For over 50 years, aerosol sampling methods for workplace air monitoring in the United States and many other countries have relied on the use of closed-face filter cassette (CFC) samplers.⁽¹⁾ Established occupational exposure monitoring practice based on CFC sampling generally entails the use of a battery-powered personal sampling pump, which pulls air via vacuum through a filter housed within the cassette.⁽²⁾ For personal monitoring of airborne particles to measure their elemental content, such filter sampling is ordinarily carried out using membrane⁽³⁾ or depth⁽⁴⁾ filters. Subsequently, the filter deposit is normally analyzed by atomic spectrometric techniques after acid dissolution of the filter medium plus collected particles.^(3–5) Unfortunately, analysis of the filter deposit alone does not always adequately measure worker exposures to airborne metals and metalloids because it can underestimate the concentration of airborne particles, thus potentially giving a false assurance of protection from overexposure. This underestimation results from accumulations of sampled aerosol particles on internal sampler surfaces, a contribution to exposure that has not always been included in analytical procedures.⁽⁶⁾ Significant amounts of particulate matter up to and even exceeding quantities of material captured on the filter can be deposited on the internal surfaces of CFCs during sample collection, transport, and handling.^(6–13) This accumulation can be substantial and yet invisible to cursory inspection, so that its significance is not always recognized.

Collection of airborne dust samples in occupational environments using CFC samplers requires the evaluation of all aerosol particles that enter the sampler, whether or not they are deposited on the filter.^{(6, 8, 14)} Methods for incorporating this material along with the filter for subsequent analysis include rinsing,⁽¹³⁾ wiping,⁽⁸⁾ and within-cassette dissolution.⁽⁴⁾ However, these procedures can be labor intensive and may result in increased experimental uncertainty as well as uncorrectable systematic errors.

An internal capsule (fused to a filter) that would contain all collected airborne particles, and that could be digested easily without compromising the analysis through high background or matrix interferences, could be used in lieu of the above-mentioned labor-intensive procedures. Currently, CFC internal capsules made of polyvinyl chloride (PVC) and used for gravimetric analysis of collected particulate matter are commercially available.^{(2, 15)} However, PVC is extremely difficult to solubilize without sizable dilution, and thus, CFC insert capsules fabricated from this polymer are not appropriate for elemental analysis by means of standardized acid digestion and atomic spectrometric detection methods. Efforts have been undertaken recently to produce CFC capsule inserts made out of material that is more easily solubilized (at the request of the ASTM International Subcommittee on Workplace Air Quality);⁽¹⁶⁾ a prototype internal capsule made entirely of cellulosic material has been developed. Such a cellulosic capsule, when inserted into a CFC sampler, can be used to collect the entire aerosol sample, which is aspirated through the inlet of the cassette. Thereafter, the cellulosic insert can be completely digested in the same strong acid solutions that are customarily employed to dissolve mixed cellulose ester (MCE) filters, which historically have been widely used for airborne elemental sampling and analysis.

The overall goal of this effort was to evaluate and validate a method that accounts for all aerosol particles that enter the CFC sampler, thereby including material that is not otherwise measured by filter-only analysis procedures. A principal aim of this work was to carry out an interlaboratory study (ILS) to evaluate the analytical suitability of cellulosic capsule inserts for their use with traditional plastic air sampling cassettes. The ILS analysis entailed fortifying the filter capsule inserts with various loadings of metal-containing aerosols and sending them to volunteer laboratories. The inserts were then acid digested and analyzed for their elemental content by atomic spectrometry.

