RTI/04562/78L-F AMERICAN INDUSTRIAL HYGIENE ASSOCIATION BULK ASBESTOS PROFICIENCY ANALYTICAL TESTING PROGRAM FINAL REPORT TO LABORATORIES - TEST ROUND A78-109 Prepared for the American Industrial Hygiene Association 2700 Prosperity Avenue, Suite 250 Fairfax, Virginia 22031 Prepared by Bruce W. Harvey, Stacy S. Doorn, J. Todd Ennis, Lisa C. Greene, Wayne G. Winstead, and Maurice Gerald Environmental and Industrial Sciences Division and Elizabeth E. Robbins Research Computing Division RTI International 3040 Cornwallis Road Research Triangle Park, NC 27709-2194 April 2009 This report may contain confidential information. If you are not the intended recipient, you are hereby notified that any review, dissemination, distribution, or duplication of this report is strictly forbidden. If you are not the intended recipient, please destroy all copies of the report and contact RTI International.
22
Embed
AMERICAN INDUSTRIAL HYGIENE ASSOCIATION … AMERICAN INDUSTRIAL HYGIENE ASSOCIATION BULK ASBESTOS PROFICIENCY ANALYTICAL TESTING PROGRAM FINAL REPORT TO LABORATORIES - TEST ROUND A78-109
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
RTI/04562/78L-F
AMERICAN INDUSTRIAL HYGIENE ASSOCIATIONBULK ASBESTOS PROFICIENCY ANALYTICAL
TESTING PROGRAM
FINAL REPORT TO LABORATORIES - TEST ROUND A78-109
Prepared for the
American Industrial Hygiene Association2700 Prosperity Avenue, Suite 250
Fairfax, Virginia 22031
Prepared by
Bruce W. Harvey, Stacy S. Doorn, J. Todd Ennis,Lisa C. Greene, Wayne G. Winstead, and Maurice Gerald
Environmental and Industrial Sciences Division
andElizabeth E. Robbins
Research Computing Division
RTI International3040 Cornwallis Road
Research Triangle Park, NC 27709-2194
April 2009
This report may contain confidential information. If you are not the intended recipient, you are herebynotified that any review, dissemination, distribution, or duplication of this report is strictly forbidden. If
you are not the intended recipient, please destroy all copies of the report and contact RTI International.
Lot B: This material consists of a self-adhesive floor tile approximately 1/16 inch (1.6millimeters) thick. The upper surface is distinguished by a textured faux stoneor marble pattern with shades of yellow, beige, brown, and rose. The tintedportion of the tile constitutes approximately one-third of the tile’s thickness. Theremaining two-thirds of the tile is off-white in color. Granular minerals such asquartz and calcite are clearly visible in the tile’s broken edges. The vinyl matrixis relatively soft. This tile is more flexible than older brittle asphaltic floor tiles.Fibrous material is not readily apparent in observation of broken edges. A thinlayer of tacky adhesive is present on the bottom surface of the tile. Theadhesive does not appear to contain fibrous components.
6
Polarized Light Microscopy (PLM)
Several subsamples were randomly extracted from the repository container of each
positive test sample lot for detailed PLM analysis prior to gravimetric sample reduction.
For each asbestos component encountered in each positive lot, the following optical
properties were determined: morphology, aspect ratio, color, pleochroism, refractive
indices (RIs), birefringence, central-stop dispersion staining colors, extinction type and
angle, and sign of elongation. These properties are summarized in Table 2.
