the site during the week of September
24-29. A complete set of 60 soil samples plus quality control samples were given to each of the participants at
the beginning of the week.
Because access to the actual soil sampling sites was limited and the local media were invited to observe activities
on selected days during the demonstration, the demonstration area was set up on an easily accessible, paved
parking lot about one-half mile from the Silver Bow sampling site. Several vans, tents and generators were
installed at the site to support the various systems. Temperatures ranged from freezing in the morning to the
mid-sixties during the day. Space heaters were used in some of the tents and vehicles during the cold morning
hours. The actual demonstration lasted 6 days; about 2 days were used for instrument setup, checkout, and
disassembly and 4 days for soil analysis. Participants worked at their own pace. A typical day during the
demonstration period began at 9 a.m. and ended at 7 p.m.
Deviations from the Demonstration Plan
A comparison of the demonstration plan prepared prior to the study and the actual conduct of the study as
recorded in the various field and data logbooks reveals a number of discrepancies, which are discussed below.
• The initial soil sampling effort at Silver Bow Creek had to be repeated because a temperature control circuit
failed during sample drying. Soil temperatures were determined to be well in excess of the 105 °C specified
in the demonstration plan. The samples were discarded and additional samples were collected and
processed.
• All soil samples were dried at an oven temperature of 175 °C instead of the 105 °C specified in the
demonstration plan. As noted in the previous paragraph, the primary oven failed and a backup oven had a
minimum temperature control level of 175 °C. In the interest of maintaining the project schedule, the
175 °C dry ing temperature was used.
• Some of the soil sampling was carried out during inclement, rainy weather. Problems were encountered
when sieving moist soil with a No. 10 screen. Larger(No. 6 and No. 8) sieve sizes were used to facilitate
soil processing of the SBLD samples in the field. These and all other samples were homogenized following
sieving so demonstrators and laboratories received comparable samples. Intercomparison of SBLD,
SBMD, and SBHD samples was not done in this study, so sieve size differences among sample sets does
not appear to be significant.
• The certificates of analysis that accompanied the soil control samples were distributed to participants after
all analytical results were submitted to SNL. Access to control soil sample results during the demonstration
was not specified in the demonstration plan, however. This procedure did not compromise the
demonstration design since final analytical data were submitted prior to access to control sample results.
• Analysis of the data from the CAS laboratory revealed beyond a reasonable doubt that two blocks of five
samples were mislabeled. The specific blocks in question were from the Mill Creek sampling site, series
MCHD and MCMD. The switch could have occurred either as a result of mislabeling of sample containers
in the field or during receipt and logging of the samples at the CAS laboratory. An investigation to
determine the source of the error was carried out; however, the source could not be determined from the
available chain-of-custodydocumentation. Despite the fact that a clear incidence of mislabeling could not
be determined, the data were corrected since the switch was unmistakable in the data analysis phase of the
project.
18
-------
Chapter 4
Laboratory Data Results and Evaluation
Laboratory Data Validation Methodology
One of the objectives of this study was to provide the technology developers with a validated set of soil analysis
results from reference laboratory methods for comparison with field results. Both qualitative and quantitative
laboratory data quality indicators were used in the data validation process for all participating laboratories.
These are described more fully in the following sections.
Qualitative Factors
Qualitative factors included degree of experience of the laboratory staff, experience in soils analysis, level of
certification, if any, and past performance on laboratory audits. These factors were used along with additional
quantitative factors in assessing laboratory data quality.
Quantitative Factors
Five specific quantitative factors were also evaluated in the soil analysis data set provided by each laboratory to
assist in the data validation process. These factors were blank sample analysis, control sample analysis,
analytical precision, instrument stability, and spike recovery. Each factor is described more fully in the
following paragraphs.
Soil Blank Analysis
The results from the blind blank soil analyses were directly compared with the information given on the
certificate of analysis accompanying the samples, which were provided by Environmental Resource Associates.
These analysis data were used as a sem{quantitative check on the methods used by the laboratories to detect
contaminant levels, because the soil contained either low or nondetectable levels of many of the target elements.
Control Soil Sample Analysis
The results from the blind control soil sample analysis from each reference laboratory were directly compared
with the certified heavy metal concentrations in the soil, as determined by interlaboratory analyses of the same
lot of soil. Environmental Resource Associates prepared the soil and coordinated the interlaboratory study. An
analysis certificate shipped with the control sample included a certified value and a "performance acceptance
limit"1 for each element in the sample. The results from the control samples from each of the laboratories were
The certificate from Environmental Resource Associates indicates that the performance acceptance limits for each element
"closely approximate the 95% confidence interval about the certified value."
19
-------
an important indicator of laboratory performance levels. Analysis results that fell within the 95 percent
confidence interval were judged to indicate an acceptable level of performance.
Duplicate Analysis Precision
Laboratory analytical precision was estimated by calculating the relative percent difference (RPD) between two
analyses of predigestion duplicate soil samples prepared by each laboratory. The following equation was used.
where
RPD = relative percent difference
Ya = sample result
Yb - duplicate sample result
Relative differences in excess of 20 percent, as specified in EPA Methods 6010A (ICP) and 7000A (AAS), are
taken to indicate questionable laboratory analytical process control.
Instrument Stability
The analytical laboratories also carried out continuing calibration procedures during their sample analyses. In
this procedure, a calibration solution for each of the target elements was analyzed at the onset of the analysis.
The same solutions were periodically analyzed throughout the course of the analysis, typically after every tenth
sample analysis. The results of each check were reported as a percent recovery of the starting calibration value.
The data give an indication of calibration drift encountered over the course of an extended analysis interval. The
control limits, prescribed in EPA Methods 6010A and 7000A, are ± 10 percent of the initial calibration value.
Calibration checks falling outside these limits indicate inadequate analytical process control.
Matrix Spike Recoveries
Some of the laboratories also conducted spiked sample recovery measurements on one or more soil samples. In
this procedure, a measured quantity of each of the target elements was added to a laboratory replicate of a soil
sample. Digestion and analysis of unspiked and spiked samples were carried out. The difference between the
spiked and unspiked sample was compared with the known spiked amount and expressed as a percent sample
recovery. Sample recoveries falling outside the range of 75 to 125 percent, as prescribed in EPA Methods 6010A
and 7000A, are indicative of questionable analytical process control.
Laboratory-to-Laboratory Data Comparison
Summary statistical parameters and data presentation formats were used to provide a quantitative measure of the
degree of comparability among the data sets from the participating laboratories. These are more fully described
below.
20
-------
Method Difference
The method difference or bias is a summary statistic of the difference observed for a particular method relative to
a reference method. The mean percent difference (MPD) of one data set versus another reference data set was
calculated using the following equation:
where,
MPD = mean percent difference
n = number of measurement values
x, = designated reference value
y, = paired value from other method
Scatter Plots
Scatter plots and associated statistical parameters were also used to compare data from one laboratory with that
from another. These plots enable a quick visual comparison. Related statistics include a least-squares method
linear regression giving the best straight line through the data. The regression line has the following equation:
Y=AX+B
where A is the slope of the line and B is the y- intercept value.
The Pearson product-moment correlation coefficient (r) was also computed. This is a measure of the degree of
linearity between the two data sets (Havlicek and Grain, 1988). A correlation coefficient of 1 suggests perfect
correlation while a correlation of 0 indicates no correlation between two data sets.
Statistical Tests
The statistical equivalence of the analytical laboratory data sets was further evaluated with the Wilcoxon
matched pair test. In essence, this nonparametric statistical test allows assessment of whether a statistically
significant bias exists between two methods on a set of paired samples. The test produces a test statistic through
an arithmetic scheme that ranks the differences encountered in sample pair results. The test statistic is essentially
a measure of the ratio of observed differences in the two data sets to expected random differences in the same
two data sets. Knowledge of the test statistic and the sample size allows one to determine whether the
differences encountered in the paired data values can be attributed to the random variation that would be
expected to occur between equivalent methods, or to bias in the methods or data sets. The quantitative aspect of
the test is related to the p-value, which is associated with the test statistic and the number of paired samples used
in the test. By convention, a p-value of 0.05 is often used as the decision point as to whether a statistically
significant bias exists. For example, the determination of a test statistic with an associated p-value of 0.05
indicates that the observed differences between two methods carry a 5 percent chance of being attributable to
random variation alone. Additional information on the use of this nonparametric test for paired-sample analysis
can be found in Conover (1980).
The statistical test results are used in conjunction with linear regression parameters such as slope and intercept to
further compare the two data sets. The statistical test provides an indication as to whether one method is
21
-------
consistently biased relative to another. A second determination is made regarding the extent of that bias, if it
exists. For example, consider the case where the statistical test indicates a significant bias between two sets of
laboratory data. Examination of the linear regression data may reveal that the methods differ by only 5 percent.
In consideration of the overall uncertainties encountered in the sampling and analytical processes, a 5 percent
method bias is tolerable and is not a reason for rejecting one data set over another. This two-phase evaluation of
the data is discussed further in the section dealing with Iaboratory-to-laboratory data comparison.
Columbia Analytical Services Data
Half of the total number of soil samples generated in this demonstration project were analyzed by CAS. A more
detailed qualitative and quantitative assessment of the laboratory's performance follows.
General Indicators of CAS Data Quality
As noted earlier, CAS is a CLP laboratory and follows standard EPA analysis protocols and procedures in its soil
analysis work. Since it is a part of the CLP program, the laboratory also undergoes periodic system audits and
analytical process audits through the use of blind control sample analyses. The laboratory provided a quality
assurance document along with the analysis results for the sample set submitted. Laboratory performance
indicators, such as matrix spike recovery data, duplicate sample summary data, laboratory internal control
sample analysis, and periodic instrument blank and calibration data collected throughout the analysis interval
were included in the report. CAS also provided copies of sample chain-of-custody forms and all raw data
generated in the analysis. No warning flags or out-of-limits quality control indicators were noted in the cover
letter provided with the quality control data package. Personnel from MSE audited the CAS laboratory. The
audit confirmed that CAS operations were in accordance with the standard procedures used in these analyses.
Quantitative Indicators of CAS Data Quality
The analytical results and an accompanying quality control data package were sent by CAS to the Sandia project
leader. The data package contained concentration levels or nondetects reported for all nine target elements in all
32 samples. Specific quantitative data quality factors are discussed in the following paragraphs.
Blank Soil Sample Results
Analytical results from the soil blank analysis are given for CAS ICP and AAS methods as well as for other
participating laboratories in Table 4-1. The "true" metal levels in the soil, as determined by round-robin analysis
of the blank soil lot number at qualified laboratories, are given in the final column of the table.
The CAS analysis results on the blank soil sample track the certified levels reasonably well. Detection levels for
the CAS ICP are slightly higher for As and Pb than for the other target elements. Iron, manganese, chromium,
and zinc are all reported at levels very close to the certified levels. During the course of the analysis, a blank
solution was periodically analyzed with the ICP instrument to check for contamination or excessive calibration
drift. The results from these periodic checks showed consistent instrument detection levels in the expected
concentration range for all target elements.
