RE: Response to Citizens Opposed to PCB Ash (COPA) Concerning Data Reports

August 21, 2000

The United States Environmental Protection Agency (U.S. EPA) has reviewed the correspondence from COPA concerning the data reports that have been placed in the Administrative Record and Information Repository. Specifically the Administrative Record Update for water sampling at Neal's Landfill, the summary of the Lemon Lane Landfill analytical results for split samples collected by Earth Tech in 1997, and the Conduit Investigation Progress Report for July to October 1998.

Issue 1: Reporting Results from the Wrong Analytical Method

As background information, CBS Corporation, now doing business as Viacom submitted the Quality Assurance Project Plan (QAPP) initially in March 1995. The document was revised twice in August 1997. The laboratory procedures are contained in Appendix A of the QAPP and those procedures have been modified a number of times. In June of 1998, CBS submitted the laboratory procedures for EPA Method 8082 for PCB analysis by Heritage Laboratories. Based upon comments from the U.S. EPA, in September 1998 CBS submitted a revision by Heritage Laboratories for EPA Method 8082. Subsequently, in March 2000, Heritage Laboratories submitted a revision to EPA Method 8082 based upon internal changes and audits of the laboratory. In addition, in April 1999 after review by the U.S. EPA, CBS submitted the laboratory procedures for PCB analysis using EPA Method 8082 for the backup laboratory Antech. Finally, October 1999, CBS submitted the laboratory procedures for fish tissue analysis by En Chem, which was revised after comments by the U.S. EPA. Enclosed with this correspondence is the latest laboratory procedures.

COPA requests in Issue 1 a confirmation that analyses done by all respective laboratories involved with the Superfund sites in Bloomington, Indiana, were appropriate for polychlorinated biphenyls (PCB) and also an explanation of why Heritage Laboratories was using an incorrect method for PCB analysis.

The original September 1986 edition of EPA's "Test Methods for Evaluating Solid Waste" (SW- 846) included Method 8080 for measuring organochlorine pesticides and PCBs using any two of a wide variety of gas chromatography (GC) columns. The second update of SW-846 dated September 1994 added Method 8081, a capillary column variant of Method 8080. Both Methods 8080 and 8081 can be used for chlorinated pesticides, for PCBs, or for both, depending on which calibration mixtures are used for the analyses. The third update of SW-846 dated December 1996 deleted Method 8080 and included Method 8081 A for chlorinated pesticides only and Method 8082 for PCBs only. Other than the calibration mixtures used, there are minimal differences between Method 8081 on the one hand and Methods 8081A and 8082 on the other. Therefore, Method 8081 is an obsolete but correct equivalent for either Method 8081 A or Method 8082 or even for both methods combined in a single analysis. It is possible that the laboratory associated with Issue 1 did not update the analytical method labels that laboratory information management system (LIMS) put on the sample analytical reports. However, with regard to analytical results, the methods discussed above are essentially identical. Therefore, Heritage Laboratories was not using an incorrect method and the appropriate method was being used. The U.S. EPA can make arrangements for COPA to review gas chromatograph charts if desired.

Issue 2: Consistently Using Comparable Methods

COPA requests in Issue 2 evidence that the analytical methods used for the sample analyses at the Bloomington, Indiana, Superfund sites were consistent with the quality assurance project plans (QAPP) that used SW-846 Method 8080, and if not, to provide an analysis of the effect of any differences to ensure comparability of the data.

The current Contract Laboratory Program (CLP) method for PCB analysis is essentially a more restricted version of the original Method 8081; earlier CLP methods were more restricted versions of the original Method 8080. For instance, Method 8081 recommends use of surrogates and suggests use of various specific surrogates, but the CLP requires use of specified amounts of tetrachloro-m-xylene (TCX) and decachlorobiphenyl (DCB). It is customary for a user of any SW-846 analytical method to specify in a QAPP, a basic ordering agreement, or another appropriate document the optional elements to be used in the analyses. The routine methods (that is, those that will be used for analysis unless specified otherwise) at most reputable environmental laboratories are generally close to the CLP procedures. The most common variant is use of more laboratory control samples than the CLP requires.