Tests performed on presumably identical materials in presumably identical circumstances do not, in general, yield identical results. This is attributed to unavoidable random errors inherent in every test procedure; the factors that may influence the outcome of a test cannot all be completely controlled. In the practical interpretation of test data, this inherent variability has to be taken into account. For instance, the difference between a test result and some specified value may be within that which can be expected to be due to unavoidable random errors, in which case a real deviation from the specified value has not been demonstrated. Similarly, the difference between test results from two batches of material will not indicate a fundamental quality difference if the difference is no more than that which can be attributed to inherent variability in the test procedure. Many different factors (apart from random variations between supposedly identical specimens) may contribute to the variability in application of a test method, including: (1) the operator, (2) equipment used, (3) calibration of instrumentation, and (4) environment (temperature, humidity, level of background contamination, and so on). Changing laboratories changes each of the above factors. The two practical extremes of precision are the repeatability (within-laboratory) and reproducibility (between-laboratory) conditions. This project yields data on both conditions by providing multiple test materials simultaneously to several laboratories. The protocol for this type of project has been codified in an ASTM International standard, ASTM E691,⁽¹⁷⁾ which describes procedures for determining method precision by means of inter-laboratory testing.

Reported results have been evaluated in accordance with ASTM E691.⁽¹⁷⁾ The analysis and treatment of the ILS test results have three purposes: (1) to determine whether the collected data are adequately consistent to form the basis for a test method precision statement, (2) to investigate and act on any data considered to be inconsistent, and (3) to obtain the precision statistics on which the precision statement can be based. The statistical analysis of the data for estimates of the precision statistics is simply a one-way analysis of variance (within- and between-laboratories) carried out separately for each level (material). The procedure to be followed is that detailed in the standard.⁽¹⁷⁾ The fundamental precision statistics of the ILS are the repeatability standard deviation and the reproducibility standard deviation. Other statistical parameters may then be calculated from these standard deviations. The repeatability standard deviation is calculated from where

s_r = the repeatability standard deviation

A provisional value ((s_R)*) for the reproducibility standard deviation (s_R) is computed from where

s_x = the standard deviation of the tabulated cell averages,

The provisional value is then compared with the repeatability standard deviation, and whichever value is larger is taken to be the repeatability standard deviation (and thus the same value can appear in both columns). Whereas such an analysis can be invalidated by the presence of severe outliers, it is necessary to first examine the consistency of the reported data as detailed in Part 17 of ASTM E691, “Flagging Inconsistent Results,” using the critical values of the h and k consistency statistics (calculated according to Paragraph 15.7, Eqs. 8 and 9 of ASTM E691) at the 0.5% significance level. After removal of outliers by this procedure, values of s_r and s_R are recalculated and then used to derive the 95% repeatability and reproducibility limits (r and R) according to r = 2.8s_r and R = 2.8s_R The final form of the precision statement is prepared in accordance with ASTM E177.⁽²⁶⁾

ILS results for measurements of cellulosic filter capsules spiked with Pb-containing performance evaluation materials, which appeared in different form in a preliminary report,⁽¹⁴⁾ are presented again in Table VI. Repeatability values were all at about 16% and below, reproducibility estimates ranged from about 5% to 24%, and recoveries were generally quantitative within ±10% of target values. The values for the high Pb aqueous spike are discussed in the reference.⁽¹⁴⁾ Overall, the ILS performance data for Pb demonstrate that cellulosic internal capsules attached to MCE filters can be used in place of wiping and/or rinsing techniques that are otherwise required to account for CFC internal non-fllter deposits. However, data from two international laboratories could not be used in the analysis because the intense jostling in international transportation had caused the loose soil and paint materials to escape from the interior of the cellulosic filter capsules.

All nine participating laboratories in the multi-element study returned results that are presented in Table VII. The samples were collected by filtration from an aerosol atmosphere, and the results are presented in terms of air concentrations. The loadings, however, are a function of both concentration and sampling time. One laboratory neglected to follow provided instructions and separated the filters from their capsules and analyzed only the filters. While this laboratory later also analyzed the capsules, this was considered too far removed from the standard procedure being evaluated for the results to be included; thus, these results were not used.