Control Soil Sample Results
The analytical results for the control soil samples are shown in Figure 4-1 as a percent difference from the
certified value for each element. The analysis certificate supplied with the control soil sample also gives a 95
22
-------
Table 4-1. Reference Laboratory Blank Soil Sample Results
Element
As
Cd
Cr
Cu
Fe
Pb
Mn
Ag
Zn
Metal Concentration Level (mg/kg)
CAS ICP
<40
<1
5
8
6,760
<20
159
<2
27
CAS AAS
1
<0.5
<10
6
7,210
<10
167
<2
28
MSE ICP
2.1
0.4
6.7
5.6
7,740
9.3
172
0.4
24.4
SNL ICP
<98
<8
<19
<76
6,350
<13
<38
<6
76
Certified Level
<2
<1
7
<5
8,180
9
159
<2
24
Notes: A "less than "(<)" symbol indicates not detected. The number following the symbol gives the detection limit. MSE and
SNL data shown are the average of two analyses.
percent confidence interval about the average value as determined by a round-robin study of the soil batch by a
number of qualified analytical laboratories. The upper 95 percent confidence limit (UCL) and lower 95 percent
confidence limit (LCL) are also plotted in Figure 4-1. The CAS results show that the results for all of the target
elements fall within these limits. Most fall within ±10 percent of the certified value for both ICP and AAS
analysis. These data indicate acceptable laboratory performance.
Duplicate Sample Analysis Results
Duplicate results from two soil samples analyzed by both ICP and AAS are given in Figure 4-2. The relative
percent difference between duplicate samples, as described earlier in this section, is plotted for each of the runs.
Plotted RPD values of unity indicate a value of less than or equal to 1. With two exceptions, all RPDs fall within
20 percent. The two exceptions are Cr by AAS and Cd by ICP. No explanation is given as to why these
duplicates showed poor agreement. In general, however, the data reveal acceptable analytical process control.
No precision data are shown for Cr analysis by AAS on sample SBLD-1-008 since a no-detect was reported for
at least one of the determinations.
Instrument Stability
An indication of instrument stability throughout the course of the analysis is given by continuing calibration
verification (CCV) analysis. A known standard is repeatedly run, typically following every 10 analyses on the
ICP or AAS instrument, in order to check instrument calibration drift. The time interval between successive
calibration checks is on the order of 1 hour. Typical CCV results for CAS ICP analysis of four elements are
given in Figure 4-3. The results are plotted in a control chart format with percent recovery relative to the starting
value of the calibration solution on the^-axis and the calibration number on the *-axis. All CCV data for all
target elements from both ICP and AAS analysis indicated recoveries between 90 and 110 percent, which is
within the quality control criteria specified in the method.
Spike Recoveries
Spike recovery data from the CAS analyses are shown in Figure 4-4. Here the deviations from 100 percent
recovery are shown for four spiked soil samples, two of which were analyzed by ICP and two by AAS methods.
In accordance with the standard method, the laboratory did not report recoveries for spiked elements when the
23
-------
80
60
40
20
o
-------
105
104-
103-
-a 102-
>
* 101-
w
° 100-
g?
CD
5 99-
o
^ 98H
97--
95
3 4
Calibration No.
'As ••*••• Cr -^-Pb -a—Zn
50
40
^ 30
§
8 20
I ,o
o
•t -10
o
'•§ -20
B
Q
-30
-40
-50
Figure 4-3. Continuing calibration verification results from CAS.
As
Cd
Cr
Cu
Fe
Element
Pb
Mn
Ag
Zn
|SBLD-1-008 (ICP) EJMCLD-1-008 (ICP) aSBLD-1-008 (Flame) >MCLD-1-008 (Flame)
Figure 4-4. Spiked soil sample recovery results from CAS. See text for
explanation of missing data.
25
-------
spike amount added was less than 25 percent of the unspiked metal content of the sample. Hence, no data are
seen for iron, which was present at high concentration levels in the unspiked soil samples. Spike levels were too
low for As, Cu, Mn, and Zn in selected samples as well, as reflected by no data entries in the accompanying
graph. Standard ICP Method 6010A specifies lower and upper recovery limits at 75 and 125 percent,
respectively (corresponding to ±25 percent difference as plotted in Figure 4-4). The data show that with the
exception of Cr, none of the valid spike recovery levels fall outside this range.
CAS Performance
The foregoing quantitative and qualitative indicators reveal that overall performance of the CAS laboratory was
acceptable. In particular, analysis of blank soil and control soil samples by ICP and AAS reveals acceptable
performance. Spiked sample analysis using the soil matrix generated in the study also gave acceptable recoveries
in all cases, except Cr, in which an adequate spike of each target element was introduced into the original soil
sample. Instrument stability, as evidenced by periodic calibration checks, was also within control limit
guidelines. Together, the quality control parameters suggest a high level of confidence in the accompanying
field soil sample data.
MSE-HKM Data
This DOE contract laboratory has not been part of the EPA CLP program; however, in practice, the laboratory
follows CLP guidelines and standard EPA analysis protocols. A more detailed qualitative and quantitative
assessment of the laboratory's performance follows.
General Indicators ofMSE Data Quality
MSE has a complete quality assurance/control plan, which was sent to the SNL project leader as a part of the
analysis results package. A member of the SNL project team also conducted an audit of the MSE laboratory
prior to the demonstration to determine compliance with standard EPA methods used in this analysis. The audit
report indicated acceptable laboratory procedures and conformance with standard methods used in these
analyses.
MSE included quality control sample documentation in its package. Laboratory performance indicators such as
matrix spike recovery data, duplicate sample summary data, results from an in-house control sample analysis,
periodic instrument calibration data throughout the analysis interval, and periodic blank analysis data throughout
the analysis interval were included. Several out-of-limits conditions were noted in the cover letter associated
with the data package. These anomalies are discussed in detail in later sections of the data presentation.
Quantitative Indicators of MSE Data Quality
Blank Soil Sample Results
Data from the quality control blank soil sample are given in Table 4-1, along with similar data from other
participating laboratories. Detectable amounts of all target elements were reported by MSE, and the agreement
between MSE values and the certified blank soil levels was the best of all three laboratories. The MSE
laboratory detection levels for most of the target elements were the lowest reported of all the participating
laboratories. During the course of the analysis, a blank solution was periodically analyzed with the ICP instrument
to check for contamination or excessive calibration drift. Results from these periodic checks showed consistent
instrument detection levels in the expected concentration range for all target elements.
26
-------
Control Soil Sample Results
The analytical results for control soil samples are shown in Figure 4-5 as a percent difference from the certified
value for each element. The analysis certificate supplied with the control soil sample also gives a 95 percent
confidence interval about the average value as determined by a round-robin study of the soil batch by qualified
analytical laboratories. The upper 95 percent confidence limit and lower 95 percent confidence limit are also
plotted in Figure 4-5. The MSB results, like those from CAS, fall within ±10 percent of the certified value for
nearly all of the target elements. Larger differences on the order of-30 percent are noted for Ag; however, the
reported results still fall within the 95 percent confidence interval about the mean certified level. These data
indicate acceptable laboratory performance.
Duplicate Analysis Results
The relative percent differences are plotted in Figure 4-6 for each laboratory analyses of the duplicate field soil
sample. All RPDs, with two exceptions, fall within the 20 percent criteria. The exceptions are an Mn
measurement with an RPD slightly in excess of 35 percent and a Cd measurement with an RPD of about 28
percent. Three other Mn and Cd precision determinations were within the 20 percent criteria specified in
standard Method 6010 A. The laboratory uses an RPD limit of 20 percent as the acceptable range of variability in
duplicate analysis. Consequently, these results reveal an acceptable degree of analytical process control.
Instrument Stability
A plot of continuing calibration verification data for MSB analysis runs is given in a control chart format in
Figure 4-7. The results for only four elements are given for one of the four batch analyses conducted by the
laboratory. All CCV data for all analyses showed acceptable (± 10 percent of original value) recoveries,
indicating acceptable instrument stability over the course of the analyses.
Spike Recoveries
Spike recovery data from the MSB analyses are shown in Figure 4-8. Element recovery values are shown for
samples that were spiked prior to digestion and analysis of the sample on the ICP instrument. The laboratory
reported recoveries for spiked elements even when the spike amount was less than 25 percent of the unspiked
metal content of the sample. For comparability of the MSE data with CAS data, however, the same spike
validation criteria specified in EPA Method 6010A were applied to the MSE data as well. If the spiked amount
was less than 25 percent of the total elemental content of the sample before the spike, the spike was judged
invalid and no data were reported. Consequently, no data are shown for Fe, Cu, and other elements in selected
instances. The valid set of spike recovery data revealed that only Pb fell outside laboratory acceptance limits of
75 to 125 percent in one of the four batch analyses.
Additional Quantitative Laboratory Data Quality Measures
The MSE quality control data package also revealed several out-of-limits conditions for a serial dilution test that
was carried out on selected field samples. In this test, the concentrations of target elements were measured by
ICP in a dilution of the sample digestate. A fivefold or greater serial dilution was then made of this original
sample and also analyzed by ICP. The measured amount in the diluted sample, taking dilution factors into
account, is expected to agree to within ±10 percent of the original sample amount. Large deviations suggest
sample matrix effects, which may affect quantitative results. The sample matrix may introduce either positive or
negative interferents for a particular element when the sample is analyzed in a relatively concentrated form. The
data from these serial dilution tests are given in Table 4-2. The data show that the ± 10 percent limit of these
27
-------
80
5
•o
40
20
U
E o
8
-20 -
I
Q
-40.
-60 .
-80
"Ag
—-^^—.^ .-. I MMi ^=^^_
i HMSE-ICP2 ;
1 »LCL
• UCL |
Cd Cr Cu Fe
Element
Pb
Mn
Zn
o>
Q.
40
35
30
25
20
15
10
5
0
Figure 4-5. Control soil sample results from MSE. The upper and lower 95
percent confidence limits with respect to the certified values are also shown in
the graph.
As
Cd
Zn
|SSBLD-1-006 (ICP) •SBMD-7-006 (ICP) ^MCLD-1-006 (ICP) QMCMD-7-006 (ICP)
Figure 4-6. Duplicate soil sample results from MSE.
28
-------
4567
Calibration Check No,
8
10
As ••+••• Cr
-a-1 Zn
Figure 4-7. Continuing calibration verification results from MSE.
60 - - - _
-40
-60
As
Cd
SSBLD-1-008 (ICP) pSBMD-7-008 (ICP) QMCLD-1-008 (ICP) jMCMD-7-008 (ICP)
Figure 4-8. Spiked soil sample recovery results from MSE. See text for explanation of
missing data.