In general, the only way that one can determine whether laboratories have used comparable analytical methods is to compare their standard operating procedures (SOP) and the results for their quality control (QC) samples. However, all laboratories used by EPA, CBS, and Earth Tech followed the EPA-approved QAPP for the Lemon Lane Landfill site.

The U.S. EPA disagrees with COPA's statement that the split sampling during the 1996 sampling event produce widely varying results. The Lemon Lane source control operable unit Responsiveness Summary contains a response to citizens concerns concerning the 1996 sampling from within the landfill.

Issue 3: Lack of Quality Control Data

In Issue 3, COPA requests the QC data for Neal's Landfill, Bennett's Quarry, and Lemon Lane Landfill. In addition, COPA would like all future reports to include this data.

The U.S. EPA will arrange for a data evaluation section to be included in its final report on its split sampling. Enclosed are copies of the Neal's Landfill, Winston Thomas Wastewater Treatment Plant and Bennett's Dump data evaluation reports developed by Tetra Tech comparing the split sampling to CBS Corporations data and a evaluation of the U.S. EPA data. The U.S. EPA deems the U.S. EPA split sampling data to be very consistent with the CBS data. CBS will provide data validation reports with the Final Reports for Neal's Landfill, Bennett's Dump and the Lemon Lane sites.

Issue 4: Inadequate Surrogate Recovery Information

COPA discusses surrogate recoveries in Issue 4 and requests that the QAPPs contain the acceptance limits for QC results and be immediately included in the information repository record at the Monroe County Public Library.

Control limits are the range of recoveries that one will obtain most of the time for samples (1) that have little or moderate matrix interference, (2) whose extracts have been subjected to minimal dilution, and (3) that have been analyzed in a comparable manner. Therefore, it is best to use laboratory-, site-, and matrix-specific surrogate limits. In most cases, however, the best available approach is to use laboratory-specific surrogate limits derived from real analytical sample analytical results using routine statistical procedures.

Surrogate recoveries of over 100 percent occur often. Recoveries of up to about 120 percent can occur as a result of the natural variation in the response of an instrument that is operating within acceptable continuing calibration limits. Much greater recoveries (up to several hundred percent) can be caused by interfering substances in a sample. Interferences are quite common for TCX because very light PCB congeners can co-elute with TCX. In addition, interferences can cause apparently low surrogate recoveries on a chromatogram by adding peaks adjacent to the surrogate peaks, thereby raising the baseline but not adding to the peak height.

It should be noted that control limits differ from laboratory to laboratory and that even within a laboratory they change from time to time. The EPA-approved QAPP has been placed in the information repository at the Monroe County Library and the laboratory procedures are enclosed for your review.

It should be pointed out that complete validation of data is performed on only certain samples. Several of the samples that COPA notes in its correspondence are karst conduit samples that were samples CBS took for its investigation. These samples many times did not receive full validation efforts because they were strictly done to determine the presence of high PCB levels in potential conduit locations. Therefore they were viewed more as a go no go test rather than a detailed highly accurate sample that may need to pass scrutiny to the level of a risk assessment or required monitoring event sample. The detailed sampling and analysis plans for each event detail what level of validation a sample will incur.

Issue 5: Discarding a Critical Portion of the Sample

In Issue 5, COPA requests an explanation of the sample mass discrepancies among the sample analytical methods, the amount of sample collected according to field records, and the amount of sample actually used in sample analysis. COPA states that discarding critical portions of the sample results in relatively low levels of PCBs being detected.

Issue 5 pertains to discarding (not extracting) parts of a sample at some point between sample collection (as documented in field notes) and analysis in the laboratory (where some sample may be needed for matrix spikes, some sample may not be removable from the sample bottle, and so on). The only significant effect of modifying the amount of a field sample that is extracted is changing the sample detection limit; however, this statement assumes that the sample is homogeneous. If the sample is not homogeneous, the laboratory will define the sample in terms of the label on the sample bottle. If the label says "water," the aqueous phase will be analyzed; if the label says "sediment," the solid phase will be analyzed. If the entire sample is to be analyzed, this intention must be made clear to the laboratory.