Table 2. Optical Properties of Asbestos in Test Sample Lots
Optical Property
Lot A Lot B Lot C
Chrysotile Chrysotile Chrysotile Actinolite
MorphologyWavy, curvilinear, with
“knees”Fine, wavy, with
splayed endsWavy, curvilinear, with
“knees”Straight,
needle-like
Aspect Ratio
Color
Pleochroism
RefractiveIndices
Birefringence
Central-StopDispersion
Staining Colors
Extinction
Sign ofElongation
Brief material descriptions acquired during the PLM examination are as follows:
7
Lot C: Slide mounts of this material are dominated by gypsum-based matrix, amicaceous mineral, and chrysotile asbestos. The matrix, which coats all othercomponents and makes their identification difficult, is moderately effervescentin room-temperature HCl. The micaceous mineral displays a variety of colorsranging from dark brown to green, but most commonly golden brown. Thechrysotile exhibits positive sign of elongation and parallel extinction. In 1.550 RIliquid, the chrysotile displays typical blue to purple dispersion staining colors.Chrysotile is visually estimated to constitute 3% of the material. A trace amountof an amphibole mineral occurs as small needle-like fibers and bundles. Thismaterial is identified as actinolite and is inhomogeneously distributedthroughout the material.
Lot D: Slide mounts of this material reveal a homogeneous mixture dominated by fine-grained, equant-shaped mineral fragments with moderate birefringence. With the addition of gentle heat from the microscope light source, the vinyl matrixdissolves in the 1.550 RI liquid in which it is mounted. Translucent occlusionscan be observed in the oil mount. These occlusions appear to be consistentwith the stereoscopic observation of a polished top surface. No fibrous materialis visible.
Gravimetric Sample Reduction
Representative subsamples from all four lots were subjected to gravimetric sample
reduction, by way of acid dissolution and/or low-temperature ashing, to confirm
qualitative results for organic and acid-soluble components and to refine
semiquantitative results derived from stereomicroscopy and PLM examinations. The
results of these test sample reductions are described below. Unless otherwise noted,
losses are quoted as a percentage of the sample’s starting weight.
Test sample compositions rendered low-temperature ashing useful on Lots A, B,
and D. Results of this sample reduction are described below. For the reduction, a
minimum of eight preweighed subsamples were ashed at approximately 480°C for 15
hours. Each residue was allowed to return to ambient temperature (stabilize) before
reweighing.
8
Lot A: The average weight loss was 26.9%, attributable to the combustion of celluloseand dehydration of mineral matrix.
Lot B: The average weight loss was 21.6%, attributable to the combustion of organic(vinyl) binders.
Lot D: The average weight loss was 51.6%, attributable to the combustion of organic(vinyl) binders.
Test sample compositions rendered acid dissolution reduction useful on all four
lots. For the reduction, a minimum of eight preweighed subsamples were treated with
3N HCl. For Lots A, B, and D, the residue after ashing was used. After the acid
treatment, each residue was thoroughly washed and allowed to dry (stabilize) before
reweighing. The approximate loss of the soluble binder/matrix was determined for each
subsample, with results as follows:
Lot A: The average weight loss was 77.3% (equivalent to 56.5% of the whole testsample), attributable to the dissolution of calcite and gypsum. The residue isdominated by perlite, fine-grained minerals, quartz, and pigments. Chrysotile isestimated to constitute approximately 15-20% of the residue, for a whole-sample equivalent of 2.5-3.3%.
Lot B: The average weight loss was 86.2% (equivalent to 67.6% of the whole test sample), attributable to the dissolution of calcite. The residue is dominated byfine-grained matrix and pigments. Chrysotile is estimated to constituteapproximately 20-30% of the residue, for a whole-sample equivalent of 2.2-3.2%.
Lot C: The average weight loss was 62.7%, attributable to the dissolution of gypsumand bassanite (both calcium sulfate minerals). The residue is dominated bymica and fine-grained matrix. Chrysotile is estimated to constituteapproximately 10-20% of the residue, for a whole-sample equivalent of 3.7-7.5%.
Lot D: The average weight loss was 97.9% (equivalent to 47.4% of the whole test sample), attributable to the dissolution of calcite. The residue is dominated byfine-grained matrix, pigments, quartz, and mica. No asbestos is visible in theresidue.