29
-------
Table 4-2. Serial Dilution Results from WISE
Element
Ag
As
Cd
Cr
Cu
Fe
Mn
Pb
Zn
Percent difference between measurements at two dilution levels
Sample No.
SBLD-1
62
24
48
17
2.7
1.9
1.7
1.4
0.2
Sample No.
SBMD-7
2.7
0.9
8.3
71
2.6
0.1
1.4
3.6
0.2
Sample No.
MCLD-1
4.2
4.1
64
7.0
2.6
1.7
0.1
4.7
0.2
Sample No.
MCMD-7
100
6.8
39
11
5.7
5.8
4.7
13
4.8
Note: Those values in excess of 10 percent are shown in bold type.
measurements was exceeded for Ag, As, Cd, Cr, and Pb in selected dilution tests. Although these results are not
cause for exclusion of the data, they do reveal that, for at least some of the samples, sample matrix effects
contribute to overall uncertainty in the analytical results.
MSE Performance
The MSE laboratory analysis results on blank and control soil samples, instrument precision and stability, and
spike recovery, in general, reveal acceptable laboratory process control. Several out-of-limits warnings were
encountered in the quality control reports; however, their presence does not warrant rejection of the data set.
Serial dilution recoveries outside the ±10 percent range indicate that sample matrix effects were influential in the
overall quantitative recovery of the field soil samples.
Sandia National Laboratories Environmental Restoration Program Laboratory Data
The SNL Environmental Restoration Laboratory was selected as an additional laboratory. This laboratory
primarily serves to provide rapid screening data which are used in conjunction with CLP-type analyses for
Sandia's internal environmental restoration program.
A quality assurance/control plan was under development during this study. In this analysis the SNL laboratory
followed formal laboratory procedures for soil analyses. Data from this laboratory were obtained with a mobile
laboratory ICP-AES system (shortened to ICP in this report). The unit is a conventional benchtop unit that has
been adapted for field use. Consequently, it exhibits higher detection limits and more calibration drift than the
ICP systems commonly used in the laboratory. A more detailed qualitative and quantitative assessment of the
laboratory's performance follows.
General Indicators of SNL Laboratory Data Quality
The SNL laboratory followed the SW-846 analysis protocols in the soil analysis. The demonstration project
leader did not receive a copy of the laboratory quality assurance plan because the plan was under development at
the time of the demonstration. The SNL laboratory did provide some quality control data such as CCV and
method blank results.
30
-------
Quantitative Indicators ofSNL Laboratory Data Quality
Blank Soil Sample Results
Blank soil data for SNL are presented in Table 4-1, along with similar data from the other participating
laboratories. Nondetectable amounts of all except two target elements were reported by SNL. Detection limits,
in general, were higher for SNL than for the other laboratories owing to the characteristics of the mobile ICP
instrument used in this analysis. Some of the elements, such as Cr and Pb, that were known to exist in the blank
were not detected in the SNL blank analysis as a result of these high detection levels.
Control Soil Sample Results
The analytical results for control soil samples are plotted in Figure 4-9 as percent difference from certified
values. The results show that, with the exception of Ag, all of the target element results fall within the lower and
upper bounds of the 95 percent confidence interval established by the quality control laboratory that developed
and tested the control sample. In general, the results for the target elements fall within +30 percent of the
certified value. Silver results fall outside the lower confidence limit by a margin of nearly 20 percent.
Discussions with laboratory personnel indicated that these results were most likely a result of the poor solubility
of silver in the microwave digestion technique used in this analysis. The microwave method relies solely on
nitric acid rather than on a mixture of nitric and hydrochloric acids used in the conventional digestion technique.
With the exception of the silver analysis, the results reveal acceptable laboratory performance.
Duplicate Analysis Results
No duplicate sample analyses were conducted by the SNL Environmental Restoration Program laboratory.
Instrument Stability
A plot of CCV data for four elements in the SNL runs is given in control chart format in Figure 4-10. Calibration
recoveries fell outside the ±10 percent limits for the following elements: Cd, Cr, Cu, Pb, and Zn. Recovery data
outside the normal control limits revealed stability problems attributable to the mobile ICP system.
Spike Recoveries
No spike recovery analysis was done by the SNL Environmental Restoration Program Laboratory.
SNL Laboratory Performance
Laboratory results for the control soil samples fell within the 95 percent confidence interval of the certified soil
concentration value of the standard for all elements except Ag. The CCV data were outside the normal tolerance
limits of ±10 percent by as much as a factor of two for some of the target elements. Duplicate analyses were not
run on any of the field samples. Consequently, no measure of instrument precision on the actual field soil sample
matrix was available. Matrix spike recovery analysis also was not carried out. In light of the limited extent of
laboratory quality control data, and the fact that a less stable mobile ICP system was used, the judgment was
made to regard these data as informational and not include them in the validated data set from the other reference
laboratories.
31
-------
80
60
40
20
1 0.
u
S
« -»-
u
i
It -40 .
-60
-80 . U-J
-100
Ag As
Cd Cr Cu Fe
Element
nSNLta,
QSNLIbj
QSNL2a;
• SNL2b
I.LCL ,
;*UCL ;
Pb
Mn
Zn
125
120
115
110
Figure 4-9. Control soil sample results from SNL. The upper and lower 95 percent
confidence limits with respect to the certified values are also shown in the graph.
« 105
co
•5
0)
o
100
95
90
85
80
75
3 4
Calibration No.
-As - •* - Cr
»- ,Pb—A—Zn
Figure 4-10. Continuing calibration verification results from SNL.
32
-------
Laboratory-to-Laboratory Data Comparison
The results of several quantitative comparisons of MSB and CAS laboratory data are given in the following
paragraphs. Included are the results and discussion of mean percent difference computations, scatter diagrams,
statistical test results, and a semiquantitative analysis of overall sample variability.
Mean Percent Difference
An estimate of MPD for the soil samples collected at the two sites is given for each target element in Table 4-3.
In this computation, CAS ICP is the designated reference data set on the basis of the laboratory's experience and
acceptable performance on the quality control samples. Thirty sample pairs from each laboratory were used for
comparison because CAS analyzed only half of the total number of soil samples collected. These percent
difference estimates provide a measure of the overall comparability of the three data sets from the two
laboratories. Low difference values reveal agreement between the analyses. The standard deviation is also given
in the table and is a measure of the degree of variability encountered in the computed MPD for each element.
With only a few exceptions, mean differences for nearly all elements are less than ±10 percent in the
comparisons of the CAS ICP reference data set with the CAS AAS and MSB ICP data sets. The comparison of
Ag and Cd between CAS ICP and CAS AAS data sets showed differences on the order of 15 percent. Chromium
by CAS AAS does not compare well at all; however, the comparison with MSE ICP Cr data is quite good. The
poor figures for the CAS AAS Cr data may be attributable to the fact that most of the soil samples had Cr levels
near the lower limit of detection of the AAS method.
Table 4-3. Mean Percent Differences from MSE ICP and CAS AAS Data
Element
Ag
As
Cd
Cr
Cu
Fe
Mn
Pb
Zn
Mean Percent Difference (ref: CAS ICP Data Set)
MSE Laboratory (ICP)
1.3±12.8
0.6±21.3
10.8 ±25.9
7.1 ±31 .4
0.2 ±13.6
6.1+20.4
0.1 ±19.7
-2.1 ± 15.3
-4.7+14.4
CAS Laboratory (AAS)
15.7±13.6
-10.9 ±7.8
-16.6 ±22.9
105.1 + 109.6
4.0 + 3.6
10.5 + 3.2
4.3 + 5.2
5.4 ±1.9
4.2 ± 22.4
Notes: The mean value is followed by the standard deviation. The CAS laboratory ICP AES data set was
used as the reference in this analysis.
Scatter Plots
Scatter plots showing intercomparisons of the CAS AAS and MSE ICP field soil sample data with the
corresponding CAS ICP analysis data are presented in Figures 4-11 through 4-16 for selected elements to
illustrate the various degrees of comparability encountered in the data. The CAS ICP data are plotted on the x-
axis with either the CAS AAS or the MSE ICP data plotted on thejy-axis. The comparison of the CAS AAS data
with the CAS ICP data was very good with the exception of Cr data, shown in Figure 4-12, corroborating the
high mean percent difference value noted for Cr in the previous section.
The MSE data show as good or better correlation with the CAS ICP data. This very close agreement is observed
despite the fact that the CAS ICP and CAS AAS samples were laboratory duplicates from the same field soil
33
-------
100
90
80
70
O)
J£
I BO
0>
s 50
CO
s 40
5
o
30
20
0
0 10 20 30 40 50 60 70 80 90 100
CAS-ICP Silver, mg/kg
Figure 4-11. CAS AAS vs. CAS ICP silver measurements on field replicate soil
samples.
35
30
25
s
O)
E" 20 -
m 15 -
5 10,
5 _
0
0 5 10 15 20 25 30 35
CAS-ICP Chromium, mg/kg
Figure 4-12. CAS AAS vs. CAS ICP chromium measurements on field replicate soil
samples. Nondetectable results are not shown in the plot.
34
-------
50,000
45,000
40,000
f 35,000
O)
30,000
< 25,000
O
20,000
15.000
10,000
10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000
CAS-ICP Iron, mg/kg
Figure 4-13. CAS AAS vs. CAS ICP iron measurements on field replicate soil
samples.
90
80
70
60 -
O)
,.- SO -
1
55
5 4°
til
s
2 30 -
20 -
10 -
0 __
0 10 20 30 40 50 60 70 80 90
CAS-ICP Silver, mg/kg
Figure 4-14. MSE ICP vs. CAS ICP silver measurements on field replicate soil
samples.
35
-------
25
E
§ 15
0
D.
O
III
U) 10
Q _ ____ _ __ __ __ _____ __ __ __ __ , ____ _____ ______ •
0 5 10 15 20 25 30
CAS-ICP Chromium, mg/kg
Figure 4-15. MSE ICP vs. CAS ICP chromium measurements on field replicate soil
samples.
60,000 - - ---- --- ------------------- ---------------------------------- ;
55.000 . •
50,000 I
45.000
3) 40,000
E i
c j
Q 35,000
* "
O !
III 30,000 . • I
2 •
2 B I
25,000 B
20,000 - V \
"• "
15,000 : ^
10,000
10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000 55,000 60,000
CAS-ICP Iron, mg/kg
Figure 4-16. MSE ICP vs. CAS ICP iron measurements on field replicate soil
samples.
36
-------
sample digestate whereas the MSB samples were from a different field sample split. A good comparison
between MSB ICP and CAS ICP data reveals that soil sample splits were chemically similar and that soil
processing and mixing produced relatively homogeneous samples.