Essentially all of the methods for extraction of aqueous samples are inappropriate for samples with a significant solid phase. Therefore, one needs to perform a phase separation and then two analyses, one of each phase. Also, the nominal amount of sample needed for an analytical method is always subject to change. The determining factor is the detection limit capabilities of the analytical instrument. As analytical instrumentation improves, one can often obtain acceptable results using a less concentrated extract-that is, one prepared using less of the original sample.

The lab will routinely use 1000 ml of sample as the method specifies except in the following circumstances:

a. Less than 1000 ml provided by the field. For some instances, less than 1000 ml is taken in the field. This can happen if the sample was taken by an auto sampler (most storm samples at Neal's Landfill and Lemon Lane are taken with auto samplers). The auto samplers used take at most 500 ml of sample. Other instances where this occurs is during drilling activities such as for the karst conduit program. Many samples were submitted to the laboratory upon the occurrence of the first encountered water during drilling. In these instances, CBS obtained a sample from what was available as an indication that the water encountered was or was not highly contaminated (sometimes what was available was much less than 1000 mls). Many times what water was available during drilling was a slurry with mud and sediments (sometimes taken from a 55 gallon drum filled mostly with cuttings). In cases where a lot of sediment is mixed with the water, the lab will analyze the phases separately (as directed by CBS).

b. Multiple analyses to be performed with one bottle of sample submitted by the field. Many times CBS analyze routine samples from the field for both PCBs and TSS. The TSS sample is drawn from the same bottle provided for PCB analysis. Typically 100 mls of sample are drawn off for the TSS analysis leaving less than the original 1000 ml for PCB analysis.

In cases where the extracted sample volume is less than 1000 ml, the nominal reporting limit may only be changed if it is significantly less than 1000 ml. This is because the actual limits of detection at the lab are less than the reported 0.1 ppb. The lab has provided specific detection limit studies to show this. For example, in cases where 890 ml of sample was extracted rather than 1000 ml, the actual reporting limit may not change.

If there is a significant change in the amount extracted, the lab will change the reporting limits. In many investigation type samples where significantly less than 1000 mls of sample were extracted (at the expense of detection limits) it was not necessary to determine the sample concentration at 0.1 ppb because CBS was looking for conduits with levels of contamination orders of magnitude above this.

All analyses conducted by the laboratories associated with Issue 5 followed the EPA-approved QAPP. Unless a specific sample, type of sample, or analytical objective is identified by COPA, Issue 5 cannot be resolved.

Issue 6: Unnecessary Dilutions

COPA discusses in Issue 6 the dilution of samples at the laboratory leading to elevated detection limits in cases where PCB concentrations are low.

Again, if the lab used significantly less sample than 1000 ml it was either because no more was available or the sample was a mud/water mix. If a small amount of sample was used, even though detection limits were raised, this higher detection limit would still meet the intent of the sampling event, which in the case of samples LL 1450/1451 was to find the highly contaminated conduit feeding the spring (concentrations greater than 2 ppb were being sought).

In cases where dilutions are required because of high levels of analyte,dilutions are performed on the extract after initial screening on the instrument. In these cases the surrogate recoveries from both the original un-diluted sample and the diluted sample are calculated and reported with the raw data package.

It is not clear what COPA is referring to by asserting that spiking the same amount of surrogate into a smaller sample volume gives false indications of surrogate recovery. The lab reports surrogate recoveries based on the amount found in the final extract volume. As long as the extract volume doesn't change from that expected, the surrogate recovery reported will be correct. In situations where dilutions are required and the final extract volume changes the lab does take the new volume into account in reporting surrogate recovery.

Issue 7: Inadequate Information to Assess Data Problems

In Issue 7, COPA requests that all field sampling reports and data be provided to the Monroe County Public Library information repository.

The U.S. EPA has provided oversight reports which include the field notes for the activities at the Lemon Lane Landfill. The turbidity and specific conductivity for the water sampling has been provided when the data has been gathered. The raw laboratory data is available if specific samples are in question. The data packages are extremely large and unless requested by COPA, the U.S. EPA will not be placing the data into the information repository.

Issue 8: Improper Quantitation Techniques

COPA requests in Issue 8 verification that the quantitation technique used for sample analysis meets the QAPP requirements and is consistent with quality analytical techniques and analytical methods.

