United States
Environmental Protection
Agency
Office of Water
(4303)
4JPA Report of EPA Efforts to R
EPA-820-R-95-003
April 1995
for the Determination of Oil and
Grease and Total Petroleu
Hydrocarbons: Phase II
place Freon
-------
!
*!;;= »
:
............
I
.
-------
Report of EPA Efforts to Replace Freon
for the Determination of Oil and
Grease and Total Petroleum
Hydrocarbons: Phase II
> Printed on Recycled Paper
-------
-------
Acknowledgments
This report was prepared under the direction of William A. Telliard of the Engineering and Analysis
Division within the EPA Office of Water. This document was prepared under EPA Contract No.
68-C3-0337 by DynCorp Environmental, Inc. Contributors to this study included Interface, Inc.,
Commonwealth Technology, Global Environmental Services, Inc., ETS Analytical Services, 3M
Corporation, Varian Sample Preparation Products, Horiba Instruments, EnSys, Inc., and members of the
Twin City Round Robin (TCRR) Group, including Bay West, Inc., Ecolab, Inc., Koch Refining Co., Land
O'Lakes, Inc., Legend Technical Services, Metropolitan Waste Control Commission, Minnesota
Department of Health, Minnesota Valley Testing Laboratories, Montgomery Watson, Serco Laboratories,
and Spectrum Laboratories, Inc.
Disclaimer
This report has been reviewed by the Engineering and Analysis Division, U.S. Environmental Protection
Agency, and approved for publication. Mention of company names, trade names or commercial products
does not constitute endorsement or recommendation for use.
Questions or comments regarding this report should be addressed to:
W.A. Telliard
USEPA Office of Water
Analytical Methods Staff
Mail Code 4303
401 M Street, SW
Washington, D.C. 20460
202/260-7120
Requests for additional copies should be directed to:
U.S. EPA NCEPI
11029 Kenwood Road
Cincinnati, OH 45242
513/489-8190
-------
-------
TABLE OF CONTENTS
Executive Summary 1
Section 1 Background 3
Section 2 Phase II Study Design 5
2.1 Study objectives 5
2.2 Sample Source Selection 5
2.3 Analytical Study Design 5
Section 3 Field Sampling 9
3.1 Sample Source Selection 9
3.2 Sample Collection and Handling Activities 9
Section 4 Data Validation and Statistical Analysis 11
4.1 Data Validation 11
4.2 Statistical Analysis 12
Section 5 Results 15
5.1 Separatory Funnel Extraction 15
5.2 Solid Phase Extraction (SPE) Using Disks 16
5.3 Solid Phase Extraction (SPE) Using Cartridges 17
5.4 Graphical Presentation of the Solvent-to-Freon Ratios 18
5.5 Graphical Presentation of the Solvent-to-Hexane Ratios 19
5.6 Graphical Presentation of RMSD Versus Acceptance Limit Results 19
Section 6 Phase II Study Discussion and Conclusions 41
6.1 Separatory Funnel Extraction 41
6.2 Solid Phase Extraction 42
Section 7 Follow-up and Future Freon Replacement Study Activities 43
7.1 Method Validation 43
7.2 Future Studies 45
Appendix A Site Summary
Appendix B Data Summary
Appendix C Statistical Techniques
-------
LIST OF EXHIBITS AND TABLES
Exhibits
Exhibit 1 Summary Statistics for Alternative Solvents in the Determination of Oil and Grease by
Separatory Funnel Extraction
Exhibit 2 Summary Statistics for Alternative Solvents in the Determination of TPH by Separatory
Funnel Extraction
Exhibit 3 Summary Statistics for Alternative Techniques in the Determination of Oil and Grease by
Solid Phase Extraction Disks - Comparison to Freon Separatory Funnel Extraction
Exhibit 4 Summary Statistics for Alternative Techniques in the Determination of Oil and Grease by
Solid Phase Extraction Disks - Comparison to n-hexane Separatory Funnel Extraction
Exhibit 5 Summary Statistics for Alternative Techniques in the Determination of Oil and Grease by
Solid Phase Extraction Cartridges - Comparison to Freon Separatory Funnel Extraction
Exhibit 6 Summary Statistics for Alternative Techniques in the Determination of TPH by Solid
Phase Extraction Cartridges - Comparison to Freon Separatory Funnel Extraction
Exhibit 7 Summary Statistics for Alternative Techniques in the Determination of Oil and Grease by
Solid Phase Extraction Cartridges - Comparison to n-hexane Separatory Funnel Extraction
Exhibit 8 Summary Statistics for Alternative Techniques in the Determination of TPH by Solid
Phase Extraction Cartridges - Comparison to n-hexane Separatory Funnel Extraction
Exhibit 9 Mean Solvent-to-Freon Separatory Funnel Extraction Ratio for Oil & Grease
Determinations
Exhibit 10 Mean Solvent-to-Freon Separatory Funnel Extraction Ratio for TPH Determinations
Exhibit 11 Mean Solvent-to-Hexane Separatory Funnel Extraction Ratio for Oil & Grease
Determinations
Exhibit 12 Normalized Root Mean Square Deviations - Oil & Grease Determinations by Separatory
Funnel Extraction
Exhibit 13 Normalized Root Mean Square Deviations - TPH Determinations by Separatory Funnel
Extraction
Exhibit 14 Normalized Root Mean Square Deviations - Oil & Grease Determinations by Solid Phase
Extraction Disks Relative to Freon-113 Separatory Funnel Extraction
Exhibit 15 Normalized Root Mean Square Deviations - Oil & Grease Determinations by Solid Phase
Extraction Disks Relative to Hexane Separatory Funnel Extraction
11
-------
Exhibit 16
Normalized Root Mean Square Deviations - Oil & Grease Determinations by Solid Phase
Extraction Cartridges Relative to Freon-113 Separatory Funnel Extraction
Exhibit 17 Normalized Root Mean Square Deviations - TPH Determinations by Solid Phase
Extraction Cartridges Relative to Freon-113 Separatory Funnel Extraction
Exhibit 18
Normalized Root Mean Square Deviations - Oil & Grease Determinations by Solid Phase
Extraction Cartridges Relative to Hexane Separatory Funnel Extraction
Exhibit 19 Normalized Root Mean Square Deviations - TPH Determinations by Solid Phase
Extraction Cartridges Relative to Hexane Separatory Funnel Extraction
Tables
Table 1
Student's t test on Difference Between n-Hexane and Cyclohexane in Percent Recoveries
of Oil and Grease
Table 2 Student's t test on Difference Between n-Hexane and Cyclohexane in Percent Recoveries
of TPH
Table 3 Student's t test on Difference Between Freon-113 and n-Hexane in Percent Recoveries of
Oil and Grease in OPR Samples
in
-------
-------
EXECUTIVE SUMMARY
In support of the Montreal Protocol on Substances that Deplete the Ozone Layer and in order to
meet the chlorofluorocarbon (CFC) phaseout requirements of the Clean Air Act Amendments of 1990,
the Environmental Protection Agency (EPA) initiated a multiphase study to determine a suitable
replacement solvent for Freon-113, a class I CFC used in several EPA wastewater and solid waste
methods for the determination of oil and grease and petroleum hydrocarbons. In Phase I of the Freon
Replacement Study, five solvents were evaluated for separatory funnel extraction and gravimetric
determination of oil and grease in aqueous samples. In addition, alternative techniques that included
sonication extraction, solid phase extraction (SPE), and a non-dispersive infra-red (NDIR) technique
were evaluated.
Conclusions from the Phase I study were used to narrow the list of alternative solvents to be
considered in Phase II to n-hexane and cyclohexane. These solvents were evaluated for separatory
funnel extraction and gravimetric determination of both oil and grease and total petroleum
hydrocarbons (TPH) in aqueous samples. Triplicate analyses were performed for each of the solvents
tested (i.e. Freon-113, n-hexane, and cyclohexane) on each of 34 samples from a combination of in-
process and effluent waste streams collected from 25 facilities encompassing 16 different industrial
categories. The objectives of Phase II were to find the alternative solvent that produced results closest
to the results produced by Freon-113 and to develop an analytical method that incorporated this
extraction solvent.
In addition to studies of alternative solvents, solid phase disk extraction, solid phase cartridge
extraction (also known as solid phase column extraction), non-dispersive infra-red spectroscopy, and
immunoassay were voluntarily evaluated by vendors of the products using splits of each sample
collected as part of the Phase II study.
Equivalency to Freon-113 separatory funnel extraction was established by generating Acceptance
Limits for the root mean square deviations (RMSDs) of the results on the basis of three sample matrix
classifications (all samples, non-petroleum samples, and petroleum samples). Examination of the
separatory funnel extraction results demonstrated that in all three matrix categories, oil and grease and
TPH results from both n-hexane and cyclohexane extraction were not equivalent to results from Freon-
113 extraction. These findings were consistent with the Phase I study conclusion that, when all
sample matrices were collectively considered, none of the alternative solvents produced results
statistically equivalent to results produced by Freon-113. Further evaluation of the Phase II data led to
the conclusion that the results produced when using n-hexane and cyclohexane were statistically
equivalent to one another.
Therefore, the decision of which alternative solvent was best suited to replace Freon-113 was
influenced by practical analytical considerations, of which the primary factor was the difference
between the boiling points of n-hexane (69°C) and cyclohexane (81°C). Based on laboratory
comments regarding the extensive amount of time required to evaporate cyclohexane, n-hexane was
chosen to replace Freon-113.
Evaluation of vendor data was limited to the SPE disk and cartridge extraction techniques with
gravimetric determination, and concluded that these techniques did not produce results equivalent to
results produced by separatory funnel extraction using either Freon-113 or n-hexane. NDIR and
immunoassay results were not considered in this report, since they represent completely different
determinative techniques. These results will be considered in other studies.
The final product of Phase II was Method 1664, a performance-based method that uses n-hexane
as the extraction solvent. The most significant changes in Method 1664 compared to other oil and
grease and petroleum hydrocarbons methods that use separatory funnel extraction and gravimetric
determination are 1) the use of n-hexane as the extraction solvent, 2) the use of standards of known
composition and purity, specifically hexadecane and stearic acid, as the spiking materials for QC
1
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
analyses, and 3) the introduction of extensive quality control (QC), including initial precision and
recovery analysis (DPR), ongoing precision and recovery analysis (OPR), reagent water method blanks,
and matrix spike/matrix spike duplicates (MS/MSDs). Though not specifically incorporated into
Method 1664, the use of alternative extraction and concentration techniques are allowed under the
performance-based option of this method, provided that performance specifications are met.
-------
SECTION 1
BACKGROUND
The discharge of chlorofluorocarbons (CFCs) has been shown to be a primary contributor to the
depletion of the earth's stratospheric ozone layer. The United States, as a party to the Montreal
Protocol on Substances that Deplete the Ozone Layer and as required by law under the Clean Air Act
Amendments of 1990 (CAAA), is committed to controlling and eventually phasing out CFCs. Under
both the Montreal Protocol and the CAAA, Class I CFCs will be phased out by January 1, 1996.
Freon-113 is a Class I CFC that is required for use in several U.S. Environmental Protection
Agency (EPA) wastewater and solid waste methods for the determination of oil and grease and
petroleum hydrocarbons. As part of the effort to eliminate the use of CFCs, EPA is studying the use
of alternate solvents to replace Freon-113. This effort is complicated by the fact that oil and grease is
a method defined parameter and, consequently, changing the extraction solvent could produce different
results. Because oil and grease is included in Clean Water Act effluent guidelines for 25 major
industries and is a regulated pollutant in over 10,000 National Pollutant Discharge Elimination System
(NPDES) permits as well as RCRA operating permits, any change to the analytical protocol has the
potential to affect permit compliance. For this reason, the objective of this study was to find a solvent
that would produce results comparable to results produced with Freon-113 for these analytes.
Initial efforts to find an alternative solvent to Freon-113 were conducted by the Office of Research
and Development's Environmental Monitoring Systems Laboratory in Cincinnati, Ohio (EMSL-Ci).
EMSL-Ci focused its study on MCAWW Method 413.1 (promulgated at 40 CFR Part 136), which is
used in Clean Water Act (CWA) programs to gravimetrically determine the oil and grease content of
surface and saline waters and domestic and industrial wastes. Aqueous samples, some of which were
synthetically prepared by spiking reagent water with various oils and greases and others that were
collected from industrial facilities, were analyzed using several different extraction solvents in place of
Freon-113. Results of the study, presented in the document titled A Study to Select a Suitable
Replacement Solvent for Freon-113 for the Gravimetric Determination of Oil and Grease, by F.K.
Kawahara, October 2, 1991, suggested the use of an 80/20 mixture of n-hexane and methyl tertiary
butyl ether (MTBE) in place of Freon-113 for oil and grease determination. Following this study, an
Office of Air and Radiation (OAR) proposal (56 FR 30519) suggested replacement of Freon-113 by
the n-hexane:MTBE mix in CWA and RCRA analytical methods for determination of oil and grease.
Based on comments submitted concerning the EMSL-Ci study results, and the need to further
investigate alternative solvents, the Office of Water and the Office of Solid Waste organized a multi-
phase Freon Replacement Study. Phase I of this study was initiated to evaluate alternative solvents
and extraction systems for equivalency across a range of real world effluent and solid waste samples
from a variety of industrial categories. This phase of the study focused on 1) the use of five
alternative solvents for gravimetric determination of oil and grease in aqueous samples by MCAWW
Method 413.1 (with modifications) and in solid samples by SW-846 Method 9071A (with
modifications) and 2) the use of alternative techniques for oil and grease determinations including
sonication extraction, solid phase extraction (SPE) using cartridges and disks, and a solvent/non-
dispersive infrared (NDIR) technique.
The results of Phase I yielded the following conclusions: n-hexane should be retained as a
possible extraction solvent for further study using gravimetric techniques; perchloroethylene should be
retained for consideration in the use of infra-red techniques; and cyclohexane should be introduced for
consideration with gravimetric techniques based on its similarity to n-hexane and because of its lower
neurotoxicity when compared to n-hexane. Results of the alternative techniques indicated that only
sonication extraction of high solids samples produced results equivalent to existing techniques that use
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Freon-113. Specifics of the study design, results, and conclusions can be found in the Preliminary
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease, September 1993.
Phase II of the Freon Replacement Study is a continuation of these efforts to determine an
appropriate replacement solvent for Freon-113 and to develop a revised method for determination of
oil and grease. Based on the conclusions from Phase I of the Freon Replacement Study, Phase II was
designed to focus on the evaluation of n-hexane and cyclohexane as extraction solvents in the
gravimetric determination of oil and grease in aqueous samples by Method 413.1 (with modifications).
In addition, gravimetric determination of petroleum hydrocarbons was included. The remainder of this
document is a report on this study.
-------
SECTION 2
PHASE II STUDY DESIGN
The second phase of the Freon Replacement Study evaluated n-hexane, cyclohexane, and
alternative extraction and concentration techniques as possible substitutes to extraction using Freon-113
in the gravimetric determination of oil and grease and TPH in aqueous samples. The original Phase n
study design is detailed in the Draft Study Plan for Phase II of the Freon Replacement Study,
September 29, 1993. The final study design is summarized below.
2.1 Study Objectives
The purposes of the second phase of the study were to: !
Continue EPA's investigation of n-hexane as a potential replacement solvent for Freon-113 in
the gravimetric determination of oil and grease in aqueous samples
Evaluate cyclohexane as an alternative solvent
Assess n-hexane and cyclohexane for determination of TPH in aqueous samples
Evaluate alternative techniques, including solid phase extraction by both disk and cartridge,
non-dispersive infrared spectrophotometry, and immunoassay
The final objectives of the Phase II study were to use the results to choose a replacement solvent
and develop a method using this solvent. Though not evaluated in this report, as part of the
development of revised analytical procedures, validation studies to support the recommended method
revisions and quality control specifications were planned. Details of these studies are presented in the
document titled Report of the Method 1664 Validation Studies, April 1995.
2.2 Sample Source Selection
Sample matrices included in Phase II of the study represented wastewaters from a variety of
industrial categories and facilities. The Office of Water's Engineering and Analysis Division
coordinated all facility contacts and planning for the sample collection efforts. Wastewater samples
containing between 40-300 mg/L oil and grease, some from petroleum and some from non-petroleum
sources, were targeted for collection. The study focused on this concentration range to avoid the
problems associated with the comparison and evaluation of non-detect results. In order to increase the
types of matrices considered by the Agency and to better assess the .effect of different matrices on
solvent extraction performance, a concerted effort was made to collect samples from a variety of
facilities that were different from those collected during Phase I of the study. Further details are
provided in Section 3 of this report, and a site summary is presented in Appendix A.
2.3 Analytical Study Design
The study focused on evaluation of three solvents, including Freon-113, for the analysis of
aqueous samples using conventional separatory funnel extraction followed by gravimetric
determination. As with Phase I, several manufacturers of alternative extraction devices and
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
measurement techniques volunteered to analyze splits of the EPA samples by their techniques, and to
provide the results of their analyses to EPA at no cost to the Agency.