The slope and intercept of the best straight line through the data and the correlation coefficient, r, which is a
quantitative measure of the degree of linearity in the data pairs, is given in Table 4-4 for CAS AAS and MSE ICP
data set comparisons with the CAS ICP data set. Coefficients greater than about 0.8 indicate a reasonably strong
linear relationship between the two data sets. Correlation coefficients less than 0.8 are encountered for Cr in
both data sets. The CAS AAS Cr data were plotted against the MSE ICP Cr data and a scatter plot much like that
shown in Figure 4-12 was obtained. This result further suggests that the CAS AAS Cr data may be suspect. The
MSE ICP Cr data show slightly better correlation when plotted against the CAS ICP data, as shown in
Figure 4-15. The slope parameters shown in Table 4-4 are a measure of the bias of one method with respect to
another. With a few exceptions the regression line slopes are in the range of 0.9 to 1.10, which corresponds to a
bias in the range of ±10 percent. Exceptions are encountered for Cd and Cr in the CAS AAS data set as well as
for Cr and Fe in the MSE data set.
Table 4-4. Reference Laboratory Linear Regression Results
Element
Ag
As
Cd
Cr
Cu
Fe
Mn
Pb
Zn
CAS AAS Data Set
Corr. Coeff.
1.00
0.99
0.85
-0.13
1.00
0.99
1.00
1.00
1.00
Slope
1.10
0.96
0.49
0.34
1.07
1.04
1.08
1.06
1.09
Intercept
0.26
-29
2.2
23
-16
1,350
-10
-3.8
-28
MSE ICP Data Set
Corr. Coeff.
1.00
0.99
0.98
0.66
0.99
0.86
0.95
0.92
0.99
Slope
1.02
1.04
0.90
0.83
0.99
1.16
0.93
0.95
0.91
Intercept
-0.1
-16
1.1
2.2
13
-1,980
36
3.1
72
Notes: The CAS ICP data set was used as the reference data set (x variable) in these regression analyses. The y variable was either the CAS
AAS or MSE ICP data set. The slope and intercept values correspond to the values A and B in the linear equation y=Ax + B.
Statistical Bias Testing
The Wilcoxon matched pair test was used to compare the CAS AAS and MSE ICP data sets with the CAS ICP
data set. The SNL laboratory data were not included in this test because they did not meet the data validation
criteria. The Wilcoxon test is a nonparametric test which enables a decision to be made as to whether a
statistically significant bias exists between two methods. The term "nonparametric" refers to the fact that the
observations (in this case the reported metal concentrations in the soil samples) need not conform to a particular
statistical distribution. The Wilcoxson test provides a quantitative measure of the likelihood or probability that
observed differences between two methods are attributable to random variation only. Application of the test
produces a test statistic and an accompanying p-value. The p-value represents the probability of observing a test
statistic value greater than or equal to that obtained in the test from the null or "no difference" distribution—the
distribution of test statistic values that would be encountered if in fact no bias is present between the two
methods in question.
A p-value of 0.05 is often chosen as the boundary point in deciding whether two methods are statistically
different. A test statistic with an accompanying p-value of 0.05 or less indicates that the two methods being
37
-------
compared are statistically different and that the decision to call them different carries a 95 percent chance of
being correct. Alternatively, it can be stated that the decision to call the methods different has a 5 percent chance
of being incorrect.
The results of the statistical test as applied to the CAS AAS and MSB ICP laboratory data sets are summarized in
Table 4-5. The test results between CAS ICP and CAS AAS data sets indicate that significant differences were
observed between the two methods for all elements. The p-values associated with the test statistics for all
elements are less than 0.01, indicating that a clearly distinguishable bias exists between the ICP and AAS
analysis. This observation is corroborated by the scatter plots shown in Figures 4-11 through 4-13. Nearly all
the plotted points fall above a diagonal line extending from the lower left to the upper right corner of the figures.
This line is the zero bias line. Points falling above the diagonal line reveal a positive bias of the AAS method
relative to the ICP method and those falling below the line reveal a negative bias.
Table 4-5. Wilcoxon Matched Pair Statistical Test Results
Element
Ag
As
Cd
Cr
Cu
Fe
Mn
Pb
Zn
Statistically Significant Bias Between Two Methods?
CAS (AAS) vs. CAS (ICP)
Yes(<0.01)
Yes(<0.01)
Yes(<0.01)
Yes(<0.01)
Yes(<0.01)
Yes(<0.01)
Yes(<0.01)
Yes(<0.01)
Yes(<0.01)
MSB (ICP) vs. CAS (ICP)
No (0.67)
No (0.91)
No (0.39)
No (0.94)
No (0.68)
No (0.31)
No (0.99)
No (0.98)
No (0.68)
Note: The p-value associated with the test statistic is given in parentheses.
A statistical comparison of the MSB ICP data with the CAS ICP data reveals that the two data sets are
statistically equivalent; thus no statistically significant method bias exists in one data set with respect to the
other. In this case all p-values associated with the computed test statistic are significantly greater than 0.05. For
example, the p-value associated with the test statistic for Cu was 0.68. This indicates that the observed
differences between the MSE ICP data and the CAS ICP data carry a 68 percent likelihood of being attributable
to random variation between two equivalent methods. These results are corroborated by the scatter plots shown
in Figures 4-14 through 4-16. The plotted points fall above and below the diagonal "zero bias" line with
approximately equal frequency, indicating no consistent bias in the results.
Intra- and Interlaboratory Variability
Each laboratory conducted a duplicate analysis of a digestate from a soil sample split made from a homogenized
bulk field soil sample. The intralaboratory ICP instrument variability was estimated by computing the RPD for
each target element from the duplicate analysis results of sample number MCLD-1 from the CAS and MSE
laboratories. The average of these RPD values is shown in column 2 of Table 4-6 for each target element. The
interlaboratory variability was estimated by computing four RPD values between the four measurement results
from both laboratories and averaging the results. These data are shown in column 3 of Table 4-6. A comparison
of the two columns of data (intra- and interlaboratory RPDs) suggests that in most cases instrument variability is
38
-------
Table 4-6. Estimates of Intra- and Intel-laboratory Sample Variation
Element
Ag
As
Cd
Cr
Cu
Fe
Mn
Pb
Zn
Average Intralab RPD
5.1
1.5
14.3
15.6
3.9
1.9
2.1
3.9
2.2
Average Interlab RPD
5.1
8.1
14.3
9.8
7.4
1.7
2.9
4.6
4.7
of the same order of magnitude as the variability arising from heterogeneity in the sample splits going to the
different laboratories and technologies.
Reference Laboratory Data Set
Based on the foregoing analyses, a reference data set was compiled by averaging the MSB ICP, CAS ICP, and
CAS AAS data sets. This reference data set was then further used for comparison with the soil analysis data sets
provided for the various demonstration technologies. A summary of the reasons for including or excluding the
laboratory data sets in the reference data set is given below.
• The CAS ICP data are judged to be valid, based on the laboratory's acceptable performance on the various
control, duplicate, and soil recovery analyses. The 30-sample CAS ICP data set is used as one component
in the reference data set.
• The 30-sample CAS AAS data set is also included in the reference data set despite the fact that the data set
was shown to be biased with respect to the CAS ICP data set. The decision to include these data in the
reference set was founded upon the linear regression results. Linear regression and correlation analysis
show a high degree of correlation and small bias between the CAS ICP and CAS AAS data. The CAS AAS
biases relative to the CAS ICP method are typically less than 10 percent for most target elements. A bias of
±10 percent is relatively small and acceptable in light of the ±20 percent tolerance in laboratory precision
that was deemed acceptable in the laboratory data validation process. The AAS Cr data, although not well
correlated with the ICP data, were also included in the reference data set. No substantive reasons to exclude
one set of measurements over another were apparent in this particular case. Consequently, both were
included.
• The MSB data are similarly accepted as valid in light of their very good correlation with the CAS ICP data
for all elements and their demonstrated statistical equivalence with the CAS ICP data set.
• The SNL laboratory data are not used in the reference data set. The data package could not be validated
because some key quality control parameters were not provided in the analysis results package.
Furthermore, a less sensitive, lower precision, mobile ICP instrument was used, which contributed to
greater uncertainty in this data set.
In summary, the reference data set is made up of an average of the MSB ICP, CAS ICP, and CAS AAS data sets
for the 30 field soil samples that were analyzed by all three methods. Single values from the MSB ICP data set
are used for the other 30 field samples not analyzed by CAS.
39
-------
The interlaboratory comparisons revealed that all validated data had either a tolerable bias or were statistically
equivalent. Consequently, no elements were excluded in compilation of the reference data set. Although all of
the target elements were included in this set, it should be noted that interlaboratory comparisons revealed that the
results from some elements should be regarded with a lower level of confidence than others. In particular, Cr
results were variable among all three methods and should be treated with appropriate caution when they are used
for comparison with field technology results.
40
-------
Chapter 5
Demonstration Results
Technology-to-Laboratory Data Comparison Methods
The Pace AAS was designated a Level 2 technology. Consequently, a more rigorous evaluation was carried out
than that done for a Level 1 technology. Performance indicators quantitatively evaluated in this process included
instrument performance for each of the nine target elements relative to well-characterized control standards;
instrument precision for the same target elements as determined from the analysis of replicate samples; and a
comparison of Pace AAS data with the reference laboratory data set. The field soil sample analysis data are
presented in Appendix A in tabular format with the results from each laboratory and the laboratory average
shown alongside the Pace AAS results, sample-by-sample, for each target element. In this chapter the Pace AAS
data are plotted against reference laboratory data. Linear regression analysis and nonparametric statistical
analyses, virtually the same as those used for the reference laboratory data set intercomparison in Chapter 4, are
also used to assess the overall comparability of the Pace AAS data relative to laboratory reference data.
Field Observations
Periodically during the demonstration, an observer checked in with the Pace AAS analysis team to monitor
progress. With the exception of one hollow cathode lamp failure, for which a replacement was quickly obtained,
no instrument malfunctions or breakdowns were encountered. The Pace team began their soil analysis on
Monday, September 25, and completed their work on Thursday, September 28, averaging about 15 soil samples
per 10-hour work day. Although the weather during the demonstration was occasionally rainy, windy, and cold,
it did not appear to adversely affect the performance of the Pace-operated system.
General Description of Pace AAS Results
The Pace AAS analysis team produced a complete report in which an analysis result (either a detected amount or
an indication of nondetectable) was obtained for all samples submitted for analysis. A total of 60 field soil
samples plus 2 control soil samples and 2 blank soil samples were analyzed with the Pace AAS system for 9
target elements. Two of the field soil samples were also analyzed a second time in order to obtain an estimate of
instrument analytical precision. The analysis team also prepared a number of internal quality control samples,
such as method blanks and calibration standards, for additional evaluation of instrument performance.
Quality Control Sample Results
The results of the Pace AAS analyses of quality control samples are presented in the following sections. Where
applicable, the results are presented in a format similar to that used in the evaluation of the reference laboratory
data in Chapter 4.