GC methods cannot be used to prove the identity of analyses. However, there are several ways to quantitate uncertain results. The following 'three basic approaches can be used in defined combinations:

• Average the results for the two GC columns. This approach would be acceptable if the entire analysis proceeded well under ideal conditions. However, this approach can be misleading and is usually biased high because of interference from nontarget compounds in one or both columns.
• Use the primary column results. This approach is best if the continuing calibration results for the confirmation column have greater variation than those for the primary column. This situation occurs frequently.
• Use the lower of the results for the two columns. This approach is acceptable if the continuing calibration results for the two columns are similar or if the column whose results are selected performed better during continuing calibration. CLP methods require use of this approach for samples for which the difference between the two columns' results exceeds 25 percent. (Such results are flagged with "P" in laboratory reports. Data validators usually change the "P" to a "J" and note that some or all of the results may be false positives.) The reason for this variation between columns is that much of the higher results actually represents interfering substances, not the analyte. Of course, some or even all of the lower results may represent interferents, but no means exist to determine whether this is the case. Therefore, the approach avoids some but not all positive bias in the results.

Most of the CBS samples have been analyzed by Heritage Laboratories. Heritage does not use two columns for analysis. The 8082 method dictates the use of two columns (one for confirmation of PCBs), but allows for the use of one column in cases where experience has shown that PCBs are present and the Aroclor pattern is well known. This is the case with Heritage Labs and the majority of the samples CBS takes at the sites.

Some of the alternate labs used by the project (for example Antech) have much less experience with the CBS sites. They prefer to use two columns for analysis (as is specified by method 8082). In the case when two columns are used, they do routinely report the lower result from the two columns. This follows the historical precedent set by the CLP program requirements and data validation guidelines for that program. The rationale for this is that the higher results is most likely due to interferences.

Method 8082 does require the reporting of the higher of the two columns in situations where there is a significant difference between the columns (about 40%). Operationally, CBS flags all data where the two columns differ by more than 25%, and does report the higher value if significant differences exist unless there is an obvious bias to one of the results found during data validation. In that case the most correct result will be reported and the discrepancy/rationale reported in the data validation report.

The laboratory followed CLP protocol for the analyses performed, verifying that the techniques used were completed properly.

Issue 9: Concerns about 1989 Westinghouse High-Flow Tracer Study Results

COPA requests in Issue 9 an explanation of the data in the above-referenced study that is being used to conclude that 98.9% of the Lemon Lane Landfill site drainage is occurring through Illinois Central, Quarry, and ICG-I Springs.

This issue has been discussed with COPA and a separate review by Gareth Davies has been released and he is agreement with the U.S. EPA experts. Further information will be gathered and documented in the second operable unit regarding groundwater flow.

For the 1989 test, the fluorescent tracers were detected qualitatively on passive detectors, originally no spectrofluorometric analysis was conducted. The data utilized to determine that about 98.9% of the site drainage appears at these springs to the south (Illinois Central and Quarry) and northwest (Packinghouse Road, Packinghouse Culvert, and Slaughterhouse) are derived from a high-flow tracer test conducted in 1990, not 1989. A couple of problems have plagued the interpretation of the earlier 1989 high-flow dye trace data with respect to the tracer detections at several monitoring sites. First, the 3 fluorescent dyes used, namely fluorescein, optical brighteners and direct yellow 96, can have a high natural background. Second, in the case of fluorescein, other trace compounds in the water may yield apparent fluorescein detections on charcoal. With the exception of two rounds of grab sampling for bromide, no background sampling was conducted to effectively evaluate the several low level tracer detections reported from this test.

In 199O, the high-flow tracer test was repeated by CBS using several procedural improvements. These improvements included the following:

• Use of a fluorescent dye (Rhodamine WT) with a comparatively low natural background. Four rounds of fluorescent background determination prior to the tracer test.
• Quantitative determination of tracer dye using spectrofluorometric techniques.
• An expanded dye monitoring network

The 1990 tracer test, as summarized by CBS in the 1994 report entitled "Summary Report Lemon Lane Landfill Dye Tracer Tests," documented that 96.85% of the Rhodamine WT tracer dye recovered was detected at a monitoring point below Illinois Central Spring (the so-called Quarry location). An additional 1.31% may have been recovered at a culvert discharge known as ICG-1, downstream from the Quarry location, but subsequent investigation by CBS suggests that this detection was related to backflooding of the monitored location by Quarry Spring Branch. An additional 0.61% was recovered from the three northwest springs mentioned above. Together, all of these discharges comprise 98.8% of the total dye recovered.

