Envlron. Sci. Technol. 1992, 26,2101-2104
Industry-Wide Performance in a Pilot Performance Evaluation Sample Program for Hazardous Materials Laboratories. 2. Precision and Accuracy of Polychlorinated Biphenyls David Eugene Klmbrough,' Rustum Chin, and Janlce Wakakuwa
Southern California Laboratory, California Department of Health Services, 1449 West Temple Street, Los Angeles California 90028-5698 The California Department of Health Services, through the Environmental Laboratory Accreditation Program (ELAP) and Southern California Laboratory developed a pilot proficiency evaluation sample program for the analysis of PCBs. One of the goals of the program was to study the overall performance of the hazardous materials laboratory industry in terms of accuracy and precision of PCB analysis. A set of five spiked soil performance evaluation samples were sent out to 20 governmental laboratories which acted as reference laboratories. Then the same soils were sent to 153 accredited laboratories. The results showed a nonnormal distribution with low bias at high concentrations and a high bias at low concentrations. The results indicate widespread and systematic errors in the analysis of PCBs.
W
Introduction There is a great amount of regulatory and public concern for polychlorinated biphenyls; their accurate and precise analysis is a matter of considerable significance. Despite this, several studies have shown a large variance and significant low bias in interlaboratory studies (1,2).Readers familiar with the U.S.Environmental Protection Agency's (EPA) water pollution (WP) performance evaluation sample results for PCBs will note the same pattern. This paper is the second part of a series of papers (3) describing an interlaboratory study of 199 hazardous materials laboratories. Here the preparation and validation of PCB-spiked multivial performance evaluation samples and the accuracy and precision of the environmental laboratory industry for PCB analysis are examined. Experimental Section (A) Experimental Design. There were three parts to this study. First, the performance evaluation samples were prepared. Second, they were validated. Third, they were then distributed among the accredited laboratories. (B) Performance Evaluation Sample Preparation. Aroclor 1260 was used to spike a native soil. The soil was analyzed and found to be free of PCBs, although low-level interferences from decomposed vegetable matter were detected. The soil was milled, sieved, and autoclaved (to kill any bacteria that might be present). Approximately 150 laboratories are certified for PCB analysis by the California Environmental Laboratory Accreditation Program (ELAP). Ten kilograms of each performance evaluation sample was prepared so that each laboratory could be provided with about 25 g per performance sample. These five samples were prepared as described in Table I. Two solutions were prepared; one contained 10 g of Aroclor 1260 in 1L of hexane, while a second was made from a 1:lOO dilution of the first solution. The Aroclor 1260 was made up from neat PCB, checked against EPAderived standards, and found to be within 5% of the expected value. These solutions were spiked as described in Table I. The samples were slurried with n-hexane, homogenized, and dried at room temperature with periodic 0013-936X/92/0926-2101$03.00/0
Table I. Preparation of Performance Evaluation Samples sample finalconc PCB, mg/kg mass of PCB, g solution spiked, m L dilution
F
G
H
I
100 1000 100
10 100 10
1.0 10 100 1:100
0.1 1.0 10 1:100
Solutions spiking material
mass, g
vol, L
conc, g/L
Aroclor 1260
10.0
1L
10
dilutions 1:"
mixing. The samples were then analyzed at the Southern California Laboratory in duplicate and found to within control limits. (C) Validation. The samples were then shipped to 20 laboratories that were not accredited by the ELAP. There were many other government or government-affiliated laboratories who agreed to do the work gratis. (D) Distribution. The samples were then shipped to 154 accredited laboratories. The laboratories were required to use only approved EPA methods (see below). Each sample was tagged with labels which had identification numbers. This was used to track them at an individual laboratory. To minimize comparisons of results by the laboratories, the ID numbers were random except for the third digit: sample
ID no.
F G H I J
XXOX or XX5X XXlX or XX6X XX2X or XX7X XXBX or XX8X XX4X or XXSX
Each laboratory was sent a letter dated April 15,1991, 2 weeks in advance of shipping the samples. The letter was addressed to the laboratory director and informed him or her to prepare for the arrival of the performance evaluation samples. It also contained the deadline for returning the results, which was June 22,1991. Instruction sheets and reporting forms were prepared and made into sets for the different laboratories. (E) Analytical Methods. At the Southern California Laboratory, the PCBs were analyzed in duplicate using EPA SW-846 ( 4 ) method 3540 using dichloromethane in a Soxhlet extractor, exchanged to hexane. The hexane was then cleaned using a Florisil column, method 3620. The hexane was analyzed using a capillary column gas chromatograph and electron capture detector, method 8081. The reference laboratories used the three types of extractions, Soxhlet, sonication, and shakers. Four types of solvent systems were used dichloromethane (methylene chloride) exchanged to hexane, 1:l dichloromethane-propanone (acetone) exchagned to hexane, 1:l propanonehexane, and hexane. All of the laboratories used gas chromatography to quantify the PCBs, although some used
0 1992 American Chemlwl Society
Environ. Scl. Technol., Vol. 26, No. 11, 1992 2101
Table 11. Comparison of Means (pg/g) and Variance sample
mean
F
reference labs SD
85.6 8.37 0.847 0.109
G H I critical 95% critical 99%
25.1 2.52 0.283 0.039
N
mean
20 20 20 17
69.9 7.01 0.764 0.168
Table 111. Rates of Out of Control Results (%) reference labs highs lows LT"
range, mg/kg 125-75 12.5-7.5 1.5-0.5 0.15-0.05