41
-------
Blank Soil Sample Analysis
A comparison of Pace AAS and certified levels for blank soil samples is given in Table 5-1. The MSB
laboratory results are also shown in the table. The Pace AAS results compare reasonably well with certified and
laboratory levels for all elements except As. The As detection limit was reported at a relatively high level of 15
mg/kg. The limit was high since analyses for all target elements, including As, were performed using direct
sample aspiration into an air-acetylene flame. A comment is made in the Pace analysis report that As analysis
should normally be done by either hydride flame AAS or by graphite furnace AAS. The system used had neither
capability, so conventional air-acetylene flame aspiration was used, with a resulting loss in performance for As.
Table 5-1. Blank Soil Sample Results for Pace AAS
Element
As
Cd
Cr
Cu
Fe
Pb
Mn
Ag
Zn
Metal Concentration Level (mg/kg)
Pace AAS
<15
0.6
6
6
8,650
9
190
<2.5
28
MSE ICP
2.1
0.4
6.7
5.6
7,740
9.3
172
0.4
24.4
Certified Level
<2
<1
7
<5
8,180
9
159
<2
24
Notes: All Pace data shown are an average of two measurements. A "less than (<)" symbol indicates
not detected. The number following the symbol is the reported detection limit for the instrument.
Control Soil Sample Analysis
Control soil samples, with well-defined concentration levels of target elements, were analyzed by all participants
in the demonstration, including the Pace analytical team. Control sample results, expressed in terms of a
percentage difference from a certified concentration level of each element in the control soil sample, are given
for the Pace AAS system in Figure 5-1. The plotted data show Pace AAS analysis results within the 95 percent
upper and lower confidence limits about the mean certified value for all elements except one As measurement.
All other target element determinations fall within ±25 percent of the certified soil control sample value. As
noted earlier, this particular instrument configuration was not optimized for As analysis. Normally, As analysis
is done with either a hydride flame or a graphite furnace accessory. In this case arsenic was analyzed by direct
aspiration, with some resulting loss in sensitivity.
Duplicate Sample Analysis
Results from Pace AAS duplicate analyses of two specified soil samples are graphically shown in Figure 5-2.
With a few exceptions, the reported relative percent difference of the target elements are 20 percent or less. A
pair of Ag measurements resulted in a difference value of 47 percent, and a pair of Cr measurements had a value
of 30 percent. The Cu analysis on sample MCMD-1-005 was reported at the same level for both analyses,
resulting in a difference value of 0 percent, which is not indicated on the graph. The results of duplicate sample
analysis indicate generally acceptable instrument performance with regard to analytical precision.
42
-------
= 400
*
O
E
S -40.0
0.
-60.0 -
1.
Ag
As Cd
L
PACE 1
• LCL ;
i«UCL
Cr Cu Fa
Element
Pb
Figure 5-1. Control soil sample analysis results from the Pace AAS. The
upper and lower confidence limits with respect to the certified levels are
also shown on the graph.
c
o
CO
Q
4-1
C
o>
o
CD
0.
-------
Recovery Analysis
Spiked sample recovery analysis results are shown in Figure 5-3. Data are shown only for Ag, As, Cd, and Cr.
The spike levels for the remaining target elements were less than 25 percent of the amount in the sample prior to
the spike; consequently no recovery data are reported for these elements. Recovery data, where available, fall
within the 75 to 125 percent (-25 to 25 percent deviation from 100 percent recovery) criteria specified in the
EPA Method 7000A analysis protocol.
Field Soil Sample Analysis Results
Analysis results were reported for all 60 field soil samples submitted to the Pace AAS analysis team during the
demonstration. The data are presented in two formats to assist in comparing the demonstration technology data
against the data set produced from laboratory analysis of field replicates of soil samples. First, a series of eight
plots (Figures 5-4 through 5-11) are given in which the Pace AAS field soil sample data for each target element
are plotted against the reference laboratory data set. A scatter plot is not shown for Cr since the reference
laboratory data were of unacceptable quality for particular elements. As a part of the laboratory data validation
process, data from the CAS ICP, CAS AAS, and MSE ICP analyses were averaged together to yield a reference
laboratory value. (See Chapter 4 for a discussion of the makeup of the reference laboratory data set.) Although
all data are plotted, in many cases the dots on the scatter plots are overlaid and are thus indistinguishable from
each other.
In general, the scatter plots reveal very good correlation between the Pace AAS and reference laboratory data
sets. Although it is not shown, the worst comparison was for chromium. Chromium analysis results from all
laboratories were highly variable and it appears that results from the mobile Pace AAS instrument are similarly
uncertain. The Cr levels encountered in the field soil samples were in general very close to the detection limits
of the various analytical methods used in these analyses. Consequently, the noise levels in the Cr determinations
from all techniques are comparatively larger, contributing to greater uncertainty in the analysis results from the
laboratory systems as well as the Pace mobile system.
CD
§
U
CD
cc
a
no-
-
"
on-
1
t£
H
1
«u
. ..
J
,
INA
Ag As Cd
Cr Cu Fe Mn
Element
Pb Zn
gg MCMD-4-OOS (AAS) H8 MCLD-1 -008 (AAS) ^ SBHD-6-OOS (AAS) | | SBHD-1 0-005 (AAS)
Figure 5-3. Spike recoveries for Pace AAS. The "NA" indicates that
the spike level was too low for reliable quantification.
44
-------
90
80
70
j? 60
O 50
55
SJ 40 .
< 30 -
20
10
0
0 10 20 30 40 50 60 70 80 90
Lab Reference Silver, mg/kg
Figure 5-4. Pace AAS vs. reference laboratory silver.
1,800 ... .
1,600
1,400
S 1,200
•| 1,000
<2 800 -
o
< 600
a
400
200 -
0 200 400 600 800 1,000 1,200 1,400 1,600 1,800
Lab Reference Arsenic, mg/kg
Figure 5-5. Pace AAS vs. reference laboratory arsenic.
45
-------
4,000
3,500
3.000
°> 2,500
o
a
O 2,000
en
g 1,500
a.
1,000
500
**
Q _ _ _ _
0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000
Lab Reference Copper, mg/kg
Figure 5-6. Pace AAS vs. reference laboratory copper.
50
45
0>
1
18 25
o
%_ 20
111
0 ._
0 5 10 15 20 25 30 35 40 45 50
Lab Reference Cadmium, mg/kg
Figure 5-7. Pace AAS vs. reference laboratory cadmium.
46
-------
50.000
45,000
40.000
*, 35,000
E
— 30.000
CO
^ 25.000
0.
20.000
15.000
mm m
10,000 . _. V •
10,000 15.000 20.000 25,000 30,000 35,000 40,000 45,000 50,000
Lab Reference Iron, mg/kg
Figure 5-8. Pace AAS vs. reference laboratory iron.
4.500 _
4.000
3.500
"S) 3.000
C 2,500 -
2 2,000
CO
<
(J 1.500
OL
1.000
500 -
0 ..^. ,
0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500
Lab Reference Manganese, mg/kg
Figure 5-9. Pace AAS vs. reference laboratory manganese.
47
-------
2.000
1.800
1,600
1.400
O)
1.200
•a
a
fl>
-J 1 ,000
tO
<
LLI 800
O
<
0.
600
200
0
0 200 400 600 800 1,000 1.200 1,400 1,600 1.800 2,000
Lab Reference Lead, mg/kg
Figure 5-10. Pace AAS vs. reference laboratory lead.
18,000 - - - -_. . _. ..- .
16,000
14,000
O) 12.000
I)
o" 10.000
c
N
(O
5 8,000
o
a. 6,000
4.000
2,000
• V
0 2.000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 18,000
Lab Reference Zinc, mg/kg
Figure 5-11. Pace AAS vs. reference laboratory zinc.
48
-------
The Pace AAS data are also presented in tabular form in Appendix A to facilitate their comparison with
individual reference laboratory results. A series of tables give the 60 sample analysis results for CAS ICP, CAS
AAS, MSB ICP, reference laboratory, and Pace AAS analysis for each of the nine target elements.
Comparison of Pace AAS Results with Reference Laboratory Data
The following analytical approaches yield a quantitative measure of the agreement between the Pace AAS data
and the laboratory reference data set.
Mean Percent Difference
The mean percent differences, as defined in Chapter 4, between Pace AAS and reference laboratory data sets are
given in Table 5-2. A small value for the mean percent difference and an accompanying small low standard
deviation is an indicator of good comparability between methods. In general, the results for the Pace AAS data
are quite good. Mean percent difference levels are generally less than 10 percent, with accompanying small
standard deviations. The poorest mean percent differences are encountered with Cr. Relatively low Cr levels
were encountered in the soil samples, resulting in greater uncertainty in these measurements from both the Pace
mobile system and reference laboratory systems used in this comparison.
Table 5-2. Mean Percent Difference for Pace AAS and
Reference Laboratory Data
Element
Ag
As
Cd
Cr
Cu
Fe
Mn
Pb
Zn
Mean Percent Difference
Laboratory Data
-6.5 ±25.7
-16.2±41.3
-6.0 ±17.5
-24.7 ±25.9
-10.2 ±10.7
-7.2 ±19.0
4.5 ±31 .6
-0.6 ±11. 8
-7.6 ±14.2
Note: The mean value is followed by the standard deviation.
Correlation Coefficients
Linear regression results and correlation coefficients between the Pace AAS data set and the reference laboratory
data set are given in Table 5-3. Correlation coefficients near unity reveal good linear correlation between the
data sets. Values near zero reveal no data correlation. Correlation coefficients are greater than 0.9 for all
elements except Cr and Fe. The slopes of the computed regression lines reveal additional information about the
linear relationship between the two data sets. Slopes near unity indicate close comparability of the two methods.
Nearly all elements have slopes in the range of 0.84 to 1.12. Two elements, Cr and Fe, have slopes significantly
different from unity.
49
-------
Table 5-3. Linear Regression Parameters for Pace AAS and
Reference Laboratory Data
Element
Ag
As
Cd
Cr
Cu
Fe
Mn
Pb
Zn
Slope
0.84
1.12
1.01
0.28
0.92
0.59
1.12
0.91
1.00
Intercept
(mg/kg)
1.0
-86
-0.5
4.9
-23
7000
0.6
33
-64
Correlation
Coefficient
0.96
0.97
0.98
0.43
0.97
0.80
0.97
0.97
0.99
Statistical Bias Testing
The Wilcoxon matched pairs test was used to compare the Pace AAS and the reference laboratory data sets for
statistically significant bias. The Wilcoxon test is a nonparametric test that assumes no underlying distribution in
the data sets being compared and is well suited for paired data such as these. The test assesses the likelihood that
observed differences between two methods are a result of random error.