Of the data submitted by vendors, only those results from the SPE disk and cartridge analyses
were considered. The results for analyses by infra-red spectroscopy and immunoassay were not
presented in this report because they represent completely different determinative techniques. Infra-red
determinative procedures for oil and grease and TPH will be the focus of future efforts under Phase III
of the Freon Replacement Study. The immunoassay procedure may be considered for use as a
screening tool but, with the measurement limitations inherent to the procedure at present, cannot be
used as a quantitative measurement technique, since the analysis was only capable of determining "less
than" or "greater than" values. Though not included in this report, summary results of the infra-red
and immunoassay analyses can be obtained from the Sample Control Center (operated by DynCorp
Environmental Programs Division), 300 N. Lee Street, Alexandria, VA 22314, (703) 519-1140.
The analytical protocol for each concentration and extraction technique considered in Phase II is
presented below.
Separately Funnel Extraction
A total of 34 aqueous sample sets were collected and sent to a single contract laboratory for
analysis in triplicate. An individual split sample in its own container was collected for each analysis.
To compare the performances of n-hexane and cyclohexane to Freon-113 in real world sample
matrices, wastewater samples were extracted using the separatory funnel techniques described in
Method 413.1. Since n-hexane and cyclohexane are less dense than water, the specifics of the
extraction procedure in Method 413.1 were adjusted to explicitly deal with the removal of each solvent
from the separatory funnel and the removal of residual water from the extract. For determination of
TPH, the oil and grease extract residue was redissolved in n-hexane and subjected to the silica gel
cleanup procedure in Method 5520F from Standard Methods for the Examination of Water and
Wastewater, 18th ed., to adsorb polar materials.
In addition, the laboratory was required to implement more stringent quality assurance/quality
control (QA/QC) tests than those required in either Method 413.1 or Method 5520F. This QC
included performance of initial precision and recovery (IPR) analyses prior to the analysis of field
samples. IPRs consisted of the extraction, concentration, and analysis of a set of four 1-L aliquots of
reagent water that had been spiked with hexadecane and stearic acid. In order to evaluate any effects
that might result from the revised analytical procedures, all IPR analyses incorporated the
modifications that were necessitated by differences in solvent densities, as well as any other changes
that were implemented during the study.
An ongoing precision and recovery (OPR) analysis, the equivalent of a single IPR sample, was
required with each analytical batch for each alternative solvent. An analytical batch consisted of a set
of samples extracted at the same time, to a maximum of ten samples.
In addition, a reagent water method blank was analyzed with each IPR set and with each analytical
batch for each alternative solvent. These reagent water blanks were run through the entire extraction
and analysis procedure by which the samples were run. The analytical protocol required that the
concentration of oil and grease in method blanks not exceed 5 mg/L and, if contamination above this
level was detected in any method blank, the laboratory was required to isolate and remove the source
of contamination.
Multiple aliquots of each sample were collected to accommodate the numerous analyses required.
For each sample, the aliquots were split from a homogenized sample and, to the extent practicable,
contained identical concentrations of oil and grease. Within each of the three different solvent
procedures and two modified methods (413.1 and 5520F), it was expected that the relative standard
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
deviation of the triplicate measurements would be less than 10 percent for those results at or above 25
mg/L and less than 20 percent for those results less than 25 mg/L.
Several procedures that were incorporated into the analytical requirements evolved as the study
progressed. These included the routine use of sodium sulfate to ensure that all residual water was
removed from the extraction solvent, the use of hexadecane and stearic acid as spiking standards for
the QC analyses, and the protocol for the use of silica gel in samples containing greater than 100 mg
of extractable material.
In current oil and grease methods, such as Methods 413.1 and 5520B, sodium sulfate use had been
limited to breaking up emulsions. With the use of extraction solvents that are lighter than water,
however, residual water from the bottom layer had a tendency to mix with the extraction solvent when
the upper n-hexane or cyclohexane phase was being drained from the separatory funnel. To remove
these trace amounts of water, it was recommended that the extraction solvent be drained through
granular, anhydrous sodium sulfate on filter paper. Phase separator paper was also shown to be
effective in the removal of water. As these or other analytical techniques evolved, they were
incorporated into the QC analyses to ensure that subsequent measurements were not adversely affected
by these changes.
Hexadecane and stearic acid were chosen as spiking standards because they are compounds of
known composition and purity and therefore allow for accurate measurement of recovery and
precision. The decision to use these compounds as spiking standards required determination of an
appropriate solvent and stock solution concentration. Various solvents and concentrations were tested
before it was determined that a hexadecane/stearic acid stock solution of 8 mg/mL total (i.e. 4 mg
hexadecane and 4 mg stearic acid) in acetone was satisfactory.
Another issue was the applicability of the silica gel extraction procedure to oil and grease (and
therefore potential TPH) concentrations in excess of 100 mg/L. The adsorptive capacity of silica gel
needed to be studied in order to determine the amount of silica gel required to adsorb increasing
concentrations of oil and grease. In addition, it was necessary to determine an appropriate cutoff for
the maximum amount of silica gel that realistically could be used.
For the purposes of this study, it was assumed (based on MCAWW Method 418.1 and Standard
Method 5520F) that 3 g of silica gel was capable of adsorbing 100 mg of adsorbable materials.
Therefore, any sample containing more than 100 mg of oil and grease that was going to be subjected
to the silica gel TPH procedure would require proportionately more silica gel to ensure that all
potentially adsorbable materials were adsorbed. It was speculated, however, that at some level
contamination from substances in the silica gel would begin to affect the TPH measurement.
In order to determine the limit at which additional amounts of silica gel might affect results, a
silica gel capacity study was conducted in which 100 mL aliquots of n-hexane spiked with different
amounts of stearic acid were subjected to the silica gel procedure. With increasing spike amounts (20
mg, 200 mg, and 2,000 mg), proportionately more silica gel (3 g, 6 g, and 60 g, respectively) was
added. Blanks comprised of n-hexane with these corresponding amounts of silica gel were also tested.
Each of the three stearic acid/silica gel combinations were prepared in triplicate. As part of this
evaluation of the capacity of silica gel, the conditions for activating silica gel were revised from the
existing Method 5520F protocol, which specified drying at 110°C for 24 hours, to drying at 200-
250°C for 24 hours, to ensure that activation was complete.
Analysis of the 20 mg stearic acid/3 g silica gel and 200 mg stearic acid/6 g silica gel
combinations produced non-detect results, thereby demonstrating that at these levels silica gel adsorbed
the stearic acid. Analysis of the 2,000 mg stearic acid/60 g silica gel combination resulted in an
average value of 9 mg/L, indicating that either the silica gel did not adsorb all of the stearic acid or
substances in the silica gel itself were contributing to contamination.
From these results, it was estimated that the maximum amount of silica gel that could be used
before potential contamination or adsorptive capacity would be of concern was 30 g. To confirm that
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
this was an appropriate choice, the procedure described above was performed on an n-hexane aliquot
spiked with 1,000 mg of stearic acid and 30 g of silica gel. Results were below the 5 mg/L detection
limit. Consequently, the maximum amount of material in a sample that could be subjected to the
silica gel procedure was limited to 1,000 mg. For those instances in which the oil and grease
concentrations exceeded 1,000 mg/L, it was necessary to split off a portion of the extract containing no
more than 1,000 mg of oil and grease for TPH determination.
Though hexadecane was not spiked into the solvent for these tests to determine if the increase in
silica gel would affect hexadecane recoveries, OPRs, which by method protocol were spiked with
hexadecane and stearic acid, were prepared using the maximum amount of silica gel that had been
required in each associated sample batch. All OPRs generated acceptable recoveries, thereby
demonstrating that increasing amounts of silica gel did not impair hexadecane recoveries.
Solid Phase Extraction (SPE)
The performance of solid phase extraction (SPE) techniques was evaluated by two manufacturers
of SPE devices, 3M Corporation and Varian Sample Preparation Products. The SPE techniques tested
have the advantage of using significantly less solvent than conventional separatory funnel extraction
techniques and, by nature of the mechanics of the technique, minimize emulsions.
The 3M Corporation extracted 29 aqueous samples for oil and grease using Empore SPE disks,
with n-hexane and cyclohexane as elution solvents, and determined results gravimetrically. The 3M
Corporation did not test samples for TPH as part of the Phase II study. Each sample was analyzed
using either 47 mm SPE disks (20 samples) or 90 mm SPE disks (9 samples), the larger of the two
being used for samples with high concentrations of particulates or matrices that might slow the
extraction process. Each sample was analyzed in triplicate. As with the separatory funnel analysis, an
individual split sample in its own container was supplied for each analysis.
Varian Sample Preparation Products extracted 33 aqueous samples with EnvirElute SPE cartridges
using n-hexane, cyclohexane, methylene chloride, and pentane as elution solvents for gravimetric
determination of both oil and grease and TPH. As with the separatory funnel analysis and SPE disk
extractions, each of the samples was analyzed in triplicate.
-------
SECTION 3
FIELD SAMPLING
3.1 Sample Source Selection
The Phase II study design required collection of a wide variety of sample matrices in order to
better assess the effect of sample type on solvent extraction performance. Both effluent and in-process
wastewater samples were collected from a variety of facilities. When selecting these sites, various
factors were considered and included the following:
The expected range of oil and grease concentrations, with a target range of 40 - 300 mg/L
The goal of collecting both petroleum and non-petroleum based matrices
Industrial categories with existing oil and grease effluent limitations
Waste streams with known analytical interference problems, such as industrial laundries
Accessibility of waste streams
Representatives of candidate facilities were initially contacted to request cooperation, verify plant
and waste stream characteristics, and to coordinate sampling schedules. Facilities volunteering to
participate were selected from EPA Regions I, II and III to minimize the costs of travel and equipment
transportation. A summary of the wastewater sources is presented in Appendix A, which shows the
facility types, industrial categories, and waste streams included in the study.
3.2 Sample Collection and Handling Activities
Field crews comprised of personnel from the Office of Water's Engineering and Analysis Division
(BAD) and/or from EAD's Sample Control Center (SCC, operated by DynCorp Environmental)
traveled to the selected sites and collected samples during one-day episodes. As with Phase I, the
purpose of this sampling effort was to evaluate alternative extraction solvents and techniques, and not
to characterize selected wastewaters for regulation development. Thus, short-term grab sampling of
the selected sources was deemed sufficient to meet the requirements of the study.
Most of the industrial facilities selected for sampling were indirect dischargers to Publicly Owned
Treatment Works, since these industrial facilities generated primary effluents that were more likely to
contain oil and grease within the desired concentration range of this study. In some instances,
untreated process waste streams were collected for analysis. In other instances, samples were collected
as mixtures of treated and untreated streams to ensure that the desired oil and grease levels were
obtained.
At each site, sample material was collected in a clean polyethylene barrel either by peristaltic
pump or by hand with a clean polyethylene beaker. Transfer of material was minimized in order to
prevent loss of extractable material during the collection process. To assure homogeneity, wastewater
in the barrel was mixed by stirring with a polyethylene paddle while sample was siphoned from the
center portion of the barrel directly into individual sample containers. Sample containers were unused,
pre-cleaned, 1-liter, wide-mouth, clear glass bottles with PTFE-lined caps.
Each sample set collected at a facility included approximately sixty 1-liter aliquots. Aliquots were
preserved on-site with HC1 (1:1) to a pH of 2. Sample bottles were cooled with wet ice prior to
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
shipment to assigned laboratories. Each sample bottle was labeled prior to collection with a unique
SCC sample number and identifying information, including source location, collection date, and
preservatives used. Pre-assigned sample numbers were the primary method of identifying and tracking
samples to ensure proper control. SCC Traffic Reports were completed at each site, and accompanied
each shipment to the receiving laboratory. Copies of these reports were used by SCC field and office
personnel for tracking purposes.
Information regarding site-specific activities were recorded on field log sheets for each sampling
episode, and included SCC sample numbers, collection date and time, description'of sample location,
sample data, and preservatives used. Field personnel double-checked all labels and Traffic Reports to
ensure accuracy and consistency. Communications were maintained among SCC field and office
personnel and the receiving laboratories for sample tracking. :
10
-------
SECTION 4
DATA VALIDATION AND STATISTICAL ANALYSIS
This section describes the approaches to evaluation of the results from those analytical techniques
specified in Section 2. Data validation, as detailed in Section 4.1, was straightforward. Statistical
analysis, as detailed in Section 4.2, included preliminary screening of data to remove outliers. Several
statistical evaluations were considered and are summarized below.
One measure of comparison, the "solvent-to-Freon ratio", indicates the amount of oil and grease or
TPH extracted by a given solvent or technique relative to the amount extracted by separatory funnel
Freon-113 extraction. A limitation of this measurement is evident when this ratio is averaged across
samples, as is required to collectively assess the data. In the process of averaging, ratios above and
below 1.00 offset one another to yield an average ratio biased toward 1.00, even when all of the
individual sample ratios are significantly different than one, thereby generating a misleading indication
of similarity to Freon-113 results.
A better indicator of similarity, though still not the most comprehensive, is the measurement of
the median absolute deviation of the result obtained using the alternate solvent or technique from the
result obtained using separatory funnel extraction with Freon-113. The median absolute deviation in a
data set is that absolute deviation value below and above which 50% of the deviations fall.
The primary measure of agreement between alternative solvents or techniques and Freon-113 used
in this study is the "normalized root mean square deviation" (RMSD). Analysis for this parameter was
complicated by the non-Gaussian nature of the error distribution of the data, which necessitated
logarithmic transformation of the data prior to performing this calculation. A more inclusive
measurement than those previously noted, the RMSD accounts for the variability within the triplicate
results of individual samples as well as the variability across samples (i.e., intra- and inter- sample
variability, respectively).
The statistics cited above were also used to compare SPE results to the results produced by
separatory funnel n-hexane extraction.
4.1 Data Validation
All data from Phase II of the study were submitted to the U.S. Environmental Protection Agency
(EPA) by May 1994. The contract laboratory and other study participants submitted data in computer-
readable or hardcopy formats. The computer-readable result files were verified against the hardcopy
data, then loaded into the study database. Results were entered into the study database manually if no
computer-readable data were provided. Hardcopy data, including all calculations submitted by the
contract laboratory, were verified from the bench sheets provided.
A few of the samples containing low concentrations of oil and grease andfor TPH yielded negative
results or results below the nominal detection limit of Method 413.1 for aqueous samples (5 mg/L).
The negative numbers may have resulted from either the subtraction of blanks containing
concentrations larger than those in the associated samples or from measurement error in the tare
weights of the sample bottles (which weighed significantly more than the oil and grease residue, i.e.,
grams vs. milligrams, respectively). As a general rule, data results at or below the detection limit of 5
mg/L were set to 0.5 times the detection limit (i.e., 2.5 mg/L) for statistical analysis.
11
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
4.2 Statistical Analysis
Preliminary statistical analyses of gravimetric results from Phase II separatory funnel extraction
and gravimetric determination were presented by the Office of Water (OW) at EPA's 17th Annual
Conference on Pollutants in the Environment in Norfolk, VA on May 3, 1994. This report includes
revised statistical analysis of those results as well as statistical analysis of the results generated by
participating vendors using solid phase extraction (SPE) techniques. A table containing all results for
the study in unreduced form is provided in Appendix B.
Outlier Screening
Prior to calculating the descriptive statistics (solvent-to-Freon ratios, solvent-to-hexane ratios and
median absolute deviations), individual ratios were screened for outliers using a robust quantile method
based on quartiles and the interquartile distance, as described in Appendix C. Ratios identified as
outliers by this method were omitted from the remainder of the statistical analyses.
This screen estimates the limits above and below which lie one percent of the data, assuming the
data are normally distributed. For the data sets evaluated in this study, the percentage of data points
falling outside of the limits ranged from 3.4% to 15.3%. Most often, those data sets with higher
percentages of outliers were associated with either of two conditions: values near or below the
detection limit, where a small difference in concentration can result in a large difference in relative
recovery; or highly variable data results from a particular solvent extraction, as was the case for
methylene chloride/SPE cartridge analysis. Only those data points for a particular data set that were
outside of the limits were removed. Data used in generating RMSDs was not subjected to this screen
since the natural logarithm transformation of the data adequately reduced the effects of potential
outliers on statistical analyses.
Data Stratification
Initial statistical analyses of all oil and grease and TPH results within each extraction technique
yielded no solvents that were within the respective RMSD Acceptance Limits for Freon-113
equivalency. Further stratification of the study database was undertaken to determine whether results
equivalent to Freon-113 could be achieved by testing subclasses of the data. Because it was expected
that oil and grease of mineral origin (petroleum) might behave differently than oil and grease of
biological origin (animal or vegetable), the data were stratified into "petroleum" and "non-petroleum"
samples, respectively. For example, the effluent from a refinery was categorized as petroleum,
whereas the effluent from a meat packing plant was categorized as non-petroleum. For the purposes
of this report, data have been assessed in terms of three sample strata, which consist of 1) all samples,
2) petroleum samples, and 3) non-petroleum samples.