9
X-Ray Diffraction (XRD)
Treated (by ashing and/or acid dissolution; see the previous section) and untreated
subsamples of all four lots were extracted for XRD analysis. All treated and untreated
subsamples were ground with a mortar and pestle, and powder mounts were prepared.
Scans were run from 5°22 or 8°22 to 62°22 at a scan rate of either 0.01°22 per second
or 0.02°22 per 0.6 second. Differences in scan rates are attributable to differences in
XRD instrumentation and have no impact on RTI’s interpretation of the diffractograms.
Following are the results of the XRD analyses:
Lot D: No asbestiform phases were detected in scans of the untreated and treatedmaterial. Confirmed nonasbestiform phases included calcite (untreated only),quartz (treated only), muscovite mica (treated only), rutile [titanium dioxide](treated only), and microcline (treated only).
10
Quantitation of Asbestos in Positive Test Sample Lots
The concentration of asbestos in each of the three positive lots was determined by
point counting, quantitative XRD, and gravimetric sample reduction, each where
applicable. An abundance of fine-grained pigments in residues from gravimetric
reduction of the Lot B floor tile material adhered to
Means and standard deviations were calculated for the point counts, XRD
quantitation values, and gravimetric residue estimates for the asbestos type(s) in each
sample and used to develop two-sided tolerance limits (acceptance ranges). Limits
were chosen so as to have 99% confidence that 95% of the reported values would be
deemed acceptable. The final acceptance range for a sample used the lowest value
among the minimum values and the highest value among the maximum values for all
techniques used. Semiquantitative results and acceptance ranges are shown in Table 3.
The number of replicate analyses used in each calculation is also indicated.
Section V - Summary of the Test RoundOf the 239 laboratories enrolled, 230 submitted results of analyses for the test
round, for a response rate of . This rate was slightly higher than the 94% average
participation rate for the 55 previous test rounds conducted under the proficiency rating
format. A laboratory not returning results was deemed “failing” (F) for the test round
unless it had notified AIHA beforehand of its intent not to participate in the test round
and provided a qualifying reason for the nonparticipation. One laboratory provided such
notification to AIHA. Of the 230 participating laboratories, 227 (99%) submitted analysis
via RTI’s Web site for the program.
The eight laboratories that did not submit results by the deadline date or have an
AIHA-approved exemption were contacted by RTI to confirm their intent not to
participate. One laboratory indicated that it would submit late results but did not. One
laboratory wishes to remain enrolled but will not submit results before Test Round 80.
11
Table 3. Semiquantitation Means and Ranges, and Acceptance Ranges
SemiquantitationMethod
Lot A Lot B Lot C
Chrysotile Chrysotile Chrysotile
Point CountMean (%)Range (%)
Replicates (#)
2.31.4 to 3.0
7not applied
5.13.5 to 6.3
8
SemiquantitativeXRD
Mean (%)Range (%)
Replicates (#)
2.72.0 to 3.9
8
3.63.2 to 4.3
10
9.06.8 to 9.8
9
GravimetryMean (%)Range (%)
Replicates (#)not applied
2.81.6 to 3.7
20not applied
99/95 AcceptanceRange (%)
Trace to 8%,inclusive
Trace to 8%,inclusive
Trace to 15%,inclusive
One laboratory returned its test samples without explanation. The other five laboratories
did not reply to RTI’s voice messages or e-mails.
Tables 4 and 5 illustrate the total test round effort, as generated by the results of
analyses submitted by participating laboratories. Table 4 shows the penalty point totals
incurred by passing (P) and failing (F) laboratories based on the grading criteria
described on page 13 of this report. A laboratory was rated F for incurring 100 or more
penalty points or for not participating. The total numbers of P and F laboratories are also
indicated. For the test round, 203 laboratories (85% of the total enrolled, 88% of the
total participating) were rated P. Of the 35 F laboratories for the test round, 26 were
rated so for incurring classification errors, with or without other error types, and 1
resulted from a combination of asbestos identification errors and/or semiquantitation
errors, without classification errors. The eight laboratories that did not submit results by
the deadline or have an AIHA-approved exemption received F ratings for
nonparticipation. This was the first test round of participation for 3 of the 35 F
semiquantitation errors are denoted by “ID” and “SQ,” respectively. Totals of all error
types incurred for the test round are shown at the bottom of the table.