The results of the Wilcoxon test for each of the nine target elements are summarized in Table 5-4. Varied results
are noted. The Pace AAS and reference laboratory data are, statistically speaking, indistinguishable from each
other for three of the nine target elements. The data sets for the other six elements are not statistically
equivalent. These results must be understood in light of the correlation data shown in Table 5-3, however. The
statistical test can detect a small bias in the two methods. For example, the slope of the comparison between Cd
is reported as 1.01, with a correlation coefficient of 0.98, revealing very good agreement between the data sets.
Nonetheless, the statistical test detects a consistent but small bias in the two data sets. The statistical test should
be used in conjunction with a regression analysis or some other measure of overall differences between a
reference and test method. The primary value of the statistical test is in the answer it yields for such elements as
As, Mn, and Pb. The test indicates that no statistically significant bias exists between the two methods and that
Table 5-4. Results from the Wilcoxon Paired
Sample Statistical Test
Element
Ag
As
Cd
Cr
Cu
Fe
Mn
Pb
Zn
Significant Bias? (p-value)
Pace AAS vs. Laboratory Ref
Yes (<0.05)
No (0.72)
Yes (<0.05)
Yes (
-------
further bias comparisons are unwarranted for these particular elements. The observed differences between the
two methods can be explained by random variability in the results. The p-value associated with the test is also
given in the table. A p-value of 0.05 indicates that a 5 percent chance is associated with the assumption that the
observed differences are caused by random variability alone. A p-value of 0.05 is normally used as the decision
point for a statistically significant bias. Values less than 0.05 indicate bias and values greater than 0.05 reveal no
bias.
The mean percent difference data in Table 5-2 give a measure of the method bias relative to the reference
laboratory data. For seven of the nine target elements, the mean percent difference is ±10 percent or less.
Exceptions are encountered for As (-16 percent) and Cr (-25 percent).
Performance Evaluation Conclusions
Accuracy
The accuracy of the Pace AAS system was assessed by comparing Pace AAS results from control soil sample
analyses with certified levels in the samples. In all determinations except one As measurement, the Pace AAS
gave results consistent with known soil concentration levels for the nine target elements. A mean percent
difference for each target element in the field soil samples was also computed using the reference laboratory data
for comparison. In general, mean percent difference estimates by the Pace AAS method were within 10 percent
of the reference laboratory data. Exceptions were noted for As and Cr. Poor results for As may be at least
partially attributable to the fact that the analytical method used in the mobile laboratory was not optimized for As
determinations. Poor results for Cr are at least partially attributable to the fact that Cr analysis results from the
reference laboratories were highly variable as well, thus compromising the quality of reference laboratory data.
Precision
The precision of the Pace AAS, determined by duplicate analysis of soil sample splits, was 20 percent or less for
most of the nine target elements. These results are consistent with the reported experiences of the Pace analysis
team, as noted in Chapter 2.
General Observations
A comparison of the Pace AAS data with the reference laboratory data reveals good agreement between the two
methodologies. This result is not surprising since the Pace AAS is essentially a laboratory instrument contained
in a mobile platform. The performance of the system for As was somewhat compromised for low-level
concentrations of As because the instrument configuration was not optimized for As analysis. Chromium
analysis with the Pace AAS appears to be the most problematic of all the target elements selected for study in
this demonstration. Chromium results were variable and did not compare well with the reference laboratory data.
The laboratory reference data were also judged to be of marginal quality.
A detailed cost analysis was not carried out for this technology because it was designated Level 2. Overall cost
comparisons between a conventional laboratory and this technology indicate generally equivalent costs since the
instrumentation and required accessories for both analytical techniques are the same. The mobile system also
requires the purchase of a mobile platform, however, which would result in higher on-site analytical costs. The
on-site method does offer relatively quick turnaround of samples compared with off-site laboratory services.
51
-------
Chapter 6
Developer's Comments
Sharyl Bergen from Pace Environmental Laboratories, Inc. of Minneapolis, Minnesota, reviewed this report in
May 1996. Pace had no comments or suggested corrections following their review.
52
-------
References
Conover, W. J., 1980. Practical Nonparametric Statistics, 2nd ed, Wiley, New York.
Havlicek, L., and R. D. Grain, 1988. Practical Statistics for the Physical Sciences, American Chemical Society,
Washington DC, pp. 84-93.
MSE, 1996. "Final Report for RCRA and Other Heavy Metals in Soils Demonstration," MSB Technology
Applications, Inc., Butte, Montana.
53
-------
Appendix A
Tabular Data
for
Pace Flame AAS
and
Reference Laboratory
Field Soil Samples
A-l
-------
Table Description
The results are organized by element with two tables for each element. The first table gives results from the
Silver Bow site and the second gives results from the Mill Creek site. The data are further described as follows:
Column 1 Sample Number
Column 2 MSB Laboratory ICP AES Results
Column 3 CAS Laboratory ICP AES Results
Column 4 CAS Laboratory Flame AAS Results
Column 5 Reference Laboratory Data Set (Average of Columns 1-3)
Column 6 Field Technology Results
A-2
-------
Table A-1. Silver Analysis Results for PACE-AAS and Reference Laboratories (Part 1, Silver
Bow Site)
Sample No.
SBHD1
SBHD2
SBHD3
SBHD4
SBHD5
SBHD6
SBHD7
SBHD8
SBHD9
SBHD10
SBMD1
SBMD2
SBMD3
SBMD4
SBMD5
SBMD6
SBMD7
SBMD8
SBMD9
SBMD10
SBLD1
SBLD2
SBLD3
SBLD4
SBLD5
SBLD6
SBLD7
SBLD8
SBLD9
SBLD10
MSE_Ag
(mg/kg)
38
53
36
34
88
58
48
73
54
35
10
22
11
76
18
28
18
18
20
10
2
2
2
3
2
3
4
3
4
5
C_IC_Ag
(mg/kg)
41
35
86
48
49
10
17
18
17
16
2
2
2
5
5
C_AA_Ag
(mg/kg)
46
39
95
53
54
11
19
20
19
18
2
3
3
6
6
Ref_Ag
(mg/kg)
42
53
37
34
90
58
50
73
52
35
10
22
16
76
19
28
18
18
18
10
2
2
2
3
3
3
5
3
5
5
PACE_Ag
(mg/kg)
39
53
34
35
85
57
43
58
47
30
18
18
9
38
31
24
18
18
17
11
3
3
3
3
4
5
6
5
6
6
A-3
-------
Table A-2. Silver Analysis Results for PACE-AAS and Reference Laboratories (Part 2, Mill
Creek Site)
Sample No.
MCHD1
MCHD2
MCHD3
MCHD4
MCHD5
MCHD6
MCHD7
MCHD8
MCHD9
MCHD10
MCMD1
MCMD2
MCMD3
MCMD4
MCMD5
MCMD6
MCMD7
MCMD8
MCMD9
MCMD10
MCLD1
MCLD2
MCLD3
MCLD4
MCLD5
MCLD6
MCLD7
MCLD8
MCLD9
MCLD10
MSE_Ag
(mg/kg)
3
5
4
4
6
6
5
5
8
5
4
4
4
4
4
5
5
5
6
7
4
4
4
4
4
5
5
5
6
7
CJC_Ag
(mg/kg)
3
4
5
4
9
4
4
4
5
5
4
4
4
5
5
C_AA_Ag
(mg/kg)
3
4
7
5
10
5
4
4
5
7
5
4
4
5
7
Ref_Ag
(mg/kg)
3
5
4
4
6
6
5
5
9
5
4
4
4
4
4
5
5
5
6
7
4
4
4
4
4
5
5
5
6
7
PACE_Ag
(mg/kg)
6
12
8
4
8
8
5
6
7
6
3
4
3
3
5
5
3
4
7
5
3
3
3
3
3
3
4
4
5
6
A-4
-------
Table A-3. Arsenic Analysis Results for PACE-AAS and Reference Laboratories (Part 1, Silver
Bow Site)
Sample No.
SBHD1
SBHD2
SBHD3
SBHD4
SBHD5
SBHD6
SBHD7
SBHD8
SBHD9
SBHD10
SBMD1
SBMD2
SBMD3
SBMD4
SBMD5
SBMD6
SBMD7
SBMD8
SBMD9
SBMD10
SBLD1
SBLD2
SBLD3
SBLD4
SBLD5
SBLD6
SBLD7
SBLD8
SBLD9
SBLD10
MSE_As
(mg/kg)
1,060
1,660
874
764
947
1,670
1,510
1,610
1,680
111
145
254
315
549
290
357
414
399
423
260
136
152
130
165
165
132
113
128
101
88
C_IC_As
(mg/kg)
1,110
866
866
1,470
1,490
162
155
316
408
405
144
156
181
148
137
C_AA_As
(mg/kg)
1,010
765
766
1,570
1,530
137
132
291
380
342
120
129
157
106
102
Ref_As
(mg/kg)
1,060
1,660
835
764
860
1,670
1,517
1,610
1,567
111
148
254
201
549
299
357
401
399
390
260
133
152
138
165
168
132
122
128
113
88
PACE_As
(mg/kg)
1,000
1,600
790
690
830
1,500
1,600
1,500
1,700
870
50
130
15
190
250
190
240
340
290
190
15
15
15
15
15
40
15
15
15
15
A-5
-------
Table A-4. Arsenic Analysis Results for PACE-AAS and Reference Laboratories (Part 2, Mill
Creek Site)
Sample No.
MCHD1
MCHD2
MCHD3
MCHD4
MCHD5
MCHD6
MCHD7
MCHD8
MCHD9
MCHD10
MCMD1
MCMD2
MCMD3
MCMD4
MCMD5
MCMD6
MCMD7
MCMD8
MCMD9
MCMD10
MCLD1
MCLD2
MCLD3
MCLD4
MCLD5
MCLD6
MCLD7
MCLD8
MCLD9
MCLD10
MSE_As
(mg/kg)
637
1,570
326
583
1,260
813
466
812
697
882
586
860
717
689
1,240
940
907
1,090
1,650
1,080
640
647
576
757
619
726
814
722
877
1,190
C_IC_As
(mg/kg)
629
347
1,240
461
716
604
757
1,190
828
1,680
704
580
668
811
837
C_AA_As
(mg/kg)
565
275
1,130
366
663
587
708
1,030
789
1,440
604
581
612
753
808
Ref_As
(mg/kg)
610
1,570
316
583
1,210
813
431
812
692
882
592
860
727
689
1,153
940
841
1,090
1,590
1,080
649
647
579
757
633
726
793
722
841
1,190
PACE_As
(mg/kg)
860
1,600
570
720
1,400
930
540
800
740
830
540
1,100
810
770
1,200
1,000
1,100
1,100
1,800
1,400
540
400
540
660
740
760
910
980
1,000
1,300
A-6
-------
Table A-5. Cadmium Analysis Results for PACE-AAS and Reference Laboratories (Part 1,
Silver Bow Site)
Sample No.