Solvent-to-Freon Ratio
A solvent-to-Freon ratio was computed to allow comparison of the average amount of oil and
grease or TPH extracted by various solvents and measurement techniques with the amount extracted by
the separatory funnel extraction technique using Freon-113. The solvent-to-Freon ratio for each
sample was formed by dividing the mean of the triplicate results produced with the alternative solvent
or technique by the mean of the triplicate results produced with the separatory funnel Freon-113
technique. The mean, standard deviation (SD), and relative standard deviation (RSD) of the solvent-
to-Freon ratios were calculated across all samples in each of the three data sets representing an
12
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase If
alternative solvent or technique, i.e., separatory funnel, SPE disk, and SPE cartridge (Exhibits 1
through 8, presented in Section 5).
Because averaging gives no indication of the variability of the data, the solvent-to-Freon ratio is a
less powerful measure of the agreement between an alternative solvent (or technique) and Freon-113
than the median absolute deviation and the normalized RMSD described below. Therefore, mean and
standard deviation values of the solvent-to-Freon ratio were limited to use as an aid in describing the
distribution of the data. The median absolute deviation and the normalized RMSD were used as the
main criteria of similarity, with more emphasis being placed on the RMSD.
Median Absolute Deviation
The median absolute deviation provides a means of gauging the difference of the alternative
solvent or technique results from separatory funnel Freon-113 results. The absolute deviation for each
sample is determined according the formula: 100 x | solvent concentration/Freon concentration - 11.
The median absolute deviation in a data set is that value below and above which 50% of the
deviations fall. Unlike the solvent-to-Freon ratio, this value is an absolute difference. Therefore,
when averaged across all samples, alternative solvent or technique results that are greater than or less
than Freon-113 results do not cancel each other. Though the median absolute deviation does provide
an indication of central tendency for how different the alternative solvent results are in relation to the
results from Freon-113, it does not consider the range of differences across all samples, and does not
show how variable this measurement of difference is across all samples. Median absolute deviations
were calculated across all samples for each of the three data sets and are presented in Exhibits 1
through 8.
Logarithmic Data Transformation
Analytical data often exhibit variability proportional to the magnitude of the signal over the
calibration range of the measuring technique (i.e., as the concentration increases, the standard deviation
will increase proportionally). Concentration dependent standard deviation was demonstrated in this
study for Freon-113, n-hexane, and cyclohexane separatory funnel extraction (results are shown in
Attachment 1-A of Appendix C). In order to evaluate data using Analysis of Variance (ANOVA)
techniques, the results of which were required for the RMSD calculations described below, it was
necessary to transform the data to eliminate the associated heteroscedasticity (non-Gaussian distribution
of standard deviations) reflected by the concentration dependent standard deviations. This was
accomplished by applying a natural log transformation using the following equation: z = In (x), where
x is the concentration (in mg/L). Results are shown in Attachment 1-B of Appendix C.
As stated in Section 4.1, data results at or below the detection limit of 5 mg/L were set to 0.5
times the detection limit (i.e., 2.5 mg/L). The transformed data (z) were then subjected to ANOVA.
Root Mean Square Deviation (RMSD)
The primary measure of similarity used to compare each of the other solvents and techniques to
separatory funnel Freon-113 extraction was the RMSD. For each technique, the RMSD represents the
standard deviation of the differences between the alternative solvent-determined concentrations for
each sample and the Freon-113 separatory funnel extraction determinations. A smaller RMSD
indicates better agreement with the Freon-113 separatory funnel procedure.
In performing this statistical analysis, the data were transformed using the logarithmic equation
listed above and were subjected to ANOVA. The triplicate concentration results for each sample and
solvent or technique were then averaged. For each alternative procedure, the RMSD was calculated by
taking the difference of the Freon-113 averages from those of the alternative procedure, squaring the
13
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
differences, averaging these squared differences across all samples, and then taking the square root of
those averages. Details of this calculation are presented in Appendix C. The resulting RMSD
accounts for variations in the alternative solvent/technique results regardless of whether they are above
or below the results from the Freon-113 extractions.
The RMSD values were normalized by dividing each RMSD by the square root of the residual
error estimate based on the replicate variability (see Appendix C for details). Normalization was
implemented to correct for any background variability produced by the procedure itself. This ensures
that the variability measured is isolated to the use of a particular solvent and/or technique.
To determine the significance of the RMSD results, a 95 percent Acceptance Limit was calculated
from each of the three data sets (separatory funnel extraction, 3M SPE disks, Varian SPE cartridges)
for the three sample strata (all samples, petroleum samples, and non-petroleum samples). The
Acceptance Limit is the 95th percentile of the normalized RMSD that would be observed if the two
methods under comparison were exactly equivalent (i.e., if Freon-113 were compared to itself 100
times, 95 times the RMSD would be at or below the Acceptance Limit). Solvents or techniques with
RMSDs less than or equal to the associated Acceptance Limit are not statistically different from Freon-
113 extraction with the approved technique. Normalized RMSD results and Acceptance Limits for
each of the alternative solvent and measurement techniques evaluated are presented in Exhibits 1
through 8.
As was stated in the introduction to Section 4, the statistical analyses are described above in the
context of comparing one or more techniques or solvents to the separatory funnel Freon-113
determination of oil and grease and TPH. The statistical analyses described above were also used to
compare the SPE disk and cartridge techniques to separatory funnel extraction using n-hexane (i.e., all
statements made above for Freon-113 apply to n-hexane).
14
-------
SECTION 5
RESULTS
The statistical results presented in Exhibits 1 through 8 are summarized in Sections 5.1 through
5.3 below. As explained in Section 4, the median absolute deviation and root mean square deviation
(RMSD) were used as the main criteria of equivalence between the results obtained with established
techniques (separatory funnel Freon-113 extraction and gravimetric determination) and all other
solvents and techniques. Other descriptive statistics (mean, standard deviation, and relative standard
deviation of solvent-to-Freon and solvent-to-n-hexane ratios) were provided in this report to
demonstrate the general distribution of the data.
Exhibits 1 through 8 are arrayed to provide a comparison of each alternative solvent or technique
for the three sample strata, i.e., 1) all samples, 2) petroleum samples, and 3) non-petroleum samples.
Solvents or techniques associated with results that produced an RMSD within the Acceptance Limit for
separatory funnel extraction using Freon-113 were not considered to be statistically different from
Freon-113. As was stated earlier, SPE disk and cartridge techniques were also compared to separatory
funnel n-hexane extraction.
5.1 Separatory Funnel Extraction
Oil and Grease
Results of the statistical analysis of data for separatory funnel extraction and gravimetric
determination of oil and grease are presented in Exhibit 1. These results were based on the analysis of
18 petroleum samples and 15 non-petroleum samples, for a total of 33 samples (one of the 34 samples
collected was not successfully analyzed due to the formation of severe emulsions), using Freon-113, n-
hexane, and cyclohexane as extraction solvents. In some cases the number of samples listed in Exhibit
1 were less than 18 and 15, respectively, due to the exclusion of sample results on a solvent-by-
stratum-specific basis by the outlier screening process.
Mean solvent-to-Freon ratios ranged from 0.74 (for petroleum samples extracted with n-hexane)
to 0.85 (for non-petroleum samples extracted with cyclohexane). Relative standard deviations of the
solvent-to-Freon ratios were between 25% and 31%; median absolute deviations ranged from 17.6% to
25.6%. The similar values in Exhibit 1 suggest that, on average, n-hexane and cyclohexane extracted
less oil and grease than Freon-113 and produced results which vary by approximately 20% from those
produced by Freon-113.
RMSD values demonstrate that when all samples were examined either as a group or on a
petroleum or non-petroleum sample basis, neither of the solvents tested yielded results within the
Acceptance Limit.
To determine if either alternative solvent performed more similarly to Freon, a paired t-test across
all samples was used to compare the n-hexane-to-Freon ratio with the cyclohexane-to-Freon ratio for
the analysis of oil and grease. The results, shown in Table 1, indicate that there was no statistical
difference in the performance of the two solvents.
Total Petroleum Hydrocarbons (TPH)
Results of the statistical analysis of data for separatory funnel extraction and gravimetric
determination of TPH are presented in Exhibit 2. As with oil and grease determination, the results
15
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
were based on analysis of 18 petroleum and 15 non-petroleum samples (though the number of samples
listed may be less due to outlier screening).
Mean solvent-to-Freon ratios ranged from 0.80 (for all samples extracted with n-hexane) to 1.40
(for non-petroleum samples extracted with cyclohexane). Relative standard deviations of the solvent-
to-Freon ratios ranged from 21% to 70%, notably wider than that for oil and grease. Median absolute
deviations ranged from 14.1% to 40.4%. The data in Exhibit 2 suggest that n-hexane and cyclohexane
performed similarly for petroleum samples, but that cyclohexane produced more deviate results than n-
hexane in non-petroleum samples, though these differences were not statistically significant.
RMSD values demonstrate that when all samples were examined either as a group, or on a
petroleum or non-petroleum sample basis, neither of the solvents tested yielded results within the
Acceptance Limit.
As with oil and grease analyses, a paired t-test across all samples was used to compare the
n-hexane-to-Freon ratio with the cyclohexane-to-Freon ratio for the analysis of TPH. The results,
shown in Table 2, indicate that there was no statistical difference in the performance of the two
solvents.
5.2 Solid Phase Extraction (SPE) Using Disks
Exhibits 3 and 4 present the results of the statistical analysis of data generated by 3M Corporation
for oil and grease determination using solid phase extraction (SPE) disks and gravimetric
determination. As was noted in Section 2.3, n-hexane and cyclohexane were evaluated with the SPE
disk technique; Freon-113 was not evaluated with the SPE disks in this study. Analysis included
determination of oil and grease only; TPH was not determined as part of the Phase II study. All
results generated using SPE disk techniques were compared to the performance of separatory funnel
extraction using both Freon-113 and n-hexane. SPE disk results were not compared to those from SPE
cartridge techniques.
Statistical results were based on the analysis of 14 petroleum and 14 non-petroleum samples. In
some cases the number of samples listed in Exhibits 3 and 4 are less than 14 due to the exclusion of
sample results on a solvent-by-stratum-specific basis by the outlier screening process.
Comparison to Separatory Funnel Extraction Using Freon-113
Mean solvent-to-Freon ratios in Exhibit 3 for the SPE disk data using n-hexane and cyclohexane
as elution solvents were all below 1.0, ranging from 0.82 (for petroleum samples extracted with disk +
n-hexane) to 0.96 (for all samples extracted with disk + cyclohexane). This indicates that, on average,
SPE disks extracted less oil and grease than separatory funnel extraction using Freon-113. RSDs of
the solvent-to-Freon ratios ranged from 29% to 54%. Median absolute deviations ranged between
22.9% and 29.7%, slightly higher than those for separatory funnel extraction using the same solvents.
RMSD values in Exhibit 3 for the use of SPE disks were higher than the Acceptance Limit in all
three sample strata, indicating that the results were not equivalent to separatory funnel extraction using
Freon-113. Of the two solvents, cyclohexane performed more similarly to separatory funnel extraction
in all three strata.
Comparison to Separatory Funnel Extraction Using n-hexane
Mean solvent-to-hexane ratios in Exhibit 4 for the SPE disk data using alternative solvents were
all above 1.00, ranging from 1.05 (for non-petroleum samples extracted with disk + cyclohexane) to
1.22 (for petroleum samples extracted with disk + cyclohexane). This indicates that, on average, SPE
disks extracted more oil and grease than did separatory funnel extraction using n-hexane. RSDs of the
16
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase If
solvent-to-hexane ratios ranged from 31% to 55%. Median absolute deviations ranged between 25.5%
and 47.1%
RMSD values in Exhibit 4 for the use of SPE disks were higher than the Acceptance Limit in all
three sample strata, indicating that the results were not equivalent to separatory runnel extraction using
n-hexane. Of the two solvents, cyclohexane performed more similarly to separatory funnel extraction
in all three strata.
5.3 Solid Phase Extraction (SPE) Using Cartridges
Exhibits 5 through 8 present the results of the statistical analyses of data generated by Varian
Sample Preparation Products using SPE cartridges and gravimetric determination. As was stated in
Section 2.3, four solvents (n-hexane, cyclohexane, pentane, and methylene chloride) were evaluated
with this technique. Both oil and grease and TPH were determined. All results generated using SPE
cartridges were compared to the performance of separatory funnel extraction using both Freon-113 and
n-hexane. SPE cartridge results were not compared to those from SPE disk techniques.
Statistical results were based on the analysis of 14 petroleum and 14 non-petroleum samples. In
some cases the number of samples listed in Exhibits 5 through 8 were less than 14 due to the
exclusion of sample results on a solvent-by-stratum-specific basis from the outlier screening process.
Comparison to Separatory Funnel Extraction Using Freon-113
Oil and Grease
Mean solvent-to-Freon ratios in Exhibit 5 ranged from 0.80 (for petroleum samples extracted with
cartridge + pentane) to 1.90 (for non-petroleum samples extracted with cartridge + methylene
chloride). With the exception of methylene chloride extraction for all three sample strata, all solvent-
to-Freon ratios were below 1.00, indicating that, on average, SPE cartridges extracted less oil and
grease than separatory funnel extraction using Freon-113. RSDs of the solvent-to-Freon ratios ranged
from 18% to 66%. Median absolute deviations ranged between 15.1% and 89.9%, appreciably higher
than those for separatory funnel extraction.
RMSDs in Exhibit 5 for the SPE cartridge results were higher than the Acceptance Limits in all
three strata, indicating that results were not equivalent to Freon-113. Of the four solvents,
cyclohexane performed closest to separatory funnel extraction for oil and grease using Freon-113 for
all three sample strata.
Total Petroleum Hydrocarbons
Mean solvent-to-Freon ratios in Exhibit 6 ranged from 0.44 (for non-petroleum samples extracted
with cartridge + n-hexane) to 1.04 (for petroleum samples extracted with cartridge + methylene
chloride). With the exception of petroleum samples extracted with methylene chloride, all solvent-to-
Freon ratios were below 1.00, indicating that, on average, SPE cartridges extracted less TPH than
separatory funnel extraction using Freon-113. RSDs of the solvent-to-Freon ratios with the SPE
cartridges ranged from 41% to 65%. Median absolute deviations ranged between 18.3% and 60.0%,
higher than those for separatory funnel extraction.
For all three sample strata, RMSDs in Exhibit 6 for the SPE cartridge results were higher than the
Acceptance Limits, indicating that results were not equivalent to Freon-113. , Of the four solvents,
cyclohexane performed closest to separatory funnel extraction of TPH using Freon-113 for all samples
and non-petroleum samples, while methylene chloride performed closest for petroleum samples.
17
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Comparison to Separately Funnel Extraction Using n-hexane
Oil and Grease
Mean solvent-to-hexane ratios in Exhibit 7 ranged from 0.89 (for non-petroleum samples extracted
with cartridge -t- n-hexane) to 2.12 (for non-petroleum samples extracted with cartridge + methylene
chloride). RSDs of the solvent-to-hexane ratios ranged from 18% to 51%. Median absolute deviations
ranged between 13.5% and 133.9%. ;
RMSDs in Exhibit 7 for the SPE cartridge results were higher than the Acceptance Limits in all
three strata, indicating that results were not equivalent to separatory funnel extraction using n-hexane.
Of the four solvents, cyclohexane performed closest to separatory funnel extraction for oil and grease
using n-hexane for all samples and non-petroleum samples, while n-hexane came closest for petroleum
samples.
Total Petroleum Hydrocarbons
Mean solvent-to-hexane ratios in Exhibit 8 ranged from 0.70 (for non-petroleum samples extracted
with cartridge + n-hexane) to 1.31 (for non-petroleum samples extracted with cartridge + pentane).
RSDs of the solvent-to-hexane ratios with the SPE cartridges ranged from 38% to 89%. Median
absolute deviations ranged between 21.6% and 79.9%.
RMSDs in Exhibit 8 for the SPE cartridge results were higher than the Acceptance Limits in all
three sample categories, indicating that results were not equivalent to n-hexane. Of the four solvents,
methylene chloride performed closest to separatory funnel extraction of TPH using n-hexane for all
samples and petroleum samples, while cyclohexane performed closest for non-petroleum samples.
5.4 Graphical Presentation of the Solvent-to-Freon Ratios
Exhibit 9 summarizes on one graph the solvent-to-Freon ratios when considering all samples
collectively for the extraction of oil and grease by all techniques. In this graph, the mean solvent-to-
Freon ratio was plotted on a logarithmic scale so that reciprocal ratios were equidistant from 1.00,
regardless of whether or not the solvent results were greater than or less than the Freon-113 results
(e.g., a solvent result that is half as much as the Freon-113 result is the same distance from 1.00 as a
solvent result that is twice as much as the Freon-113 result). These conditions apply to the graphs
described below for TPH analysis and in Section 5.5 for solvent-to-hexane ratios.