The following evaluation criteria were used to assign these classification,
identification, and semiquantitation errors:
Proficiency Grading Criteria Penalty Points
Failing to submit analysis results Automatic NP
Failing to report asbestos in a positive sample (FN) 100
Reporting 1% or greater asbestos in a blank sample (FP) 100
Failing to report the correct asbestos type, in a one-type sample (ID) 50Failing to report second asbestos type, in a multi-type sample (ID) 50Failing to report third asbestos type, in a multi-type sample (ID) 25
Reporting 1% or greater incorrect asbestos type (ID) 45/type
Reporting trace asbestos in a blank or trace incorrectasbestos type(s) in a positive sample No penalty
Failing to report trace asbestos when RTI QC confirmsit in all samples 25/type
Failing to report trace asbestos when RTI QC does notconfirm it in all samples No penalty
Per sample, first asbestos semiquantitation outside acceptance range 20Per sample, second asbestos semiquantitation outside acceptance range 10Per sample, third asbestos semiquantitation outside acceptance range 5
Section VI - Individual Laboratory ResultsPlease refer to the computer printout on page 14 for a tabulation of your individual
laboratory results and a comparison of those results with the reference laboratories’
analyses. The total penalty points incurred by your laboratory during this test round and
the status of your laboratory as a result of your performance this test round are listed in
the upper right-hand corner of the form. The number of penalty points per sample is
highlighted in the section entitled “Analysis Results for Laboratory _____.”
14
One-page
“computer printout of graded results”
to be inserted here
15
A laboratory achieved a program status of proficient (P) if two of its three most
recent test rounds were rated P. A laboratory also achieved a program rating of P if it
had participated in only two test rounds but had been rated P in each. A laboratory was
rated nonproficient (NP) for the program if it was rated F for more than one of the three
most recent test rounds. If a laboratory’s program rating is listed as not applicable (NA),
that laboratory has not participated in the number of test rounds required to establish a
program rating. Through the completion of this test round, 204 laboratories (85% of the
total enrolled) have a P program rating, 25 (11%) have an NP program rating, and 10
(4%) have no program status because they have not completed the requisite number of
test rounds.
Section VII - Analysis Problems and SuggestionsThe following appraisals are provided concerning trends or patterns seen in sample
classification, asbestos identification, and semiquantitation errors for each of the four
test sample lots.
A total of 30 classification errors were incurred, for an overall classification error
rate of 3.3%. This was only slightly higher than the program’s 3.1% historic classification
error rate. Twenty-four false negatives resulted in a false negative error rate of 3.5%;
this was slightly higher than the program’s historic false negative error rate of 3.0%. Six
false positives resulted in a false positive error rate of 2.6%; this was considerably lower
than the program’s historic false positive error rate of 3.3%.
Twenty-one of the 24 false negatives were incurred on Lot B, a vinyl floor tile
containing approximately 3% chrysotile. These false negatives were easily avoidable if
the laboratory had the ability to perform a gravimetric matrix reduction prior to looking
for and quantifying any asbestos present. Ashing and acid dissolution removed
approximately 22% and 68% of the sample’s weight, respectively. Chrysotile was easily
detected in the residue, constituting 20-30% of the remaining material.
The results on Lot B illustrate that laboratories not routinely performing gravimetric
matrix reduction on floor tiles potentially run a high risk for failing to detect asbestos
(when present) and incurring an avoidable false negative. RTI recommends that
laboratories employ a full spectrum of sample treatment processes, such as those
16
described in the EPA Test Method for the Determination of Asbestos in Bulk Building
Materials, EPA/600/R-93/116. Performing ashing and acid reduction serves to
concentrate any asbestos present into the residue, where it is more easily detected and
better quantified.