SBHD1
SBHD2
SBHD3
SBHD4
SBHD5
SBHD6
SBHD7
SBHD8
SBHD9
SBHD10
SBMD1
SBMD2
SBMD3
SBMD4
SBMD5
SBMD6
SBMD7
SBMD8
SBMD9
SBMD10
SBLD1
SBLD2
SBLD3
SBLD4
SBLD5
SBLD6
SBLD7
SBLD8
SBLD9
SBLD10
MSE_Cd
(mg/kg)
23
13
22
10
49
18
19
18
16
24
6
12
8
17
15
8
10
5
7
6
5
4
4
3
3
5
7
3
2
4
C_IC_Cd
(mg/kg)
27
22
54
18
14
6
15
11
7
6
5
3
3
8
4
C_AA_Cd
(mg/kg)
24
15
20
16
12
3
3
7
10
5
4
3
3
7
3
Ref_Cd
(mg/kg)
24
13
20
10
41
18
18
18
14
24
5
12
9
17
11
8
9
5
6
6
5
4
3
3
3
5
7
3
3
4
PACE_Cd
(mg/kg)
23
12
19
10
47
16
18
16
14
22
5
10
4
11
12
7
7
4
6
6
4
3
3
2
2
4
7
3
3
4
A-7
-------
Table A-6. Cadmium Analysis Results for PACE-AAS and Reference Laboratories (Part 2, Mill
Creek Site)
Sample No.
MCHD1
MCHD2
MCHD3
MCHD4
MCHD5
MCHD6
MCHD7
MCHD8
MCHD9
MCHD10
MCMD1
MCMD2
MCMD3
MCMD4
MCMD5
MCMD6
MCMD7
MCMD8
MCMD9
MCMD10
MCLD1
MCLD2
MCLD3
MCLD4
MCLD5
MCLD6
MCLD7
MCLD8
MCLD9
MCLD10
MSE_Cd
(mg/kg)
4
4
3
4
5
3
4
6
27
4
3
4
4
4
6
5
7
7
8
6
2
4
4
4
4
4
5
4
5
7
C_IC_Cd
(mg/kg)
3
3
4
3
25
4
3
5
5
7
2
3
4
4
3
C_AA_Cd
(mg/kg)
3
2
3
2
22
3
3
5
5
6
2
3
3
4
3
Ref_Cd
(mg/kg)
3 _j
4
3
4
4
3
3
6
25
4
4
4
3
4
5
5
5
7
7
6
2
4
3
4
4
4
4
4
4
7
PACE_Cd
(mg/kg)
3
3
2
3
3
2
2
4
23
4
4
3
3
3
6
5
5
7
8
6
3
3
4
4
4
5
5
5
4
8
A-8
-------
Table A-7. Chromium Analysis Results for PACE-AAS and Reference Laboratories (Part 1,
Silver Bow Site)
Sample No.
SBHD1
SBHD2
SBHD3
SBHD4
SBHD5
SBHD6
SBHD7
SBHD8
SBHD9
SBHD10
SBMD1
SBMD2
SBMD3
SBMD4
SBMD5
SBMD6
SBMD7
SBMD8
SBMD9
SBMD10
SBLD1
SBLD2
SBLD3
SBLD4
SBLD5
SBLD6
SBLD7
SBLD8
SBLD9
SBLD10
MSE_Cr
(mg/kg)
6
4
5
6
7
7
7
6
7
9
13
20
27
21
25
14
14
14
10
8
15
14
13
13
12
14
13
13
12
14
C_IC_Cr
(mg/kg)
8
5
8
5
6
17
12
16
10
9
15
14
14
17
16
C_AA_Cr
(mg/kg)
23
15
11
11
Ref_Cr
(mg/kg)
7
4
5
6
13
7
6
6
6.6
9
15
20
18
21
17
14
12
14
9
8
15
14
13
13
13
14
15
13
14
14
PACE_Cr
(mg/kg)
8
4
7
7
10
9
7
6
8
9
9
11
14
8
7
9
7
12
5
8
10
8
12
10
16
17
11
11
6
7
A-9
-------
Table A-8. Chromium Analysis Results for PACE-AAS and Reference Laboratories (Part 2, Mill
Creek Site)
Sample No.
MCHD1
MCHD2
MCHD3
MCHD4
MCHD5
MCHD6
MCHD7
MCHD8
MCHD9
MCHD10
MCMD1
MCMD2
MCMD3
MCMD4
MCMD5
MCMD6
MCMD7
MCMD8
MCMD9
MCMD10
MCLD1
MCLD2
MCLD3
MCLD4
MCLD5
MCLD6
MCLD7
MCLD8
MCLD9
MCLD10
MSE_Cr
(mg/kg)
8
10
4
6
11
10
7
10
10
13
11
10
11
13
13
10
13
12
12
13
7
8
9
12
8
11
10
7
9
11
C_IC_Cr
(mg/kg)
7
3
10
6
9
10
14
14
12
14
8
7
12
10
8
C_AA_Cr
(mg/kg)
10
11
28
12
21
30
17
21
27
33
19
21
32
15
Ref_Cr
(mg/kg)
8
10
4
6
11
10
14
10
10
13
14
10
18
13
15
10
15
12
18
13
16
8
12
12
14
11
17
7
11
11
PACE_Cr
(mg/kg)
6
8
3
7
6
6
4
6
6
10
9
8
8
8
6
8
9
11
7
8
8
4
7
9
7
8
9
7
8
8
A-10
-------
Table A-9. Copper Analysis Results for PACE AAS and Reference Laboratories (Part 1, Silver
Bow Site)
Sample No.
SBHD1
SBHD2
SBHD3
SBHD4
SBHD5
SBHD6
SBHD7
SBHD8
SBHD9
SBHD10
SBMD1
SBMD2
SBMD3
SBMD4
SBMD5
SBMD6
SBMD7
SBMD8
SBMD9
SBMD10
SBLD1
SBLD2
SBLD3
SBLD4
SBLD5
SBLD6
SBLD7
SBLD8
SBLD9
SBLD10
MSE_Cu
(mg/kg)
1,570
1,330
2,460
991
2,620
1,680
1,010
, 1,030
1,620
1,970
281
864
788
2,180
1,090
780
1,270
449
608
710
394
351
339
414
347
404
566
414
305
363
C_IC_Cu
(mg/kg)
1,670
2,510
2,410
1,010
1,400
385
512
1,240
1,290
644
374
357
332
647
376
C_AA_Cu
(mg/kg)
1,790
2,700
2,620
1,060
1,500
371
522
1,270
1,280
635
376
359
338
648
370
Ref_Cu
(mg/kg)
1,677
1,330
2,557
991
2,550
1,680
1,027
1,030
1,507
1,970
346
864
607
2,180
1,200
780
1,280
449
629
710
381
351
352
414
339
404
620
414
350
363
PACE_Cu
(mg/kg)
1,400
1,200
2,100
870
2,100
1,400
920
780
1,300
2,200
370
850
340
1,600
1,100
580
950
420
500
650
380
360
340
370
290
360
560
350
270
320
A-ll
-------
Table A-10. Copper Analysis Results for PACE-AAS and Reference Laboratories (Part 2, Mill
Creek Site)
Sample No.
MCHD1
MCHD2
MCHD3
MCHD4
MCHD5
MCHD6
MCHD7
MCHD8
MCHD9
MCHD10
MCMD1
MCMD2
MCMD3
MCMD4
MCMD5
MCMD6
MCMD7
MCMD8
MCMD9
MCMD10
MCLD1
MCLD2
MCLD3
MCLD4
MCLD5
MCLD6
MCLD7
MCLD8
MCLD9
MCLD10
MSE_Cu
(mg/kg)
682
792
419
687
956
533
589
859
3,340
889
631
532
585
632
825
893
890
871
1,020
784
476
477
595
554
721
971
853
784
624
1,090
C_IC_Cu
(mg/kg)
663
400
828
626
3,490
631
621
795
821
1,010
513
598
775
878
598
C_AA_Cu
(mg/kg)
701
420
880
668
3,640
657
651
845
885
1,130
535
610
837
916
622
Ref_Cu
(mg/kg)
682
792
413
687
888
533
628
859
3,490
889
640
532
619
632
822
893
865
871
1,053
784
508
477
601
554
778
971
882
784
615
1,090
PACE_Cu
(mg/kg)
620
760
360
630
740
540
510
870
3,700
1,000
630
580
590
650
770
910
740
870
940
800
400
340
580
490
650
740
750
680
540
950
A-12
-------
Table A-11. Iron Analysis Results for PACE-AAS and Reference Laboratories (Part 1, Silver
Bow Site)
Sample No.
SBHD1
SBHD2
SBHD3
SBHD4
SBHD5
SBHD6
SBHD7
SBHD8
SBHD9
SBHD10
SBMD1
SBMD2
SBMD3
SBMD4
SBMD5
SBMD6
SBMD7
SBMD8
SBMD9
SBMD10
SBLD1
SBLD2
SBLD3
SBLD4
SBLD5
SBLD6
SBLD7
SBLD8
SBLD9
SBLD10
MSE_Fe
(mg/kg)
28,700
19,200
21,900
26,300
30,100
26,500
20,700
22,900
29,800
23,700
27,800
43,000
55,200
46,700
49,000
31,300
34,100
33,100
26,300
23,200
15,500
13,900
13,300
13,900
13,500
12,700
11,100
12,000
10,700
11,400
C_IC_Fe
(mg/kg)
30,800
21,700
28,100
19,400
25,400
34,000
29,200
33,100
28,000
25,400
13,000
12,300
13,400
12,400
13,300
C_AA_Fe
(mg/kg)
35,100
24,400
31,600
22,000
28,400
36,900
32,600
37,600
31,200
28,200
14,200
13,500
14,900
13,700
14,400
Ref_Fe
(mg/kg)
31,533
19,200
22,667
26,300
29,933
26,500
20,700
22,900
27,867
23,700
32,900
43,000
39,000
46,700
39,900
31,300
31,100
33,100
26,633
23,200
14,233
13,900
13,033
13,900
13,933
12,700
12,400
12,000
12,800
11,400
PACE_Fe
(mg/kg)
28,000
20,000
21,000
25,000
28,000
25,000
21,000
19,000
26,000
22,000
21,000
26,000
32,000
27,000
21,000
24,000
22,000
33,000
22,000
23,000
14,000
10,000
12,000
9,200
14,000
12,000
12,000
11,000
9,400
9,100
A-13
-------
Table A-12. Iron Analysis Results for PACE-AAS and Reference Laboratories (Part 2, Mill Creek
Site)
Sample No.