Exhibit 9 shows that the solvent-to-Freon ratio ranges from approximately 0.76 to 1.69, depending
on the solvent and technique. The use of n-hexane and cyclohexane in both separatory funnel
extraction and SPE disk and cartridge techniques allowed the variables of solvent and technique to be
considered independently of one another. This graph demonstrates that separatory funnel extraction
with either n-hexane or cyclohexane extracted less oil and grease, on average, than SPE using either
disks or cartridges.
Exhibit 10 summarizes the mean solvent-to-Freon ratios for the extraction of TPH. The solvent-
to-Freon ratio ranged between 0.72 and 0.95, narrower than that for oil and grease. Since analysis for
TPH was not performed using SPE disks as part of the Phase II study, only separatory funnel
extraction and SPE cartridge extraction techniques could be compared directly for n-hexane and
cyclohexane use. Contrary to the results from oil and grease analysis, separatory funnel extraction
with either n-hexane or cyclohexane extracted more TPH, on average, than SPE cartridges.
18
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase (!
5.5 Graphical Presentation of the Solvent-to-Hexane Ratios
Exhibit 11 summarizes the solvent-to-hexane ratios for the extraction of oil and grease. The
conditions described above for the solvent-to-Freon ratio graphs apply to the graphs addressed in this
section. The solvent-to-hexane ratios for the three techniques ranged from 1.02 to 1.93, higher than
the solvent-to-Freon ratios. Because solvent-to-hexane ratios were not determined for separatory
funnel extraction (i.e., hexane and cyclohexane results were not directly compared), the variables of
solvent and technique could not be separated for separatory funnel comparison to SPE disk and
cartridge techniques. A comparable graph for TPH analysis was not prepared, since the only
alternative technique by which TPH analyses were run as part of the Phase II study was the SPE
cartridge.
5.6 Graphical Presentation of RMSD Versus Acceptance Limit Results
To provide a better understanding of the relative performance of all of the alternative solvents and
techniques evaluated, the RMSD and Acceptance Limit data are presented graphically in Exhibits 12
through 19. Separate graphs were prepared for each combination of extraction technique (i.e.,
separatory funnel, SPE disk, or SPE cartridge), oil and grease or TPH determination, and Freon-113 or
n-hexane Acceptance Limits.
In these Exhibits, the RMSD for a particular solvent or technique is represented by a solid or
hollow square (for petroleum and non-petroleum sample categories, respectively), and Acceptance
Limits are indicated by a horizontal dotted or horizontal dashed line (for petroleum and non-petroleum
sample categories, respectively). Where these RMSD squares fall within (i.e., at or below) the relevant
Acceptance Limit line, that solvent or technique is equivalent to Freon-113 separatory funnel
extraction. Where no solvents or techniques yielded results equivalent to Freon-113, no squares are
within the Acceptance Limit line.
19
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 1
Summary Statistics For Alternative Solvents in the Determination of Oil and Grease
by Separatory Funnel Extraction
All Samples
Solvent
Freon
Hexane
N
33
29
32
Mean
1.00
0.76
0.83
SD
0.20
0.22
RSD
(%)
26
27
Median
Deviation
(%)
0.0
21.8
18.4
Non-Petroleum
Solvent
Freon
Hexane
N
15
13
14
Mean
1.00
0.80
0.85
SD
-
0.21
0.26
RSD
(%)
26
31
Median
Deviation
(%)
0.0
25.3
25.6
Petroleum
Solvent
Freon
Hexane
Cyclohexane
N
18
16
18
Mean
1.00
0.74
0.81
SD
0.19
0.20
RSD
(%)
26
25
Median
Deviation
(%)
0.0
19.2
17.6
RMSD
1.22*
4.50
3.34
RMSD
1.33*
4.08
2.80
RMSD
1.30*
5.15
4.14
* Acceptance Limit
N = Number of Samples
Mean = Mean of Solvent to Freon Ratios
SD = Standard Deviation of Solvent to Freon Ratios
RSD = Relative Standard Deviation of Solvent to Freon Ratio = 100 x SD/Mean
Median Deviation = median, across samples, of 100 x I Solvent to Freon Ratio -1
RMSD = Normalized Root Mean Square Deviation of Sample x Solvent Means
20
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil land Grease and TPH: Phase ((
Exhibit 2
Summary Statistics For Alternative Solvents in the Determination of TPH
by Separately Funnel Extraction
All Samples
Solvent
Freon
Hexane
Cyclohexane
N
33
29
28
Mean
1.00
0.80
0.95
SD
0.22
0.36
RSD
(%)
28
38
Median
Deviation
(%)
0.0
20.2
21.9
Non-Petroleum
Solvent
Freon
Hexane
Cyclohexane
N
15
15
13
Mean
1.00
1.12
1.40
SD'
0.74
0.99
RSD
(%)
66
70
Median
Deviation
(%)
0.0 '
40.4
31.7
RMSD
1.22*
1.68
2.66
RMSD
1.33*
1.53
2.70
Petroleum
Solvent
Freon
Hexane
Cyclohexane
N
18
18
16
Mean
1.00
0.85
0.85
SD
0.22
0.18
RSD
(%)
26
21
Median
Deviation
(%)
0.0
16.2
14.1
RMSD
1.30*
2.61
2.47
* Acceptance Limit
N = Number of Samples
Mean = Mean of Solvent to Freon Ratios
SD = Standard Deviation of Solvent to Freon Ratios
RSD = Relative Standard Deviation of Solvent to Freon Ratio = 100 x SD/Mean
Median Deviation = median, across samples, of 100 x I Solvent to Freon Ratio - 1
RMSD = Normalized Root Mean Square Deviation of Sample x Solvent Means
21
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 3
Summary Statistics For Alternative Techniques in the Determination of Oil and Grease
by Solid Phase Extraction Disks - Comparison to Freon Separatory Funnel Extraction
All Samples
Solvent
Freon Sep. Funnel
Hexane
Cyclohexane
N
26
23
25
Mean
1.00
0.88
0.96
SD
~
0.42
0.44
RSD
(%)
~
48
46
Median
Deviation
(%)
0.0
29.7
27.2
RMSD
1.25*
6.12
5.35
Non-Petroleum
Solvent
Freon Sep. Funnel
Hexane
Cyclohexane
N
13
11
12
Mean
1.00
0.85
0.91
SD
0.41
0.49
RSD
(%)
48
54
Median
Deviation
(%)
0.0
29.7
27.2
RMSD
1.36*
6.29
6.02
Petroleum
Solvent
Freon Sep. Funnel
Hexane
Cyclohexane
N
13
11
11
Mean
1.00
0.82
0.87
SD
0.30
0.25
RSD
(%)
37
29
Median
Deviation
(%)
0.0
28.9
22.9
RMSD
1.36*
5.85
4.13
* Acceptance Limit
N = Number of Samples
Mean = Mean of Solvent to Freon Ratios
SD = Standard Deviation of Solvent to Freon Ratios
RSD = Relative Standard Deviation of Solvent to Freon Ratio = 100 x SD/Mean
Median Deviation = median, across samples, of 100 x I Solvent to Freon Ratio -1
RMSD = Normalized Root Mean Square Deviation of Sample x Solvent Means
22
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Pftase (f
Exhibit 4
Summary Statistics For Alternative Techniques in the Determination of Oil and Grease
by Solid Phase Extraction Disks - Comparison to n-hexane Separate ry Funnel Extraction
All Samples
Solvent
Hexane Sep. Funnel
Hexane
Cyclohexane
N
26
24
24
Mean
1.00
1.12
1.13
SD
0.50
0.47
RSD
(%)
45
42
Median
Deviation
(%)
0.0
38.0
28.7
RMSD
1.25*
6.22
5.29
Non-Petroleum
Solvent
Hexane Sep. Funnel
Hexane
Cyclohexane
N
13
12
12
Mean
1.00
1.08
1.05
SD
0.56
0.55
RSD
(%)
52
55
Median
Deviation
(%)
0.0
47.1
40.2
RMSD
1.36*
5.59
5.10
Petroleum
Solvent
Hexane Sep. Funnel
Hexane
Cyclohexane
N
13
12
12
Mean
1.00
1.17
1.22
SD
0.45
0.38
RSD
(%)
38
31
Median
Deviation
(%)
0.0
33.1
25.5
RMSD
1.36*
7.25
5.63
* Acceptance Limit
N = Number of Samples
Mean = Mean of Solvent to Freon Ratios
SD = Standard Deviation of Solvent to Freon Ratios
RSD = Relative Standard Deviation of Solvent to Freon Ratio = 100 x SD/Mean
Median Deviation = median, across samples, of 100 x I Solvent to Hexane Ratio - I
RMSD = Normalized Root Mean Square Deviation of Sample x Solvent Means
23
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 5
Summary Statistics For Alternative Techniques in the Determination of Oil and Grease
by Solid Phase Extraction Cartridges - Comparison to Freon Separatory Funnel Extraction
All Samples
Solvent
Freon Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
27
26
25
25
25
Mean
1.00
0.88
0.90
0.85
1.69
SD
0.48
0.37
0.34
0.64
RSD
(%)
55
41
40
38
Median
Deviation
(%)
0.0
42.0
20.4
31.0
48.7
RMSD
1.24*
5.28
3.99
4.05
5.47
Non-Petroleum
Solvent
Freon Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
14
13
13
13
13
Mean
1.00
0.91
0.94
0.89
1.90
SD
0.60
0.52
0.54
0.69
RSD
(%)
66
55
61
36
Median
Deviation
(%)
0.0
45.1
47.3
41.9
89.9
RMSD
1.33*
5.29
4.23
4.28
5.28
Petroleum
Solvent
Freon Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
13
13
12
11
11
Mean
1.00
0.85
0.88
0.80
1.33
SD
0.34
0.20
0.14
0.30
RSD
(%)
40
23
18
23
Median
Deviation
(%)
0.0
29.2
15.1
20.0
24.4
RMSD
1.34*
5.02
3.24
3.32
5.77
* Acceptance Limit
N =» Number of Samples
Mean = Mean of Solvent to Freon Ratios
SD = Standard Deviation of Solvent to Freon Ratios
RSD = Relative Standard Deviation of Solvent to Freon Ratio = 100 x SD/Mean
Median Deviation = median, across samples, of 100 x I Solvent to Freon Ratio - 1
RMSD = Normalized Root Mean Square Deviation of Sample x Solvent Means
24
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase »
Exhibit 6
Summary Statistics For Alternative Techniques in the Determination of TPH
by Solid Phase Extraction Cartridges - Comparison to Freon Separatoiry Funnel Extraction
All Samples
Solvent
Freon Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
27
17
18
18
18
Mean
1.00
0.72
0.81
0.78
0.84
SD
0.41
0.45
0.42
0.47
RSD
(%)
57
56
54
56
Median
Deviation
(%)
0.0
39.0
39.8
31.5
46.1
RMSD
1.24*
3.50
3.23
3.43
3.45
Non-Petroleum
Solvent
Freon Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
14
7
7
7
7
Mean
1.00
0.44
0.54
0.51
0.51
SD
0.23
0.29
0.33
0.28
RSD
(%)
52
54
65
55
Median
Deviation
(%)
0.0
54.3
45.6
60.0
47.4
RMSD
1.34*
3.21
2.85
3.15
3.24
Petroleum
Solvent
Freon Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
14
10
11
11
11
Mean
1.00
0.92
0.98
0.95
1.04
SD
0.39
0.45
0.39
0.46
RSD
(%)
42
46
41
44
Median
Deviation
(%)
0.0
27.5
26.0
18.3
33.1
RMSD
1.35*
4.43
4.43
4.31
4.06
*'Acceptance Limit
N = Number of Samples
Mean = Mean of Solvent to Freon Ratios
SD = Standard Deviation of Solvent to Freon Ratios !
RSD = Relative Standard Deviation of Solvent to Freon Ratio = 100 x SD/Mean
Median Deviation = median, across samples, of 100 x I Solvent to Freon Ratio -1
RMSD = Normalized Root Mean Square Deviation of Sample x Solvent Means
25
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 7
Summary Statistics For Alternative Techniques in the Determination of Oil and Grease
by Solid Phase Extraction Cartridges - Comparison to n-hexane Separatory Funnel Extraction
All Samples
Solvent
Hexane Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
27
26
25
22
24
Mean
1.00
1.02
1.07
1.04
1.93
SD
0.47
0.43
0.21
0.72
RSD
(%)
46
40
20
37
Median
Deviation
(%)
0.0
27.0
32.2
13.5
79.4
RMSD
1.24*
4.68
3.66
3.98
7.34
Non-Petroleum
Solvent
Hexane Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
14
13
13
13
13
Mean
1.00
0.89
0.94
0.90
2.12
SD
0.45
0.44
0.36
0.90
RSD
(%)
51
47
40
42
Median
Deviation
(%)
0.0
36.5
41.0
14.2
133.9
RMSD
1.33*
4.81
3.15
3.65
6.31
Petroleum
Solvent
Hexane Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
13
11
11
11
11
Mean
1.00
1.01
1.13
1.06
1.71
SD
0.27
0.30
0.19
0.34
RSD
(%)
--
27
27
18
20
Median
Deviation
(%)
0.0
21.2
20.4
13.5
75.6
RMSD
1.34*
4.01
4.80
4.77
9.60
* Acceptance Limit
N = Number of Samples
Mean = Mean of Solvent to Freon Ratios
SD = Standard Deviation of Solvent to Freon Ratios
RSD = Relative Standard Deviation of Solvent to Freon Ratio = 100 x SD/Mean
Median Deviation = median, across samples, of 100 x I Solvent to Hexane Ratio -1
RMSD = Normalized Root Mean Square Deviation of Sample x Solvent Means
26
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase \\
Exhibit 8
Summary Statistics For Alternative Techniques in the Determination of TPH
by Solid Phase Extraction Cartridges - Comparison to n-hexane Separatory Funnel Extraction
All Samples
Solvent
Hexane Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
27
18
19
19
20
Mean
1.00
1.03
1.03
1.07
1.19
SD
0.58
0.50
0.61
0.73
RSD
(%)
56
49
57
61
Median
Deviation
(%)
0.0
33.9
24.8
40.4
45.2
RMSD
1.24
2.93
2.63
2.84
2.55
Non-Petroleum
Solvent
Hexane Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
14
7
8
9
9
Mean
1.00
0.70
0.87
1.31
1.18
SD
0.47
0.50
1.16
0.99
RSD
(%)
--
67
57
89
84
Median
Deviation
(%)
0.0
28.5
21.6
79.9
71.1
RMSD
1.34*
2.65
2.30
2.63
2.32
Petroleum
Solvent
Hexane Sep. Funnel
Hexane
Cyclohexane
Pentane
Methylene Chloride
N
13
10
11
10
11
Mean
1.00
1.12
1.14
1.11
1.20
SD
0.43
0.50
0.44
0.48
RSD
(%)
38
44
40
40
Median
Deviation
(%)
0.0
33.9
; 35.1
34.6
25.1
RMSD
1.35*
3.81
3.70
3.53
3.30
* Acceptance Limit
N = Number of Samples
Mean = Mean of Solvent to Freon Ratios
SD = Standard Deviation of Solvent to Freon Ratios
RSD = Relative Standard Deviation of Solvent to Freon Ratio = 100 x SD/Mean
Median Deviation = median, across samples, of 100 x I Solvent to Hexane Ratio - 1
RMSD = Normalized Root Mean Square Deviation of Sample x Solvent Means
27
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 9
Mean Solvent-to-Freon Separatory Funnel Extraction Ratio
for Oil & Grease Determinations
MeCl-
Ideal
Ratio
Pentane
I
Cyclohexane
Hexane
0.5
A
1
Solvent-to-Freon Ratio
Sep Funnel
SPE Disk
A SPE Cartridge
28
-------
g
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 10
Mean Solvent-to-Freon Separatory Funnel Extraction Ratio
for TPH Determinations
MeCl2
Ideal
Ratio
I
Pentane
Cyclohexane
Hexane
0.5
Sep Funnel
A SPE Cartridge
1
Solvent-to-Freon Ratio
29
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 11
Mean Solvent-to-Hexane Separately Funnel Extraction Ratio
for Oil & Grease Determinations
I
*-<
o
C/3
MeCl-
Pentane
Cyclohexane
Hexane
Ideal
Ratio
SPE Disk
A SPE Cartridge
0.4
1
Solvent-to-Hexane Ratio
2.5
30
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 12
Normalized Root Mean Square Deviations
Oil & Grease Determinations By Separatory Funnel Extraction
4-
3-
2-
1-
0-
n
n
RMSD Acceptance Limit
Non-Petroleum
Petroleum
Hexane Cyclohexane
Solvent
NOTE Points below the respective Acceptance Limit are not significantly different from separately flannel extraction wilh Freon
31
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 13
Nonnalized Root Mean Square Deviations
TPH Determinations By Separately Funnel Extraction
4-
3-
.-
0-
D
D
RMSD Acceptance Limit
Non-Petroleum
Petroleum
Hexane Cyclohexane
Solvent
NOTE: Points below the respective Acceptance Unit are not significantly different from separately funnel extraction with Freon
32
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 14
Normalized Root Mean Square Deviations
Oil & Grease Determinations By Solid Phase Extraction Disks
Relative to Reon-113 Separately Funnel Extraction
6-
5 -
4-
P
00
D
D
RMSD Acceptance Limit
Non-Petroleum n
Petroleum m
Hexane Cyclohexane
Solvent
NOTE Points below tte respective Acceptance limit are not significantly different from sepantory fimnel extraction with Reon
33
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH:
Phase II
Exhibit 15
Normalized Root Mean Square Deviations
Oil & Grease Eteterminations By Solid Phase Extraction Disks
Relative to Hexane Separatory Funnel Extraction
5-
4-
3-
2-
1-
0-
D
D
RMSD Acceptance Unit
Non-Ifetroleum n
Pfetroleum
Hexane Cyclohexane
Solvent
^}aIE Points belowtherespective Acceptance limit are not significantly different from separately funnel extraction with Hexane
34
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 16
Normalized Root Mean Square Deviations
Oil & Grease Determinations By Solid Phase Extraction Cartridges
Relative to Freon-1 13 Separatory Funnel Extraction
6-,
5-
4-
3-
2-
0
D
n
n
n
KMSD Acceptance limit
Non-Petroleum D
Petroleum "
Hexane Cyclohexane Pentane Methylene Chloride
Solvent
NOTE Pbints below the respective Acceptance Limit are not significantly different from separately funnel extraction with Freon
35
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 17
Normalized Root Mean Square Deviations
TPH Determinations By Solid Phase Extraction Cartridges
Relative to Freon-113 Separatory Funnel Extraction
4-
3-
§
2-
i _
D
n
D
RMSD Acceptance Limit
Non-Petroleum D
Petroleum B
Hexane Cyclohexane - Pentane Methylene Chloride
Solvent
NOTE: Points below the respective Acceptance Limit are not significantly different from separatory funnel extraction with Freon
36
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 18
t
Normalized Root Mean Square Deviations
Oil & Qease Determinations By Solid Phase Extraction (Iktridges
Relative to Hexane Separatory Funnel Extraction
lO-i
9-
8-
7-
6-
5-
4-
3-
2-
i _
o-
RMSD Acceptance limit
]Sfon-Petroleum
Petroleum
n
n
n
n
Hexane Cyclohexane Pentane Methylene Chloride
Solvent
NOTE Points below the respective Acceptance Limit ate not significantly different from separately funnel extraction with Hexane
37
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Exhibit 19
Normalized Root Mean Square Deviations
TPH Determinations By Solid Phase Extraction Cartridges
Relative to Hexane Separatory Funnel Extraction
5-1
4-
3-
2-
1-
0
D
D
D
n
RMSD Acceptance Limit
Non-Petroleum
Petroleum
Hexane Cyclohexane Pentane Methylene Chloride
Solvent
NOTE Points below the respective Acceptance Limit are not significantly different from separately funnel extraction with tfexane
38
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Table 1
Student's t test on Difference Between n-Hexane and Cycllohexane
in Percent Recoveries of Oil and Grease1
Sample Type
Non-Petroleum
Petroleum
Average
Difference
20.6
3.0
SD
96.6
30.7
t-value
0.82
0.42
Probability2'3
0.424
0.683
1 Percent Recoveries Relative to Recovery Using Freon-113
2 Probability that difference = 0
3 Probability values greater than 0.05 are not statistically significant
Table 2
Student's t test on Difference Between n-Hexane and Cyclohexane
in Percent Recoveries of TPH1
Sample Type
Non-Petroleum
Petroleum
Average
Difference4
-114.3
-13.7
SD
236.1
42.0
t-value
-1.87
-1.38
Probability2'3
0.082
0.184
1 Percent Recoveries Relative to Recovery Using Freon-113
2 Probability that difference = 0
3 Probability values greater than 0.05 are not statistically significant
4 A negative difference indicates that recoveries using cyclohexane were greater
than those using n-hexane
39
-------
-------
SECTION 6
PHASE II STUDY DISCUSSION AND CONCLUSIONS
6.1 Separatory Funnel Extraction
The purpose of Phase II of the Freon Replacement Study was to evaluate n-hexane and
cyclohexane as potential extraction solvent replacements for Freon-113 in the separatory funnel
extraction/gravimetric determination of oil and grease and TPH in aqueous matrices. For both oil and
grease and TPH determinations, neither n-hexane nor cyclohexane produced results that were
equivalent to results produced by Freon-113.