If a laboratory does not have the equipment or capability to perform these sample
reductions, it may be well-advised to submit floor tile samples to another laboratory with
gravimetric reduction and/or transmission electron microscopy (TEM) capability, in order
to confirm or refute the negative PLM results. If a laboratory does this as standard
practice on all floor tiles, it may seek a permanent “floor tile waiver” from AIHA. This
waiver allows the laboratory to choose to either: 1) not submit results on any floor tile
PT sample, or 2) submit results on floor tile PT samples but not have them graded by
RTI.
Six false positives were incurred on Lot D, a sheet linoleum containing no fibrous
components. Five laboratories reported the presence of chrysotile, in amounts ranging
from 2% to 15%. A sixth laboratory reported 10% each of chrysotile and tremolite. That
laboratories would report any asbestos type at all is surprising, as RTI detected no
fibrous material to be present in the material. Three of the six laboratories also incurred
false negatives on Lot B, suggesting the possibility that analysis results for the two
samples were transposed in the process of being reported.
Twenty identification errors were incurred, for an identification error rate of 2.9%.
This was less than one-half the program’s historic error rate of 6.5%.
Fifteen of the 20 identification errors were incurred on Lot A, a textured ceiling
spray containing approximately 3% chrysotile. Nine errors were assigned for reporting
amosite or tremolite but not reporting the chrysotile known to be present; six errors were
incurred for reporting amosite or tremolite in addition to the chrysotile.
The other five identification errors were divided between Lots B and C. Both errors
on Lot B were incurred for reporting tremolite in addition to the chrysotile known to be
present. On Lot C, two errors were assigned for reporting anthophyllite but not reporting
the chrysotile known to be present. The other was incurred for reporting amosite in
addition to the chrysotile. RTI and the two reference laboratories reported the presence
of trace-level amphibole exhibiting inclined extinction.
17
The number of semiquantitation errors was low compared to that in recent test
rounds, but not unexpected given the specific nature of the test samples in this test
round. On Lot B, all 34 semiquantitation errors were incurred by laboratories reporting
chrysotile in amounts exceeding the upper limit of RTI’s acceptance range for that
asbestos type. Ten errors were incurred on Lot C, all for reporting asbestos types in
amounts exceeding the upper limits of RTI’s acceptance ranges. Two errors were
incurred on Lot B, also for reporting asbestos types in amounts exceeding the upper
limits of RTI’s acceptance ranges.
Section VIII - Test Method/Analytical Technique SummaryRTI solicited information about the test methods and analytical techniques
employed by the program participants on this test round’s test samples. The cited
methods and techniques are listed in Table 6. RTI then correlated false negative, false
positive, asbestos identification, and asbestos semiquantitation errors to the
method/technique used. Based on the recommendations of its statisticians, RTI
performed two simple statistical analyses – an analysis of variance (ANOVA) and a
mean squared error (MSE) test – of the semiquantitative data on positive sample lots.
The ANOVA determines if there is a statistically significant difference among the
consensus means of the reported values, for each test method/technique. Its use is
appropriate when comparing multiple data sets of widely varying numbers of individual
measurements. If the ANOVA indicates that a significant difference exists among the
results for the methods/techniques used, the MSE test attempts to identify the
method/technique combination yielding the best semiquantitative results by comparing
the accuracy of reported values to the true value for the test sample and also factoring
in the precision of the reported values. If a statistically significant difference is found, the
results of the evaluation of the data are included in this report.
It appears that the relatively small number of errors of any type make presentation
of these comparative data on a per-sample basis to be of limited value and practicality
to participating laboratories. Therefore, and effective with Test Round A71-207, RTI has
elected to present the error type/frequency data for each test method used in
cumulative terms for all samples in the test round. These data are provided in Table 7.