MCHD1
MCHD2
MCHD3
MCHD4
MCHD5
MCHD6
MCHD7
MCHD8
MCHD9
MCHD10
MCMD1
MCMD2
MCMD3
MCMD4
MCMD5
MCMD6
MCMD7
MCMD8
MCMD9
MCMD10
MCLD1
MCLD2
MCLD3
MCLD4
MCLD5
MCLD6
MCLD7
MCLD8
MCLD9
MCLD10
MSE_Fe
(mg/kg)
24,100
21,400
23,900
25,900
24,700
23,800
23,000
22,800
49,800
15,400
14,800
16,700
15,600
16,200
17,700
15,700
18,200
16,300
16,200
16,300
18,600
20,100
20,600
21,700
18,400
20,600
20,200
18,400
22,700
18,200
CJC_Fe
(mg/kg)
23,100
26,300
24,400
24,100
49,200
13,500
17,200
17,500
16,500
16,800
19,200
18,500
20,400
18,900
20,600
C_AA_Fe
(mg/kg)
25,100
27,300
26,800
26,600
49,500
14,800
18,900
19,200
19,800
18,900
21,000
20,000
22,900
20,800
22,800
Ref_Fe
(mg/kg)
24,100
21,400
25,833
25,900
25,300
23,800
24,567
22,800
49,500
15,400
14,367
16,700
17,233
16,200
18,133
15,700
18,167
16,300
17,300
16,300
19,600
20,100
19,700
21,700
20,567
20,600
19,967
18,400
22,033
18,200
PACE_Fe
(mg/kg)
23,000
22,000
24,000
27,000
25,000
22,000
20,000
20,000
42,000
16,000
12,000
16,000
14,000
15,000
16,000
23,000
23,000
27,000
19,000
23,000
20,000
16,000
21,000
19,000
17,000
21,000
19,000
20,000
20,000
16,000
A-14
-------
Table A-13. Manganese Analysis Results for PACE-AAS and Reference Laboratories (Part 1,
Silver Bow Site)
Sample No.
SBHD1
SBHD2
SBHD3
SBHD4
SBHD5
SBHD6
SBHD7
SBHD8
SBHD9
SBHD10
SBMD1
SBMD2
SBMD3
SBMD4
SBMD5
SBMD6
SBMD7
SBMD8
SBMD9
SBMD10
SBLD1
SBLD2
SBLD3
SBLD4
SBLD5
SBLD6
SBLD7
SBLD8
SBLD9
SBLD10
MSE_Mn
(mg/kg)
2,460
2,940
2,100
1,380
2,000
2,700
1,920
2,930
1,800
1,960
935
2,500
1,150
3,320
2,560
2,090
1,070
549
1,080
850
708
478
409
379
563
556
391
505
434
513
C_IC_Mn
(mg/kg)
2,520
2,130
1,900
1,870
1,780
794
2,270
2,130
1,530
608
730
458
532
448
540
C_AA_Mn
(mg/kg)
2,720
2,320
2,070
2,040
1,970
847
2,320
2,380
1,680
652
794
456
564
456
593
RefJVIn
(mg/kg)
2,567
2,940
2183
1,380
1,990
2,700
1,943
2,930
1,850
1,960
859
2500
1,913
3,320
2,357
2,090
1,427
549
780
850
744
478
441
379
553
556
432
505
522
513
PACE_Mn
(mg/kg)
2,700
3,300
2,600
1,800
2,300
3,000
2,300
3,200
2,100
2,300
2,100
2,300
1,500
4,200
1,900
2,800
2,000
660
2,000
710
670
410
490
320
530
470
400
510
410
820
A-15
-------
Table A-14. Manganese Analysis Results for PACE-AAS and Reference Laboratories (Part 2,
Mill Creek Site)
Sample No.
MCHD1
MCHD2
MCHD3
MCHD4
MCHD5
MCHD6
MCHD7
MCHD8
MCHD9
MCHD10
MCMD1
MCMD2
MCMD3
MCMD4
MCMD5
MCMD6
MCMD7
MCMD8
MCMD9
MCMD10
MCLD1
MCLD2
MCLD3
MCLD4
MCLD5
MCLD6
MCLD7
MCLD8
MCLD9
MCLD10
MSE_Mn
(mg/kg)
113
122
71
94
121
75
88
161
2,180
222
284
202
263
210
222
170
288
235
219
188
119
102
120
139
132
146
163
139
115
357
C_IC_Mn
(mg/kg)
115
78
124
96
2,230
267
266
206
270
225
124
113
138
155
108
C_AA_Mn
(mg/kg)
115
82
105
98
2,350
280
271
210
277
236
125
113
143
158
108
Ref_Mn
(mg/kg)
114
122
77
94
117
75
94
161
2,253
222
277
202
267
210
213
170
278
235
227
188
123
102
115
139
138
146
159
139
110
357
PACE_Mn
(mg/kg)
110
120
74
97
100
78
81
150
2,400
230
250
200
240
200
180
160
250
230
200
180
110
86
110
120
110
120
140
120
100
340
A-16
-------
Table A-15. Lead Analysis Results for PACE-AAS and Reference Laboratories (Part 1, Silver
Bow Site)
Sample No.
SBHD1
SBHD2
SBHD3
SBHD4
SBHD5
SBHD6
SBHD7
SBHD8
SBHD9
SBHD10
SBMD1
SBMD2
SBMD3
SBMD4
SBMD5
SBMD6
SBMD7
SBMD8
SBMD9
SBMD10
SBLD1
SBLD2
SBLD3
SBLD4
SBLD5
SBLD6
SBLD7
SBLD8
SBLD9
SBLD10
MSE_Pb
(mg/kg)
1,170
1,010
946
936
2,080
1,310
1,030
1,770
1,500
1,140
410
631
456
779
677
836
696
466
537
342
200
166
139
245
173
217
324
193
264
294
C_IC_Pb
(mg/kg)
1,220
902
1,850
992
1,310
1,260
513
823
798
471
166
147
161 _j
374
302
C_AA_Pb
(mg/kg)
1,290
955
2,000
1,060
1,370
1,290
539
883
840
494
171
154
165
393
315
Ref_Pb
(mg/kg)
1,227
1,010
934
936
1,977
1,310
1,027
1,770
1,393
1,140
987
631
503
779
794
836
778
466
501
342
179
166
147
245
166
217
364
193
294
294
PACE_Pb
(mg/kg)
1,100
940
870
920
1,900
1,200
1,000
1,500
1,400
1,380
490
630
370
870
720
880
560
550
470
390
190
190
150
220
180
240
360
230
270
310
A-17
-------
Table A-16. Lead Analysis Results for PACE-AAS and Reference Laboratories (Part 2, Mill
Creek Site)
Sample No.
MCHD1
MCHD2
MCHD3
MCHD4
MCHD5
MCHD6
MCHD7
MCHD8
MCHD9
MCHD10
MCMD1
MCMD2
MCMD3
MCMD4
MCMD5
MCMD6
MCMD7
MCMD8
MCMD9
MCMD10
MCLD1
MCLD2
MCLD3
MCLD4
MCLD5
MCLD6
MCLD7
MCLD8
MCLD9
MCLD10
MSE_Pb
(mg/kg)
391
582
384
312
520
484
355
376
388
332
229
350
267
256
442
265
362
349
484
413
298
316
291
331
264
342
350
432
540
499
C_IC_Pb
(mg/kg)
391
415
521
361
396
235
283
424
336
495
306
279
275
330
497
C_AA_Pb
(mg/kg)
417
452
556
388
423
240
294
441
362
529
319
288
297
340
530
Ref_Pb
(mg/kg)
400
582
417
312
532
484
368
376
402
332
235
350
281
256
436
265
353
349
503
413
308
316
286
331
279
342
340
432
522
499
PACE_Pb
(mg/kg)
390
600
410
310
490
400
320
390
400
390
230
400
280
270
420
270
340
390
510
440
300
290
300
330
280
360
330
440
560
510
A-18
-------
Table A-17. Zinc Analysis Results for PACE-AAS and Reference Laboratories (Part 1 Silver
Bow Site)
Sample No.
SBHD1
SBHD2
SBHD3
SBHD4
SBHD5
SBHD6
SBHD7
SBHD8
SBHD9
SBHD10
SBMD1
SBMD2
SBMD3
SBMD4
SBMD5
SBMD6
SBMD7
SBMD8
SBMD9
SBMD10
SBLD1
SBLD2
SBLD3
SBLD4
SBLD5
SBLD6
SBLD7
SBLD8
SBLD9
SBLD10
MSE_Zn
(mg/kg)
7,410
3,830
6,750
3,620
14,800
5,530
5,440
5,870
4,710
7,160
1,270
3,440
1,410
3,760
4,080
2,550
2,810
1,300
2,020
1,660
426
327
310
404
343
413
543
363
325
420
C_IC_Zn
(mg/kg)
8,590
6,540
16,600
5,240
4,510
2,040
2,810
3,390
2,640
2,040
396
343
341
700
429
C_AA_Zn
(mg/kg)
9,340
7,130
18,000
5,810
5,000
2,190
3,040
3,860
2,950
2,230
442
381
375
773
473
Ref_Zn
(mg/kg)
8,447
3,830
6,807
3,620
16,467
5,530
5,497
5,870
4,740
7,160
1,833
3,440
2,420
3,760
3,777
2,550
2,800
1,300
2,097
1,660
421
327
345
404
353
413
672
363
409
420
PACE_Zn
(mg/kg)
8,100
4,400
6,800
4,400
16,000
6,000
5,700
6,300
5,100
7,600
1,400
2,500
1,100
3,600
4,300
2,100
2,100
1,500
1,700
1,700
390
330
310
380
310
410
550
230
330
420
A-19
-------
Table A-18. Zinc Analysis Results for PACE-AAS and Reference Laboratories (Part 2, Mill
Creek Site)
Sample No.
MCHD1
MCHD2
MCHD3
MCHD4
MCHD5
MCHD6
MCHD7
MCHD8
MCHD9
MCHD10
MCMD1
MCMD2
MCMD3
MCMD4
MCMD5
MCMD6
MCMD7
MCMD8
MCMD9
MCMD10
MCLD1
MCLD2
MCLD3
MCLD4
MCLD5
MCLD6
MCLD7
MCLD8
MCLD9
MCLD10
MSE_Zn
(mg/kg)
609
525
577
662
640
474
538
669
4,080
430
388
447
406
387
722
596
746
657
858
698
437
543
587
529
591
687
755
642
654
994
C_IC_Zn
(mg/kg)
640
623
611
588
4,130
417
441
686
671
887
468
541
651
751
610
C_AA_Zn
(mg/kg)
141
137
673
657
4,600
462
483
739
741
1,000
517
586
727
805
668
Ref_Zn
(mg/kg)
463
525
446
662
641
474
594
669
4,270
430
422
447
443
387
716
596
719
657
915
698
474
543
571
529
656
687
770
642
644
994
PACE_Zn
(mg/kg)
549
490
540
590
500
420
470
590
4,300
490
340
440
350
360
590
540
550
610
760
660
400
450
550
480
500
650
660
640
570
890
A-20
------- |