On average, both n-hexane and cyclohexane extracted less oil and grease and TPH than Freon-113.
RMSDs for n-hexane were higher than cyclohexane for oil and grease analysis. On the other hand,
RMSDs for n-hexane were lower than cyclohexane for TPH analysis, with the exception of petroleum
samples. Despite these variations, a paired t-test across all samples demonstrated that the percent
recoveries between n-hexane and cyclohexane relative to Freon-113 were not statistically different
from one another for both oil and grease and TPH determination (see Tables 1 and 2).
Because the performances of both solvents were similar, the decision of which alternative solvent
was best suited to replace Freon-113 was based on practical analytical considerations, of which the
primary factor was the difference between the boiling points of n-hexane (69°C) and cyclohexane
(81°C). Based on laboratory comments regarding the extensive amount of time required to evaporate
cyclohexane, and the potential for greater loss of material when extracts were heated to the higher
temperatures required for cyclohexane evaporation, n-hexane was determined to be a more suitable
replacement for Freon-113.
As part of this decision, health and safety concerns related to the neurotoxicity of n-hexane
compared to cyclohexane were considered. The Occupational Safety and Health Administration
(OSHA) permissible exposure limits (PELs) for n-hexane and cyclohexane were compared and showed
that n-hexane is only 1.7 times more toxic than cyclohexane. Further, the Time Weighted Average for
n-hexane is 300 ppm, compared to 500 ppm for cyclohexane. Time Weighted Average is defined as
the employee's average airborne exposure, which shall not be exceeded in any 8-hour work shift of a
40-hour work week. Therefore, although it is desirable to limit the amount and degree of occupational
hazard presented by solvent substitution, the toxicity of n-hexane is not significantly higher than that
of cyclohexane and can be minimized by implementing effective safety controls and procedures in the
occupational setting.
EPA's decision to use n-hexane over all other solvents considered under the Freon Replacement
Study was further supported by the following reasons: 1) n-hexane had been previously used as the
extraction solvent for permit compliance analysis of oil and grease and TPH prior to the advent of
Freon-113, 2) EPA Phase I and Phase II studies indicate that n-hexane produces results that are as or
more comparable to Freon-113 results than other solvents (although no solvent produced results
exactly equivalent to Freon-113), and 3) a t-test run on QC data from the Phase II study showed that
there was no significant difference in results produced by n-hexane and Freon-113 for the analysis of
reagent water samples spiked with reference standards (see Table 3).
The final product of Phase II was Method 1664 (the "Method"), a performance-based method that
uses n-hexane as the extraction solvent. The quality control criteria incorporated into the Method
exceeds and improves upon that of the currently approved 40 CFR part 136 oil and grease methods,
and is consistent with the 40 CFR 136, Appendix A methods for determination of organic analytes.
An important component of these and other QC tests required in Method 1664 is the use of
hexadecane and stearic acid as the reference standards for spiking. Hexadecane was chosen to
41
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II '
simulate petroleum hydrocarbons; stearic acid was chosen to simulate animal fats and detergents, and
serves to test the effects of the silica gel procedure.
6.2 Solid Phase Extraction
In regard to alternative techniques, neither SPE cartridge or disk extraction produced results
equivalent to Freon-113. Use of these techniques are allowed, however, under the performance-based
option of Method 1664, provided that the equivalency procedures in the Method are followed and all
QC acceptance criteria are met.
Table 3
Student's t test on Difference Between Freon-113 and n-Hexane in
Percent Recoveries of Oil and Grease in OPR Samples
Solvent
Freon
Hexane
Average
% Recovery
91.3
86.2
Average
Difference
5.09
SDof
Difference
10.30
t-value
1.64
Probability1'2
0.13
1 Probability that difference = 0
2 Probability values greater than 0.05 are not statistically significant
42
-------
SECTION 7
FOLLOW-UP AND FUTURE FREON REPLACEMENT STUDY ACTIVITIES
7.1 Method Validation
Several method validation studies, ranging from interlaboratory studies used to derive QC
specifications to single laboratory Method Detection Limit (MDL) studies, have been performed as
part of the development of Method 1664. Descriptions of these studies are provided below
Twin City Round Robin Group Interlaboratory Study
A total of eleven laboratories, working cooperatively as part of the Twin City Round Robin
(TCRR) Group, performed an interlaboratory study for the determination of n-hexane extractable
material (HEM) by Method 1664. Results of this study are presented in the document titled Report of
the Method 1664 Validation Studies, April 1995. This study consisted of two parts- 1) the
performance of an initial precision and recovery (IPR) test requiring the analysis of four spiked reagent
water samples to demonstrate the laboratory's ability to generate acceptable precision and accuracy
and 2) the analyse of two sets of field samples, one from a petroleum source and the other from a'
non-petroleum source, in triplicate, for HEM.
Most laboratories did not encounter difficulties with the analysis of IPR and ongoing precision and
recovery (OPR) samples and were able to achieve acceptable recoveries of hexadecane and stearic
acid. Statistical evaluation of the results produced few outliers, indicating that Method 1664 is a
reproducible procedure sufficiently reliable to be used by a variety of laboratories. In addition the
mean RSD of field sample results across all laboratories and all samples was 13 6 thereby
demonstrating that the Method is capable of producing precise results on real world samples.
As part of the TCRR study, participants submitted comments related to the Method most of which
focused on difficulties related to extracts containing excessive amounts of water and the longer time
required for the evaporation of n-hexane from the extracted material. These issues have been
addressed, the former by recommending more careful separation of the aqueous and solvent phases to
avoid carryover of the water into the extract and that more sodium sulfate be used in the filtering
process, and the latter by allowing the use of either a water bath or steam bath set at a temperature
that results in evaporation of the solvent within 30 minutes. These revisions have been incorporated
into Method 1664. v
MDL studies
To date, five single-laboratory MDL studies have been performed as part of the effort to determine
MDLs and MLs for HEM and SGT-HEM. Results of these studies are detailed in the document titled
Report of the Method 1664 Validation Studies, April 1995. The MDL is defined as the minimum
concentration of a substance that can be measured and reported with 99 percent confidence that the
analyte concentration is greater than zero. To determine the MDL, the laboratories were required to
follow the procedure in Appendix B to 40 CFR part 136. This procedure consists of the analysis of
seven ahquots of reagent water that are spiked with the analyte(s) of interest. For EPA's MDL
studies, the hexadecane and stearic acid specified in the quality control tests in Method 1664 were
used. The MDL is calculated by multiplying the standard deviation of the seven replicate analyses by
the Student's t value for (n - 1) degrees of freedom, where n equals the number of replicates The
Student's t value for seven replicates is 3.143.
43
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
The Minimum Level is defined as the level at which the entire analytical system produces a
recognizable signal and an acceptable calibration point, and is determined by multiplying the MDL by
3.18 and rounding the resulting value to the nearest factor of 10 multiple of 1, 2, or 5. The value
"3.18" represents the ratio between the Student t multiplier used to determine the MDL (3.143) and the
10 times multiplier used in the American Chemical Society (ACS) Limit of Quantity (LOQ)
calculations (i.e. 10 -r 3.143 = 3.18). For example, if the calculated MDL is 5.8, the ML will be equal
to 5.8 times 3.18, which equals 18.4. Rounding to the nearest factor of 10 multiple of 1, 2, or 5 then
establishes the ML at 20.
The first MDL study was performed in a commercial laboratory by an analyst at the Ph.D. level
who has more than 20 years of experience in the determination of oil and grease and TPH. This study
yielded an MDL of 0.91 mg/L and a resultant ML of 2 mg/L for HEM and an MDL of 1.6 mg/L and
a resultant ML of 5 mg/L for SGT-HEM. Based on the disparity between the results obtained by this
laboratory and the lower limit of the range in Method 413.1, it was decided that a second MDL study
should be conducted in another commercial laboratory to verify the values obtained in the first study.
The second MDL study was also performed by a laboratory experienced in the determination of
oil and grease and TPH, though the analysts performing the study were not at the Ph.D. level. In
order to move expeditiously, the laboratory was required to perform the second MDL study within 24
hours. An MDL of 5.4 mg/L and an ML of 20 mg/L for HEM, and an MDL of 2.6 mg/L and an ML
of 10 mg/L for SGT-HEM was determined in the second MDL study.
The second laboratory was contacted to determine if they encountered difficulties in performing
the study. They stated that the results were the best that could be obtained under the imposed 24 hour
turn-around time constraint, and that they believed they could achieve lower MDLs given more time.
Based on these circumstances, the Agency decided that the MDLs to be included in the next revision
of Method 1664 should be those representing the better performing laboratory. Therefore, the MDL
and associated ML values from the original Method 1664 MDL study were incorporated into the
October 1994 revision of the Method.
The high results produced in the second MDL study brought into question the reasonableness and
effect of requiring a 24-hour turnaround. As a result, the second laboratory performed another MDL
Study (MDL study #3), this time without the turnaround constraint, and with the analytical objective of
confirming the MDLs/MLs that had been obtained in the first MDL study. An MDL of 2.4 mg/L and
an associated ML of 10 mg/L for HEM, and an MDL of 1.7 mg/L and an associated ML of 5 mg/L
for SGT-HEM were obtained from this third MDL study. Although closer to the MDL and ML for
HEM obtained in the first MDL study, the ML of 10 mg/L for HEM is still above the equivalent level
in Method 413.1, and the result for SGT-HEM, the more complex procedure, was still less than the
result for HEM.
From these results, the Agency concluded that the MDLs/MLs for HEM and SGT-HEM produced
in the first MDL study were self-consistent, whereas the results produced in the second and third MDL
studies were not. Therefore, the MDL and ML limits in the proposed January 1995 version of the
Method were those from the first MDL study.
The Agency still needed to address the issue that the HEM MDL values in both the October 1994
and January 1995 versions of Method 1664 had not been verified with follow-up MDL studies. In
contrast a comparison of SGT-HEM results shows that the MDL/ML for SGT-HEM from the third
MDL study supports the first MDL study results for SGT-HEM. (Both the first and third MDL
studies produced an ML of 5 mg/L for SGT-HEM.)
To verify the HEM MDL and ML values specified in the October 1994 and January 1995 versions
of Method 1664, which were the results obtained in MDL study #1, the laboratory that performed this
MDL study conducted another study (MDL study #4). As with MDL study #1, the same Ph.D. level
chemist with extensive analytical experience performed the analyses. Because the spike level in MDL
study #1 was greater than five times the resulting MDL, the spike level was lowered to 5 mg/L. An
44
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
MDL of 0.88 mg/L, with a resulting ML of 2 mg/L was obtained, thereby supporting the original
MDL results.
In response to comments received from laboratories and other interested parties regarding the
difficulties encountered when attempting to achieve the HEM MDL of 0.91 mg/L specified in the
October 1994 and January 1995 versions of Method 1664, and because most technicians performing
HEM analysis for commercial laboratories will not have the experience or qualifications of the Ph.D.
level chemist who performed MDL studies #1 and #4, an analyst with a bachelor's degree and one
month's laboratory experience performed another HEM MDL study at this laboratory. The results of
MDL study #5 were an HEM MDL of 1.4 mg/L and a resulting ML of 5 mg/L.
EPA has concluded that the MDL appropriate for Method 1664 should be representative of a
better performing laboratory. However, to realistically address the qualifications of the laboratory
personnel most likely to perform this procedure, the MDL should reflect the results obtained when
using qualified, but not Ph.D. level, personnel. Therefore, the HEM MDL specified in the April 1995
version of Method 1664 (the version being proposed) is 1.4 mg/L and the HEM ML is 5 mg/L.
Unchanged from the January 1995 version of Method 1664, the SGT-HEM MDL is 1.6 mg/L and the
SGT-HEM ML is 5 mg/L.
7.2 Future Studies
Phase III - IR Study
Evaluation of alternative solvents and techniques for infra-red determination of oil and grease and
TPH is being considered for future work under Phase III of the Freon Replacement Study. As was
decided in Phase I of the study, the use of perchloroethylene will be evaluated as a possible
replacement solvent for IR analyses. Preliminary studies of the use of perchloroethylene in place of
Freon-113 indicated that the stabilizer present in the perchloroethylene solvent could interfere with
analysis. As a result, J.T. Baker developed a perchloroethylene standard that contains lower levels of
stabilizer to minimize these effects. Formal evaluation of perchloroethylene is pilanned under Phase
III.