Based on low-level tracer dye detections at 10 other locations observed in 1990, very small amounts of Groundwater peripheral to the landfill may be diverted to other drainage basins during storm water flow conditions. However, these tracer detections were sporadic, commonly associated with natural background, and many were single occurrences. These detections, if real, constitute only about 1% of the dye recovered, and indicate at best, a very minor and ephemeral connection to Groundwater in the vicinity of the landfill.

PCB sampling by CBS at the springs around Lemon Lane can be subdivided into the following three categories: (1) sampling of water at low-flow conditions; (2) sampling of water at high- flow conditions (storm event sampling); and (3) sampling of sediment.

Illinois Central Spring and Quarry Springs, ICG-1, Slaughterhouse, PH Road, and PH Culvert Springs were sampled by CBS for 8 consecutive weeks beginning in August 1995, and then monthly until July 1996 during low-flow conditions. Only Illinois Central and Quarry Springs had PCBs detected in these low-flow water samples. The nine peripheral springs (Stoney West, Stoney East, WN-1, WS-2, ICG-6, Snoddy, Bypass 37, Urban, and Crestmont) were sampled one time during low-flow conditions. Three springs were dry at the sampling time (WN- I, WS-2, ICG-6. The sample collected from Snoddy spring (which had a PCB concentration of 0.15 ppb) was actually the surface water flowing where commercial construction had destroyed the previous spring location. The remainder of the sample analytical results were non-detect for PCBs at 0.1 ppb.

Storm event sampling was first conducted by CBS in May 1995 before sampling in the landfill was conducted. Illinois Central Spring and Slaughterhouse Spring were sampled. Illinois Central Spring had a high concentration PCB pulse detected while no PCBs were found at Slaughterhouse Spring for that same storm event. A smaller storm event was sampled in October 1995 and this time low level PCB concentrations were detected at Slaughterhouse Spring.

In May 1996, sampling was conducted by CBS at Quarry Springs, downstream Clear Creek, and Illinois Central Spring. It was determined from this sampling event that Illinois Central Spring had the highest PCB concentrations and also that stations sampled downstream had PCB concentrations that were diluted compared to the Illinois Central Spring results. Subsequent storm event sampling has focused on Illinois Central Spring only.

In addition, sediment sampling was conducted at the 9 peripheral springs. Samples were collected at each spring orifice and four downstream locations where sediment deposition was observed. Except for a 0.67 ppm PCB result for a CBS split sample at Stoney West Spring and a similar low level PCB result for a split sample (collected by Earth Tech) at Urban Spring, no PCBs were detected at the peripheral springs. However, PCBs were detected at Slaughterhouse, Packinghouse Road, Packinghouse Culvert Springs, and also Illinois Central and Quarry Springs.

Issue 10: Review of Past Investigations

The U.S. EPA disagrees with COPA regarding the statement that questions on data quality for water and sediment exist. The previous dye trace data was evaluated and the U.S. EPA has no plans to reevaluate the data.

Issue 11: Review of Design Details

The U.S. EPA has sent the bid specifications and drawings to MVA Consulting. In addition, the U.S. EPA has provided a copy of the response to CBS comments on the water treatment plant. The US. EPA will keep COPA and others involved in the development of the plant monitoring plan as the information becomes available.

Issue 12: Improvement in the Information Repository

The Administrative Record when placed into the Monroe County Public Library contains an index which should make finding documents fairly easy. Updates to the Administrative Record along with additional Administrative Records for additional operable units will be placed in the library as they become available.

The material for the information repository located in the Monroe County Public Library has not been organized in the past. The U.S. EPA previously has not placed a cover sheet which references the documents contained in the package. Since COPA identified this problem, the information in the repository has been listed in a cover letter for easy access. In addition, COPA has suggested that documents be provided in electronic format. The U.S. EPA will make critical documents available in electronic format but cannot commit to providing all documents in electronic format. Please contact us when you would like a document in electronic format and the U.S. EPA will try to meet your request.