In addition to evaluating alternative solvents by separatory funnel extraction, alternative techniques
are being considered for evaluation and include:
1) A procedure proposed by Chem-Check Consulting that consists of non-dispersive infra-red
(NDIR) measurement of sample extracts, the residue of which are evaporated onto cavity cells.
Since this technique removes the extraction solvent from the measurement step, any difficulties
inherent to the solvents are eliminated.
2) 3M Disposable IR Cards, which consist of either a polyethylene substrate or a PTFE substrate.
As with the NDIR/cavity cell procedure, the solvent is evaporated prior to measurement and
therefore is not a concern when using IR determination.
Assessment of these techniques will most likely be accomplished through cooperative studies with the
product vendors.
45
-------
-------
APPENDIX A
SITE SUMMARY
-------
-------
CD
o
M
g
co
CM CM CM CM
CM CM
CM CM
CM CM CM CM
CM CM
g
CD
CD
M
1 i
CO
cu
_|__1
CO
cd
3
CD
rH
ft
cd"
CO
^
in
0
tn
CD
4-1
cd
U
r-H
cd
H
i-l
] >
CO
3
a
d
H 1
£*1
4-1
-H
rH
-rH
O
cd
-P 4-1
d d
CU CD
3 3
rH rH
4-1 IH
4-1 4H
Ed Ed
M >i
0 M
4-1 cd
cd 73
M d
cd o
ft o
CD CU
CO CO
O rH
r r-
CO CO
^ji Tj1
CM CM
tn
d
H
d
-H
CD
05
£i
3
CU
rH
0
i-l
J _\
CD
CM
!>1
H
cu
d
4-1
CD
05
g
R
CD
, 1
O
i-l
4-1
CD
CM
4-1
4-1 d
d CD
CD 3
3 rH
rH 4H
4H 4H
1 CO
M co
cd CD
g O
H O
M W
CM CM
V£) [ -
r- i
CO CO
*3< «rji
CM CM
tn
d
H
4-1
CO
cd
o
13 tn
d
tn-H
d .d
iH CO
73 -H
rH d
O -H
S fa
rH rH
cd cd
4-1 4-1
CD CD
2 2
4-1
d
cd
rH
4-1 CM
d
cd tn
CM -H
.d
tn co
d -H
H d
4-1 -H
CO fa
cd
U rH
cd
CD 4-1
H CD
Q 2
4H
4-1
Cd
O
rH
'
(H
CU
4-1
cd
CD
4-1
CO
cd
3
CO
CO
CD
o
o
in
CM
00
r
CO
^
CM
4H 4-4
4-1 4-1
Ed Ed
CO rH
CM rH
* '
i-1 M i-l
CD CD CD
4-14-14-1
cd cd cd
% Z Z
CD CU CD
4->4-14->
CO CO CO
cd cd cd
333
en co co
CO CO CO
CD CD CD
O O O
O O O
CM CM CM
ai o CM
I CO O
co oo en
^tt «3i ^ji
CM CM CM
tn
d
-rH CO
.d 4-1
CO O
H 3
d -a
-rH O
fa SH
CM
rH
cd 4-1
4-1 Cd
CD CD
2 2
4-1
d
cd
rH
CM
tn
4-1 d
d -H
cd co
rH CO
CM CD
o
o
tn i-i
MH CM
2
d
to
o
cd
EQ
4H
IH
ft
cu
CO
n
rH
iH
CD
4-1
cd
S
CU
-P
CO
cd
3
1^1
rH
rH
O
rH
CD
00
^l1
CM
4-1
H
rH
-H
O
cS
fa
d
O
-H
4-1
ft
CD
O
CD
&
CD
O
CO
o
cd
fa
d
o
-rH
4-1
ft
CD
O
CU
CD
i-l
O
CO
4H
4-1
Ed
&
CO
IT)
(o
i-l
CU
4-1
cd
CD
4-1
CO
cd
3
^1
rH
-H
O
CM
CO
CO
^
CM
4-1
4H
Cd
ft
CD
CO
rH
rH
rH
CD
-P
cd
*£
CD
4-1
CO
id
3
J>^
rH
-H
O
rH
O
rH
m
CM
4H 4H
4H 4H 3
Cd Cd EH 3
r- n 3 rH
CM CM rH
- rH rH
M M rH HH 4-14-1 EH 4-1 4-1
CDCD cucu dd dd
4->4-l 4-14-14-l-P-P4-14-14-lCDCD4-14-lrHCUCU
SSd CDCDCDSCDgCDCDrHrHlDCDrHrHrH
CDCUCD 333CU3CD33CH4-133- 4-14-1
4-1-P3 rHrHrH4-lrH4-lrHrH4H4HHrH 4-14-1
CO CO rH *| [ '] | 1] | (0 4-1 CO '1 1 4H 1V1 [VI 4-] '[ | t] | r£j ryi
edcd4-H 4H4H4-lcd4-lcdlH4H 4-14H4H
"334H rHCdEdEdSHBCdCd M MEdCdEd M M
H CD O O O 0
COCO 4->>l>l>iCO>,CO>1>14J4-l>i>i>i4-14-l
COCOi-lCdrHrHi-ltOi-ICO(-li-ICdCd.HrHiHCdCd
cDcDcDScdcdcdcDcdcucdcdi-irHcdcdcdi-irH
o o 4-i ,d gggogoggcdcdSggcdcd
O O rH CO -H -H -rH O -H O -H -3 ft ft -H -H -H ft ft
HrH-HttirHi-lrHrHiHi-li-lrHCDeDrHi-lrHeDCD
CMCMfa3CMCMCMpjCMCMPMD-lCOCOCMCMCMC/!CO
Cv)^l1CDfOCOr--CrlOrH^J1V£)'Nj1I V£>COO^OrHin
CDOOOOC^COCOCOCricriO^C'^OCriOC'^Cr\OOO
ODOOOOCOCOCOCDOOOOCOCOrHCOrHOOCOCriO->rH
^tr^31^1^Tl^ll^Il^TlirJr^tr.r/1^il/)i o
-H en 3 co cd H CD
d 4-lrH4-lrH fa Tjfa
o tn d cd o cd d
Hdcuocdo d 3d
4-1 -rH tn-rH 4-1 -rH CO O CO CO Cd O
tn tn-rH d M g 3 g 4-1 -H 4-1 4-1 J -H
dd-d dcucudcDOcn -P oo 4-1
-H -rH d cd 4-i rd cd _c~l 3 73 cd 33 < 1 cd
4-14-1O EHCUCJ2OT3O 4J 7373134-1
cdcdU Q OO S-l OO-Hi-1
OOCU (H OCDOiHfa O SHrHi-IO
OOB5 CUi-S-HrH-rHCM ft CMCM4-lft
ft d -rH d -CO CO 0)
rHrHg 4-lftCd4->Cd4-lU d 4-14-13 d
H-H3 cdcdtnxtncdtn cd cd.
cd cd 4-J 4-1 -P 4-1
rH rH -H dd-d a)
CMCMrH 4-1-l-lcdCdOcd O
-rH d drHrH CdrH >ld
CJitJiU 4-lcd cdCMCM fa CM Mcd
ddcd drHrH 73 d
-H-rHfa >,(dCM P-itntn cu tn4-idO)
H H H rH dd 0 dd34.J
3 3 tn cu CM -H -H d -H cd cd d
4-i4-id d tn tncotn cd tOrHi-q-rH
OO-H dtmn 4Hcocd d COCM 05
cdcdrH cddS 2cu-« CD CD MS:
4H4H73 EH-H 4-1 OO 4-1 OOlcd
ddd 4-irHdrHOcd d od-HT5
ddcd rHcdcdcdcdr-iCM -H n-HHca
cdcdtC CDrHOrHOCM cd CMM4-1O
22 jd3-HCM-rH CD 2 (UcniH
g 4-lgg g4-l> 4-1733r-H
dd3 cdRcDcucucd-rH co cddT3-H
cdcdrH cuOrd>i.dcDrH 3 a w a l
rH
-H
o
r-
o
rH
m
CM
4-1
-r-l
rH
rH
O
cd
fa
d
0
-r-l
4-1
id
4-1
I-l
o
ft
CO
d
(d
1-1
EH
0)
o
d
cd
d
CD
4-1
d
-rH
cd
*>*
"O
4-1
rH CD Cd
O CM 3
I
CD
3
rH
4H
4-1
Ed
i-l
O
4-1
cd
r4
4-1 cd
d ft
CD CD
3 CO
Ml MH II
4-1 d ft4H
CD O cd Ed 4-1
CM 3 EH 4H
II Ed
II II II
4H ft
CM CM 3 4-1 CD
S EH Ed CO
-------
-------
APPENDIX B
DATA SUMMARY
-------
-------
Q >
> UJ
CO Q
M in co st
*- oj ro NO
in cq o co
OJ O CO <
M CO M
OJ OJ OJ
IN- OJ
in o CO
CD * CD
S- co st in
CD rO ON CD
IN- NO
in rvj
CO O4 O
cn
03
v v
4- IN- NO O O
O O O CD
CO CO O
«- ro CM
O CD CD
O CO* 0
V V V
o o o
O O 0 0
IN- OJ ~*
«- «* ro
NO O IN-
N- in N.
o o
in in
o o
in in
o OJ CD in
in I1*- o in
«- ro
CD CD CD IN-
NO in ro in
0 0
in ro I
i- OJ
o o o
ro oj co
*- OJ «-
r «- «-
^ NO O*
o o m
in in N.
O 0 0
« in r\j
ru m ro
O O CD
LTt O O
r\i r\j NO
0 0
NQ r-
OJ fO
o o o
OJ N- «-
0 0
ro o
OJ CD
CD
in
CD
in
0
OJ
OJ
0 0
in ro
OJ C\J
CD
;£
0 0
in o
0
NO
o
NO
OJ
CD
OJ
0
O
NO
0 0
CD O
ro ON
OJ *
NO oj o
in IN- (M
O O IN-
m in in
0 O O 0
NO NO OJ
«- ro oj
O 0
in in
o o
oj ro
rvi oj
D
Kj
O
in
o
OJ
CD
in
OJ
o
in
0
in
0
OJ
OJ
CD
o
in
0 O
0 O
in in
OJ *
SOLVENT
ID
C
ro
X
QJ
J= 0)
o c c
o ro
O . I- - C. d>
0 U- 31
01
C
ro
X
oj
J= 0)
0 C C
o ro
~ i-
0 3= £ Q-
0)
C
ro
IS
Cyclohe;
Hexane
I- 0)
O u- n:
01
CQ at
ro
x o
a-
in
a
UJ
a.
m
§
CO
Ul
o
4->
m
i_
a.
^j "o
- t_
Q_
g
U>
(0
CJ
-------
O >
) Q
Os f--
CNJ so
so so in
«- «- c\j
ro ro -* co
ro ro os so
CO N-
ro «-
s* so CM
CM CM ro
ro «-| -* in
ro ro |s- co
CM in ro ro in c\j o*
r- O
CO O
(S- OS
o co co «-
!«- CM CM
CM
- so
000
O ts- to
CM ro ro
oooo
o *- o ro
ro -* -3- ro
2
LU
ce
OOOO
CD O
CM CM
OOOO
co so si- ro
so so co in
oooo
o o
o o
N- in
CD CD CD
CD O O
Os in «
CM ro ro
o o o CD
o o o
oooo
CO CM O «-
to sr in ro
o o
«- in
* ro
CM
O-
oooo
§§3°.
o o
o o
so in
C\J CM
o o
o ro
o o o o c
so rs- os os s
D 00 0
o o ro o
o o
O 0
ro CM
oooo
oooo
so in |s- m
o o
So
in
o o o
O O CD
co in CM
CM ro ro
oooo
O O CD CD
CM «- «- *
sf sf in ro
o o
o o
o *
ro ro
000
sf N- in
in in si-
< oooo
OL
Ul
a;
CD CD CD O
oooo
oooo
so sQ CO so
o o o
t- o -
ro
^t-
ro
o o
O sO
vt ro
1
3 o
gg g
» CO 4-»
u x u c
>» CO CJ Cl
0 3= 35 n.
01
c
CO
X
CD
°l
' CD
u x
>s . t_ Q)
U u. 3=
. 0)
C
CD
3
SS
^^ ro
0 X
>- CD
CJ 3=
S
ro
x
01
j: 0)
o C C
* o ro
o . c. 0)
CJ U. 31
01
c
ro
x
0>
.C 0)
o c
O X
>* 01
O 3C
01
c
ro
o c
01 01
£: Q.
01
c
ro
!s , I_ 01
o uu ac
c
ro
2
g S
0 X O
X » 01
o 3:
c
ro
x
01
f 01
o c c
£ Jf g
O U- 3=
8
LLJ
a.
CO
ro
CM
1
ce
>
co
ui
fc
CO
CO
si-
CM
CO
-a-
csj
o
CO
CO
Q-
CO
C
JZ
U)
'E
CL
CL
0)
u
o
Q.
CD
II
3
_
-------
Q >
1 114
CO Q
tnNOMin NOc\l NO NO o N- ON o NO N-i/^^) NO co * NT. ol
>d- ro -d- st- ONQ cDrom in ^ r--
S- in Nl
in CM 03
NO CO *- Nl
co ro
& co -
CD in
-a- >o in «J
co oo r\j«-
co in eo co
ffl
o o o o
in co [O -d-
in -d- S- in
CD CD O O
O CD O O
o ON ro ro
«- «- to CM
CD CD O O
o o o o
CD o o
o o o o
N- NO in v-
J- -J- CN O
<^ CO
-d- ro
O trv CD 1
ro ro o
2
LU
CD CD CD CD
CD NO IO O
in st S- in
O O O CD
0 O O 0
o o
N- N1
000
p tfl pj
o o o &
c\l N| NO -o'
o o o
o in NO
o o o o
*- st- CM O
sf N-
0 0 0 O
O ON O 0
«- *- ro rvi
N- CM CN
vf -* CO >
oj co
-d- rn
«- «- ro rj
o o o o
O CD CD O
CD CD CD O
CD CD O O
O O
o rn
CD o o o
in ro co CN
in in NO in
o o o o
ON O O vO
«- OJ M OJ
«- ON fr,
NO NO NO
>
_J
o
OJ
c
o c
O X
cj re
CJ
c
(0
CJ C
CD CJ
£ Q-
0)
c
CD
CJ
c
ro
X
. CJ
CD
C
CD
X
s t_
O u-
i
to
g
m
0)
C
(D
S
.£1
O
O
CJ
cu
c
CO
X CJ
c
CD
4-t
g
Q_
£
CO
I
o c.
o
U CJ
>- !-
CJ u-
CJ
c
ra
X
CJ
re
(U
c
CD
X
CJ
JZ CJ
o c
- CD
O X
>. 0
o re
CJ
c
CD
CJ C
CJ CJ
S 0.
v CD
CJ re
CJ
c
CD
g
JT
O
U
o
CJ
c c
O CD
CJ X
t_ CJ
u- re
X CJ C
. I- CD >.
CJ u- re CJ :
CD
C
CD
ca u
CD
X
-------
£Z
to « CM in
ro «- CM CM
co in
NO ro
CD o in
«- ro ^
CM CO O NO
in co o< «-
ro o
CM in
ro o ro
o co ro
. so NO ^o
> in sf CM
CO NO
ro o
I CM
ro
sf «- o sr
in T CM CD
o o o
[" o NO
o o o o
o o
CD O
o o o o
ON ^3" O ON
CD O CD
ro sr N-
o o o o
MD CO *-
M -a- to
* in co o»
-j- to N- rvj
>o r~-
to ro
o CD o ro
J- CVJ CM CO
«- «- ro
o o o o
>d" *«J" tn ON
sr ro N- CM
o o o o
o o
CD CD
o o o o
o o o
CM Nfl %J-
ro ro CM
o o o o
co t>- to
M sf M
O O O CM
sf CM « CO
T v ro
CM
a.
UJ
ee
o o o o
CD CD O O
o o o CD
ON CD O ON
ro to NO ro
o o o o
- 0
KI * ro
o *o o ro
* ro co ro
co r«-
ro ro
O O O CM
in CD ro r^
«- «- ro
«- o o
0 0 O O
0000
O CD
O CD
CD CD CD
Is w- sf
0000
CD CM O NO
in ro NO ro
O CD O O
IS 3
o o st
CN in CO
CD o o in
-* sf CM O
»- «- ro
«- \ in
ro ro CM
uj jo
V
$
g£
- m
U X
>. cu
03:
at
c
ro
X
o
Jgf
O O X
>. t- » fl> CJ CO
o = x a.
^ g g
O Q) X
>. L. (U
cj u_ :c
%
to
g
-C 0)
o c
' (D
O X
&£
0)
c
(D
CJ C
CU CU
s: a.
~ 0
CJ :n
N I_ fl)
O u- 3=
8
8 o
JC CU C
O C ID
> (D ' 4-1
U X O C
>N dj (D » C- CU
O U- SC
8
CD
X
CU CU
jn cu c
0 C CO
CO 4-<
O X O C
>* CU CU CU
CJ 3= E D-
I
1
«1
m
x
m
§
CU
m
Q
C
a:
co
tu
Z
Q
I
S
8
ro
sf
£
8
s
a.
c
i
I
?
s
Q,
c
CD
O
£
EQ
-------
CO
1
3:
1^
C
(0
(U
U
jo
1
s
t?
o
*t
Uj
5
o
t
o
"£
. .
I U4
ss.
UJ
^
It
UJ
CM
Q-
UJ
«-
CL.
UJ
>
1
O
CO
UJ
cu 2
co in "
co -,
X
^
CJ
n
CO
_J
CJ
<
u_
z
CJ
£
0 CM
CM |N-
O CD
sr sr
it sr
CD CD
sr CM
rt in
O CD
NO CM
ro sr
o o
CM CO
rt rt
cu
c:
Cyclohexal
Hexane
CO
o
Ul
CL
co
rt CM sr
o * rt
o o o
000
«- ro NO
CM It «-
0 O O
O O O
CM O IN.
CM It «-
CD CD O
0 O O
t- NO IN-
CM It «-
O O O
000
O CM It
£
Cyclohexal
Freon
Hexane
UJ
4-1
C
CU
M-
S_
LU
>.
CD
*~
I
CL.
CO
CO
CM
X
£_
CU
^
CO
h-
CU
JZ
4->
CO
O
It
O IN- ST It
in CD co ro
O O O 0
So o o
o CM in
r- CM «-
O O 0 0
ro NO o o
co o sr sr
CM «-
O O 0 0
CO O O 0
IN- «- CM Sf
«- CM «-
O O 0 0
ON CM 0 O
CO ON O IN-
CM «-
CU
c
CD
CU CU
J= CU C
O C CO
'CO '4->
U X CJ C
X cu cu cu
CJ 3= £ 0.
.
o
o
UJ
0.
CO
O NQ
ro CM
0 0
CD CD
CM NO
* «~
0 O
CD CD
- IN-
0 0
0 0
sr NO
0 O
O CD
it sr
cu
c
Cyclohexa
Hexane
CO
Q
UJ
O.
CO
CM co sr
v sr CM
o o o
in CD CO
ON «- CO
O O O
in o o
IN- NO t-
0 0 O
NO CM ro
ON CO ON
O O O
o o ro
«- CO NO
cu
c
Cyclohexa
Freon
Hexane
UJ
c
cu
M-
UJ
£_
CO
E
£
IN.
8
stf-
c\i
4-*
c
m
CL.
O)
C
'+J
m
^
L.
0
u_
-N CU CU CU
,
o
CJ
UJ
o_
CO
in O
ro «-
CD CD
in o
sr it
CD CD
in sr
sr it
O CD
0 <^
sr ro
o o
CM It
sr in
cu
c
Cyclohexa
Hexane
CO
o
UJ
CL
t/>
sr o NQ
NO ST IN-
o o o
IN- CO NO
NO IN NO
o o o
RISS
CD CD O
o ro eo
NO co in
o o o
g (C p;
cu
c
Cyclohexa
Freon
Hexane
UJ
4-)
C
CU
H-
Ul
>-
CO
e
0.
o>
CO
CO
CM
C
CO
CL
4-
,
CO
O
ci
JZ
CJ
in
NO o V sr
CO NO in o
O O O 0
ST ON IN- «-
NO in to NO
o o o o
o in ' «-
IN- NO (O in
o o o o
ON rt in «-
NO in
. O X CJ C
>. CU CO CU
CJ a: E: a.
,
o
o
UJ
O-
v>
CM
ro
o
o
CM
o
CN
0
sr
CM
o
co'
cu
c
Cyclohexa
CO
a
UJ
a.
CO
Q.
CO
-------
in CM
«^ o
O tO
to «-^
*-N-
CM T-
O O CM
CM «- in
CM *f to
to
in 00 vO CM
CM to vd «
in «
to ^
* to «-
to ^ co
in co to
*- in to
O O O
in N- in
co o o o
1^ in h- o
CM «-
CD O
to oj
«- OJ
o o o
0 0 O
C\J xO v±
CM CM tO
o o
o «-
CM ro
o o
-* o
to to
o o o o
CO CM 1^ CO
2
Ul
ce
Si
Ul
Of
oo
CM
o
to
O CM O
in N- in
V V
0 CO 0
in >o in
o o o CM
o in to o
CM
V
& o o o
*o in o «
CM «-
O CD
CM «-
r- CM
O O
^- CM
O O O
CD CD CD
tO CO N-
CM CM tO
O O O
CD CD CD
CM CM CM
O O
O CD
CM «
to to
O O O O
3:' ' '
o o
CO 1^
CD O
CD «
CM 10
o o
O CD
IO CM
00 fO O
tO vf CM
CM
O O O O
^ t>^ c\i o!
t^ ** r^- in
O O O
O CD CD
»0 CO O
CL
Ul
ce
o o o * o o o
o
to
CD O O
CD CD CD
CM CM to
o
o
IO
O O
CD CD
in -J-
to to
CD O O O
O «- N.
fO I CM
CD
X
0)
§
u
G)
O (D
O X O C
>. CO CO CO
CJ = £ 0.
CD
C.
CD
X
CD
r* (D
O X
>. CD
o z
CD
CO
X
CD
O ID
U CO X
>- t- CO
CJ u- 3:
O
C
ro
X
o
g£
CO '
o"^ S.
£
(D
M
s
d.
CO
c
CD
X
CO
JZ CD
0 C
CD
CJ X
S£
cu
c
ID
X
CD
JZ
o
"o
u
CD
§c
CD
CD X
CD
C
CO
X
CD
r- Q)
O C
CO '
U X O
>- CD CO
CJ 3C E
01
C
CO
c
CO
Q.
0)
c
CD
X
CD
j: CD
0 C
' CD
CJ X
>* CD
CJ n=
CD
C
CD
X
CO
.C CO
o c c
O CD
U CD X
>» t- CD
O u- 3=
CD
C
CO
X
o
O CO
0 CD X
>- I- CD
CJ u. n:
LU
>-
I
5
CM
I
CL
CD
CJ
O
3
D.
OO
-------
I
1
8
CD
13
§
.§
+3
.1
« o -N
fO O
sf (M
0 O
0 O
£1 r«-
o o
o o
OJ ON
IV. NO
o o
0 O
r- -*
NO r-
CD O
0 0
;ss
in NO NO
IN- NO ST
ro «- N
O CD CD
0 0 O
«- sf IN-
l\J CM «-
O CD O
(> NO NO
v- CM <
0 O O
0 0 O
in ro o
IM CM T
CD O O
O O CD
- cu
o nr
o
CJ
UJ
n
£
ra
CJ C
(U CU
£ CL
0)
c
CD
Cyclohex
^
to
C"
UJ
n
Hexane
cu
c
ra
Cyclohex
LU
i
g
CU
t
Hexane
c
ra
Cyclohex
LU
C
O
cu
L.
U-
Hexane
cu
c.
Cyclohex
o
CJ
LU
CU
C
CO
X CJ
cu cu
Pentane
CU
c
ra
Cyclohex
Hexane
_^
CO
Q
LU
CU
C
ID
Cyclohex
UJ
1
u_
Hexane
cu
c
ra
Cyclohex
o
CJ
LU
CU
C
ra
X CJ
cu cu
Pentane
cu
c
ra
Cyclohex
^
u>
a
LU
Kexane
u>
c
tn
Cyclohex;
LU
Freon
Hexane
cu
c
ra
Cyclohex:
o
U
LU
CU
C
ra
x o
cu cu
3: £
Pentane
cu
c
m
Cyclohexi
Hexane
\/
CO
Q
LU
.
o
I
O
O
ON
1§
ta
o
<
"8
cu
a.
ro
-------
o
trt
ra
§
3j
^
5,
§
Q)
g
OJ
to
S;
^
o
1
.c
t
^2
QJ
Q
1
^
c
Q
8j
IP
u
I
QC
o
t;
2
Ui
1
S«»
O
t:
o
Q.
. .
in a
g
i
2
UJ
a:
CVJ
Q.
UJ
CC
2j
oe
i
a;
g
UJ
-1 z
13 3*
X o S
t"~ t3 Jo
.« "
** '5
^.
O£
U)
UJ
U)
^g
§
M
^»
t
_J
U
2
f
u
u.
in O- o
>* r>-o
000
o o o
(M «- CVJ
o o o
o o o
CVJ » t- CJ
o u. a:
UJ
_J
s
3
>t
UJ
o
13
E.
g.
OJ
1
CVJ
0)
g
OJ
'ra
X
a
ra
o
^*
'ra
a:
CVI
CO Ov CM O
«- CVI O. -d-
o o o o
O O O 0
vO -* CM v£>
«- «- CVJ «-
o o o o
o o o o
«- »- CM «-
O O 0 0
o o o o
in to cvj in
o o o o
o o o o
cu
c
ca
S OJ
j; C
>. 01 0
co *
o o
0 0
o «-
« CM
0 O
O CD
CO «-
«- CM
0 0
o o
o o
o o
ra
Jo
ca
u x
>. OJ
O 31
C/l
O
UJ
a.
to »o to
in ^^ ^o
CVJ T-
O CD CD
000
in co «-
000
CD CD CD
to o to
""""
000
O 0 CO
-J- CO O
CD O O
O O 1^-
CO CO O
ca
g
"o c c
o ra
o 01 x
>. I- CD
CJ U. 3=
UJ
'
c
3
H-
H-
UJ
0
ra
c_
g.
cu
in
o
in
CM
to «- o
"""
0 O O
o o in
CM CM CM
0 0 O
CD CO *O
tO CM CM
000
CO O 1^
CM tO CM
O O CD
O O ~J
to to cvi
CD
C
ra
2
J= CD
o c c
o ca
U CD X
>* t- CD
CJ U. 3C
UJ
t
_J
3
1
in
CD
ra
3
CD
in
ra
3
.
O
tv.
CD
in
CM
CD
0
C
S
CD
C
CD
E
-a
CD
O
^
CO
Cti
in
CM
a
CD
ii
3
-------
QJ
-C
CL
j^
fc
&
i
Q)
fo
CU
o
C
5
s
o
c:
.g
TO
S
1
+S
Q
1
C
o
Q)
t
2
1
S
">
t:
£
^*4
s
Uj
O
t
0
C3,
DC
Q >
} UJ
!
a.
in
OJ
Q.
in
OL
UJ
Q:
a %
2 °
CO
0 UJ
ID
"f 1
X 31 3;
1 § S
(U a) *~
D-
Q. O
< t-
4-»
CU
O-
.
CO
0
_
<;
UJ
^-
t/>
UJ
Ul
§
CJ
CJ
CO
^
_J
^
u
u.
V V V
000
in in in
V V V
o o o
in in in
V V V
o o o
in in in
V V V
O 0 0
Cyclohexane
Freon
Hexane
UJ
i
4-1
C
s
*fr-
H-
LLJ
I_
O
M
ta
CO
Q.
|s.
o o o
«- |s- O
O CO CO
Cyclohexane
Freon
Hexane
UJ
-1
^
j*^
UJ
o
^
c
c-
co
CU
to
CD
3
Process
CO
sf
CM
o «- ro
sf T- CN
*
0 O CD
O CD C3
CO CO C-J
^- T- rj
o o o
O CD O
Os Os OJ
«- «- OJ
o o o
0 O O
CO Is. r-
«- «- OJ
0 O O
0 O O
OS IS. K)
«- «- CM
Cyclohexane
Freon
Hpxanp
UJ
J
t-
ro
CU
4^
CO
CD
3
CO
CO
CO
a.
1
,_,
c
CO
a.
.
CD
Q,
^
a.
IS,
sf N- Is, CO
c^ so sf co
^
O O 0 O
CD O O O
h- O IS. sf
CM CM CM CM
CD CD O O
CD 0 CD CD
h- 0 CO N1
CM M CM CM
O 0 0 O
O CD CD O
CO O O sf
CM rO CM CM
O 0 0 0
o o o o
s<3 Os >o in
OJ OJ OJ CM
Cyclohexane
Hexane
MeCl
Pentane
o
o
Ul
a.
u>
N- in ro
ro CM CM
0 O 0
CJs ^- xO
ro CM *
o o o
sf in co
ro CM «-
000
T so ro
sf CM »-
^- ^- fs.
sf OJ «-
Cyclohexane
Freon
Hexane
Ul
-1
^^
.j
in
ro
CM
c_
"
CO
CO
CD
3
CO
CO
CU
0
o
L.
a.
CO
CO
sf
CM
C
CO
a.
w
Q)
8
t_
a.
o
CO
DO
CO
CM CM
CD" ro'
V V
O 0 r- O
in in in rs,
V V V
o o o o
in in in «
V V V V
O CD CD O
in in in in
V V
in in in |s.
Cyclohexane
Hexane
MeCl
Pentane
1?
CJ
UJ
CU
tn
a.
ta
n
3
-------
Q)
ca
§
{i
^
s
(0
01
%
CO
^^
2
(Q
^
o
H^
o
C
5
GJ
«fe
g
a
1
§
c
i
fc
Qj
O
9
§
Q^
S
§
1
5
U}
"o
s
CJ.
a
. .
§1
Z
m
£
10
UJ
a:
CM
o.
UJ
oe
a.
UJ
ce.
1
3:
15
t_
ra
O UJ
m "a 2
X as =5
"""" O
0 ty *~
ul'o
< H
e>
0.
"a
M
O
t
s
o:
to
uj
%
3
u
^"
u
u.
s
3
§
CM tO O
CM CM tO
O O O
*3- T CM
»» T- tO
000
» to o»
*3" < CM
0 00 0
in co *
>» to
o o o
« IO -3-
-a- v- to
Cyclohexane
Freon
Hexane
UJ
^^
«*-
H-
»-
**
\^
O
la
I
«-*
CM CM tO
O O O
CN O O-
to d- in in
Cyclohexane
Freon
Hexane
LLJ
1
1
H-
S
UJ
Q.
(U
t/3
in
v
^^
£_
0)
M
ra
3
0)
4-»
(A
3
__,
0
CM
CM
in co si- sf
in o ro o
0000
CD (M vO CD
ro CM to to
O O CD O
o CM to o
ro CM to to
0000
in «- st «
CM CM fO CM
CD O O O
CO CM 0 T-
CM CM -s± rO
Cyclohexane
Hexane
MeCl
Pentane
"o
(J
UJ
d-
C/}
>o >o ro
o -* ro
0 O O
l> CO O\
CM vf «-
O 0 0
o in in
CM >* v-
0 0 O
0 TO 0
ro in CM
000
ON in CM
CM SJ- IM
Cyclohexane
Freon
Hexane
UJ
1
M-
M-
UI
D.
IU
u>
*~
''
*->
L.
O
ID
N-
C
ro
c
ID
CJ
Q
ON in ON FN-
CD CM CD O
O O O O
o sf in sf
ro CM CM CM
o o o o
0 «- sf sf
ro CM CM CM
0 0 0 O
o NO in in
ro CM CM CM
0000
«- sf O to
tO OJ CM CM
Cyclohexane
Hexane
MeCl
Pentane
o
o
LJJ
Q_
C/)
Sf 0 «-
«- 00 TN-
CM
O O O
N- NO CJ*
o o o
&>££
o o o
Ch -
M
CO
3
I/)
O
O
0.
a
CO
3
4-»
C
to
CL
D)
C
-M
O
CO
M-
c
CD
C
(0
CJ
OS *O *4" «~
««-** in
o o o o
in in in o
in in -d- sf
0 0
33
O O O O
\O *Q CD **i"
in in in ro
o CD o o
ro ro -a- ro
in in -d- -*
Cyclohexane
Hexane
MeCl
Pentane
o
UJ
CL
V)
3
Q.
PQ
-------
^
I
i?
i:
fc
-Q
CO
cu
CO
CD
cu
C3
C
CO
§
M-
0
c
nj
.g
cu
fi
Q
O>
£
c:
o
CJ
-5
1
S
"H
,g
fij
s
CD
t
O
Q.
fv-
Q >
t LU
CO Q
^
UJ
E
UJ
oi
a.
UJ
£
UJ
arbons
SOLVENT
0 UJ
o ^
CQ -a a
CD CD '
«3T c
CU
CL
.
CD
O
1
s:
UJ
t
CO
UJ
H-
CO
O
CO
>-
_J
CJ
2
z
o
u_
OJ r- *-
0 O 0
~ fc^
o o o
in «^ IN-!
O 0 0
< CO O
CO 0 O
CO ^- -»
OJ «-
Cyclohexane
Freon
Hexane
UJ
fl^
3
h-
£
4-J
C
CD
M-
M-
UJ
t_
CD
4-1
1
OJ
O
to
u~
O)
.^
^
c
CD
3
Q
"~
*o ro
V V
o in oj CD
in N- in in
V V V
0 O O 0
in o in in
V V V V
0 0 O 0
in in in in
V V
CD in in CD
in N- in in
Cyclohexane
Hexane
MeCl
Pentane
.
o
CJ
111
a.
V)
CD in o
OJ * v-
o ro «-
^ N- \O
O CO OJ
ro* co N^
CD O CO
« N- in
in co OJ
co in in
Cyclohexane
Freon
Hexane
UJ
I
Q)
4-»
CD
3
r-
ro
3
8
CO
OJ
CO >* 1^ OJ
O O O O
oj o o in
CO t^- ^ CO
CM ~* o ro
co r^- oj co
CD O CD in
o co o' co
-* ro o r--
t^ CO t- CO
Cyclohexane
Hexane
MeCl
Pentane
o
CJ
UJ
CL
CO
cu r*- Is-
O h- CO
ro ro vt
000
r*- -*
01
£
ro
t
c_
0)
4->
ro
cu
r
to
o r~- r^
ro o oj
'
O 0 0
O N- CO
0 OJ O
J- \O ^
CO N- CD
OJ
-
co
CD
o
o
Q.
0
a
3
c
ro
CL
1
-O
m
m
cu
CD
-------
000
in in in
o o o
in in in
o o o
in in in
V V V
o o o
in in in
Cyclohexane
Freon
Hexane
UJ
4-»
c
3
»+-
Ul
L.
r-
o.
i
CM
M
C
CO
a.
m
H-
«s
Q
U
§
5
^
o o o o
in in in in
o o o o
in in in in
O CO O CD
in in in in
V V V
o o o
in in in
4)1 * QJ QJ O
0 =c E a.
o
CJ
UJ
a.
CM NO NO
IN^ ON sf
O O O
O IN. O
O CM sf
000
CM IN- o
O. ro in
O 0 O
to co o
CO CM CM
O O O
r co o
O «- in
Cyclohexane
Freon
Hexane
UJ
S
^i
L.
-------
in
OJ
N ro in o is, sr
N rvi
o co i
i o
ro
sr vo in co
-o «- in co
in >o
N- m
«- in ro *-
OJ O O |s!
in «- sr
O CD I
V
CD o in
1000
i ro «- rvi
0000
O O CD CD
ooo
FC^IC
CD CD O CD
000
O O N.
ro «- i-
O O i OJ
in o co ro
in CD 1s-
co *^- in
O CD N-
OJ OJ «-
ooo
oj in in
ooo
o ro ro
ooo
o co rs.
ro »- «-
o CD CD o
g' o\ co r-
o co ro
O O CD CD
0 0 C3 O
0 0 O O
oj o d-
in in o
v y
ooo
oj in in
(\l
\O O O CD
~or- ^- <-
ooo
OJ O «
ooo
ooo
o o ro
OJ OJ «
ooo
ooo
r o rvi
rvi «- r\i
o o
0 0
ro *-
O CD
o ro
>o o
o o
ro in
co ro
0 O
ro o
N. OJ
ooo
ooo
Is. OJ OJ
ooo
CD O CD
«- fO 0
ooo
000
o oj sr
ooo
ooo
o ro oj
CD
0
OJ
CD
0
CD CD O
in CO Is.
ooo
(2 fl r8
0 0 CD O
0 O CD 0
oj CD rvi oj
0 0 CD O
O CD CD O
rvi CD rvi ro
CD CD CD
o o ro
N- o in
O CD CD
rvi o ro
O « in
ooo
O 0 CO
« OJ x
ooo
« « in
OJ OJ «-
SO
C
X
OJ
0
c
ro
X
0
J= 0
O C
' ro
o x
>. CO
tj "y"
0
c
ro
0 0
E: o.
0
c
ra
X
0
j=
o c
o
U CO
X L-
CJ u-
0
c
ra
X
0
^
0
c
ra
X
0
J= CO
o c
ro
o x
>* 01
cj a:
0
C
ra
-M
o c
0 0
y Q_
C
ro
X
CO
JZ
o c
' O
U 0
X i-
CJ U.
0
c
ra
g
3:
0
c
ra
x
0
-C 0
o c
ro
0 X
>~ 0
CJ 3=
(J
0
0
c
ra
4-*
C
0
Q_
0
C
ro
x
0
jr 0
o c c
o ra
0 0 X
>. I- 0
u u. :n
CJ
CO C
c ra
ra ' 4->
X O C
CO (U CJ
o ro
0 0 X
>. i- 0
o u- :E
J= 0
O C C
o ra
0
o
0
: o
o
u
ui
o.
CO
3
I
sr
3
H-
O
C£
I
CO
S.
0
CO
3
X
c?
CO
a
o
o
o
Ul
CO
U
o
O CO
< CO
g
o
II
3
-------
-------
APPENDIX C
STATISTICAL TECHNIQUES
-------
-------
APPENDIX C
STATISTICAL TECHNIQUES
Outlier Screening Methods
A robust quantile screening method suggested by Emerson & Strenio (in Hoaglin, et al., 1983)
was used to screen across the set of laboratory results for each compound (oil and grease, TPH) and
sample stratum (all samples, petroleum samples, non-petroleum samples) for outlying values. This
method estimates the levels above and below which fall 1% of the data, according to the following
formulae:
Lower 1%
Upper 1% level=Qz+l.5IQR,
where Q, is the first quartile, i.e., the value below which fall 25% of the data; Q3 is the third quartile,
i.e., the value below which fall 75% of the data; and IQR is the interquartile range (75th percentile-
25th percentile). Solvent-to-Freon ratios (or solvent-to-hexane ratios) were determined for each
sample. These ratios were then subjected to the outlier screen and points outside the range were
rejected and removed from descriptive analyses (i.e., solvent-to-Freon ratios, solvent-to-hexane ratios,
and median absolute deviations).
Data Transformations
The standard deviations of the measured sample concentrations were roughly proportional to the
mean sample concentrations, as is common with analytical data. The graph in Attachment 1-A to this
Appendix, a plot of the replicate standard deviations versus the mean of the replicate analyses for
Freon-113, n-hexane, and cyclohexane separatory funnel extraction, demonstrates this trend of
increasing standard deviation with increasing concentration. This heteroscedasticity violates the basic
assumptions of Analysis of Variance (ANOVA), and needed to be corrected prior to calculating the
Root Mean Square Deviation (RMSD).
To eliminate the heteroscedasticity, the concentration results were transformed to produce data
with constant standard deviations. Since the intercept in the standard deviation regression was zero,
the data were transformed using the equation
z = In(x),
where x is the sample concentration (in mg/L) for each individual result within all triplicate analyses,
and In indicates the natural logarithm function. As is demonstrated by the graph in Attachment 1-B to
this Appendix, following this conversion, the data no longer showed a statistically significant
correlation between standard deviation and mean concentration. Consequently, the transformed data
were suitable to the ANOVA analysis described below.
Negative Concentrations
While it is counterintuitive to find negative concentrations, it can be expected to occur, at least
occasionally, due to analytical variability whenever very small concentrations are measured via
methods involving blank-subtraction. It is possible that some sample/solvent combinations could be
C-3
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
removed from this study by setting up statistical limits on whether the sample can statistically be
shown to contain a non-zero amount of oil and grease, for instance simply by computing a one-sided
hypothesis (at say 5%) that the mean concentration for a sample/solvent combination is greater than
zero using the three replicate measurements (either with or without the data transformation). However,
it is desirable that the study remain balanced across solvents, and since some solvents show
concentrations clearly greater than zero while other solvents do not, such a rule may not be useful.
Instead, negative concentrations were treated as non-detects, and their values were set to one-half of
the detection limit (2.5 mg/L).
ANOVAs
The ANOVA model deemed best for this analysis was
y=a+bi+gj+dij+eijk
where i=l...I solvents (l=Freon), j = 1...J samples, and k=l...K (3) replicate analyses. Here a, bi; gj;
and djj are all fixed effects that were estimated1, and eijk is the random measurement error, with
mean=0 and variance=o2. This model is a standard ANOVA model, and was fitted to obtain estimates
of each of these parameters and standard errors associated with each estimate. Testing of this model
in Phase I (see Appendix C, Attachment B of Preliminary Report of EPA Efforts to Replace Freon for
the Determination of Oil and Grease, Revision 1, September 1993) demonstrated that the interaction
term is significant (i.e., the effect of solvents depends on the sample matrix). This association
prompted evaluation of the results on the basis of similar matrix groupings, or strata (i.e., all samples,
non-petroleum samples, petroleum samples).
Evaluation of Solvents
To adequately compare solvent performance, the statistic used needed to provide an overall index
of performance that summarizes the similarity of each alternative solvent to Freon-113, while taking
into account the differences in the outcomes for each solvent on different samples. One measure for
this is the root mean square deviation, across samples, between the alternative solvent and the Freon-
1 1 3 results. In terms of the model above, the mean result for each solvent for a sample is,2
so the root mean square deviation for solvent i is
'With Ebp 0, £gj= 0, and Edy= 0 over i for fixed j and over j for fixed i.
2 For typographical simplicity, the carats over each parameter are omitted from this point on even though
the formulae refer to the sample estimates rather than the theoretical model values.
C-4
-------
Heport of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
This can also be computed as the root mean square deviation between the sample* sol vent cell
means for the alternative solvent and Freon-113. The smaller this measure, the more closely the
results using the alternative solvent approximate the results using Freon-113. RMSD computes the
squared deviation of the average analytical results using alternative solvents on a sample from that of
Freon-113 on the same sample, and accumulates this over all samples to provide an overall measure of
agreement. The data show significant interaction between solvent and sample in the statistical model
that is, whether alternative solvents extract more or less oil and grease than Freon-113 varies according
to the sample matrix. RMSD was chosen as the primary statistical measure because it assesses
variations both above and below the Freon-113 results, which was necessary in order to account for
the possibility that an alternative solvent may extract significantly less oil and grease than Freon-113
on some samples, and more on other samples.
Acceptance Limits were derived by computing the RMSD that would be expected by chance
a one, *g., ,f Freon-113 were tested by this protocol and compared with itself using separate analyses
Under the null hypothesis that there is no actual difference in the procedures, the square of the RMSD
appropriately normalized by the residual error estimate, will have an F distribution Therefore
K(RMSD?
2s2
~F
JJJ(K-l)
where s is the root mean square error (RMSE) of the model, and I, J, and K are as above A 95%
acceptance region for the equality of the test is ,
RMSD<
K
r' J,U(K-l)(0.95)
or, in terms of the normalized RMSD,
RMSD< /_!
K
Examples of these analyses are shown in Attachment 2.
C-5
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Appendix C
Attachment 1-A
Triplicate Standard Deviation as a Function of Mean Triplicate Concentration
0
100
200 300 400
Mean Triplicate Oil & Grease (mg/L)
500
600
C-6
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Appendix C
Attachment 1-B
Triplicate Standard Deviation as a Function of Mean Triplicate Concentration
Following Natural Log Transformation
§
Q
GO
1-
0.1
0.01
11-1
II
' n' ' I
1 10
Mean Natural Log of Triplicate Oil & Grease (mg/L)
C-7
-------
Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease and TPH: Phase II
Appendix C
Attachment 2
Example of RMSD Acceptance Limit Analysis
Sample * Solvent Means (after Natural Log Transformation)
Sample
24870
24871
24873
24874
24875
24876
24877
24878
24879
24880
24881
24882
24883
24884
24886
24887
24888
24889
24890
24891
24893
24896
24897
24898
24899
24900
24901
24902
25101
25104
25105
25106
25107
Freon
1.89976
1.08922
3.62188
2.85957
3.19262
5.44335
4.29667
4.88645
5.90111
4.00163
4.19625
4.47511
3.87275
4.99640
4.02527
4.62359
5.78441
4.35926
2.34573
1.94084
3.52070
3.74459
4.16967
2.96354
6.22000
5.49763
4.65902
4.66147
4.20046
5.16971
5.21582
3.63882
3.37937
Hexane
2.24750
1.78270
3.36226
2.31012
3.11160
5.14898
4.11848
4.78586
5.78575
3.86160
3.98256
4.27638
3.43653
4.55492
3.55624
4.45701
5.05031
4.18910
1.88590
0.91629
3.29609
5.23774
3.38595
3.20073
6.12417
5.15191
4.77981
5.18971
3.60976
5.21991
4.66391
3.14549
3.23419
Cyclohexane
1.69108
1.10924
2.87459
2.30687
3.08939
4.98348
3.88191
4.83458
5.66966
3.97592
4.00258
4.23395
3.59227
4.56941
4.05484
4.53798
5.33054
4.20105
2.62744
0.91629
3.48958
3.61525
3.44812
3.41185
6.14840
5.33943
4.77752
5.00874
3.70926
5.08786
5.01186
3.26684
3.37465
RMSD
0.49173
0.36581
s(RMSE) I J K
0.13397 3 33 3
C-8
-------
Report of EPA Efforts to Replace Freon for the Determination of oil and Grease and TPH: Phase II
Acceptance Limit =
2s^
K
J,IJ(K-l)(.9S)
2(0.13397)2 , c
3
0
Normalized RMSD =
RMSD
K
Normalized RMSD for Hexane =
0-49173
=45Q
2(0.13397)2
Normalized Acceptance Limit =
-5 =1.22
Since 4.50 > 1.22, Hexane is not equivalent to Freon
C-9
-------
-------
APPENDIX D
REFERENCES
-------
-------
APPENDIX D
REFERENCES
1.
2.
4.
5.
6.
7.
8.
9.
10.
Draft Study Plan for Phase II of the Freon Replacement Study, September 29 1993
Available from the Sample Control Center (operated by DynCorp Environmental Programs
Division), 300 N. Lee Street, Alexandria, VA 22314, (703) 519-1140.
Emerson and Strenio In Understanding Robust and Exploratory Data Analysis Hoaglin D C
Ed.; John Wiley & Sons: New York, 1983; Chapter 3. >&>-,
F.K. Kawahara, A Study to Select a Suitable Replacement Solvent for Freon 113 for the
Gravimetric Determination of Oil and Grease; Environmental Monitoring Systems Laboratory
Office of Research and Development, U.S. EPA: Cincinnati, Ohio 45268, October 2, 1991.
Method 1664: N-Hexane Extractable Material (HEM) and Silica Gel Treated N-Hexane
Extractable Material (SGT-HEM) by Extraction and Gravimetry (Oil and Grease and Total
Petroleum Hydrocarbons), April 1995, Document No. EPA-821-B-94-004b EPA Water
Resource Center, Mail Code RC-4100, 401 M Street, S.W., Washington, D.C. 20460.
"Methods for Chemical Analysis of Water and Wastes", 3rd Edition; Environmental Protection
Agency, Environmental Monitoring Systems Laboratory-Cincinnati (EMSL-CiV Cincinnati
Ohio 45268, EPA-600/4-79-020, 1983; Method 413.1 and Method 418.1.
N.R. Draper and H. Smith, Jr., Applied Regression Analysis. Second Edition- John Wilev &
Sons: New York, 1981, pp 238. '
Preliminary Report of EPA Efforts to Replace Freon for the Determination of Oil and Grease
Revision 1, September 1993, Document No. EPA-821-R-93-011, EPA Water Resource Center'
Mail Code RC-4100, 401 M Street, S.W., Washington, D.C. 20460. ;
Proceedings of the Seventeenth Annual EPA Conference on Analysis of Pollutants in the
Environment, January 1995, Document No. EPA-821-R-95-008, EPA Water Resource Center
Mail Code RC-4100, 401 M Street, S.W., Washington, D.C. 20460.
Report of the Method 1664 Validation Studies, April 1995. Available from the Sample
Control Center (operated by DynCorp Environmental Programs Division), 300 N Lee Street
Alexandria, VA 22314, (703) 519-1140.
"Standard Methods for the Examination of Water and Wastewater", 18th Edition; American
Public Health Association: 1015 Fifteenth Street, NW, Washington, D.C 20005 'l992- Method
5520B and Method 5520F.
D-3
-------
-------
ft'
-------
Ill III II I II II I I II
ipi n null n in i|n i in 11 in i n in in nil i n i iiiiii i mi n n
in ill in 11 inn i in mi in in 11 nil in n iiiiii iiininn ill
n mi null n nil in i n in i iiiiii i iinnnin in in n i|iiiiin INI in in 11 i in in n 11 In I In 11 n in n n iiiliin
linn in i in ill i n i n in in nil 11 n n i in n linn i inn linn in nun n inn inn in ill nn i n linn i
iv a
, ui««»
Illn
IIIIP in in
-L niliiiii i
' iiiri I
nn.tjHi
I
si
(ill*
u 1
1' i'«"
"Bill llllf-
w II
Lf "liraI
" iii!
i i''ii''iiiiiiiii(iiiiii^
,1 i»iiii iiiii UK < iiii 111 'i HI
S' nj
i*'
fit
v\t
4
LI.
nun I
i
in
I
I1
i
------- |