Multi-element data from the ILS round are presented in Table VII; eight participants returned acceptable results for most elements, as has been explained above. Reported results returned from the laboratories were provided in units of mass (μg) per filter. These results were recalculated as air concentrations (μg/m³) using the measured Aow rates through the individual samplers to remove the effect of minor variations in Aow rates. Concentration values were then used in succeeding statistical computations. Results are provided for the evaluation of all eight laboratories unless outlier laboratories were removed as was indicated by h and k statistics.⁽¹⁷⁾ Two cases where outliers were identified but could not be removed were indicated. Overall means are given in units of elemental concentration in air (μg/m³).

Not all laboratories analyze As on a routine basis, so not every laboratory reported values for this element. The results from those laboratories that did report As were highly variable, and given that the minimum number of six participants was not reached, the As data were not included in the ASTM E691 statistical analysis. For the lowest As level, the reported values varied from 0.07 to 3.0 μg/sample, for the second level from 0.99 to 5.0 μg/sample, and for the highest level from 5.0 to 10 μg/sample. The target loadings at each level were anticipated to be 0.15, 1.5, and 7.5 μg/sample, respectively, and thus, the reported range of As values included the target values. However, the large variations encountered here were not anticipated, since such variability was not obvious in a prior round-robin study where As loading levels were 0.25, 5, and 100 μg/sample.⁽²²⁾ However, the As levels in this study are similar to the two lowest levels in the previous study and high variation was encountered at the very lowest level in that work. Previously, two laboratories had analyzed blank cellulosic filter capsules from the same batch as used here and had reported all samples with arsenic < 0.8 or < 1.4 μg As per sample, respectively. While these reporting limits are higher than the lowest level of anticipated sample loading provided, they are at least an indication that background was not a contributor to variance in the higher loadings.

The lowest level in this study for Cd was also problematic, being below the reporting limit for two of the participating laboratories. A third laboratory reported Cd results ranging from <0.06 to 0.11 μg/sample, and this laboratory could have been classified as an outlier. Since removing this laboratory would have left only five laboratories in the study (less than the minimum of six specified by ASTM E691), these results were also not included in the evaluation. However, the range for all the other Cd values (0.11 to 0.14 μg/sample, compared with an expected 0.15 μg/sample) suggests that the outcome would probably have been within acceptable limits. Nevertheless, these results were not used.

Finally, several laboratories were identified as outliers in the analysis of one or more levels of one or more elements using critical values of the h and k statistics, as per the ASTM E691 procedure.⁽¹⁷⁾ This finding was spread fairly evenly among the laboratories, and no one laboratory was consistently an outlier in any one element. Identified outlier laboratories were removed and the statistics recalculated. In two cases, this was a meaningless procedure, as removal of suspected outliers led to yet additional outliers. This situation is especially prevalent when the variance in the inter-sample analysis is very tight and considerably smaller than the between-laboratory variance. Hence, the results in these cases are given for all laboratories and the issue is noted. In the other cases, removal of the outlier laboratory or laboratories generally (but not always) improved the variance, and this is how the results are reported, with a notation that outliers were removed. No more than two laboratories have been removed from any level for any particular element, so that the final statistics are provided for at least the ASTM E691 -specified minimum of six laboratories in every case.

The repeatability standard deviation (r) and reproducibility standard deviation (R) are functions of the absolute value of the mean, which is typical for results away from the limit of detection. In the final columns of Table VII, r and R are divided by the overall mean to give more meaningful estimates of variability. With one exception (i.e., Cd Level 2), all results for repeatability are considered acceptable (i.e., r/x <0.25). The low loading level (<5 μg Cd per sample) explains the higher variability for this sample. The RSD for reproducibility (R/x) can be used as an estimate of the overall variance of the method. Most of the observed ILS elemental analysis results for reproducibility (Table VII) are acceptable (i.e., <0.30). We cannot postulate an explanation for the higher interlaboratory variability observed for Ni measurements. All values of s_x/x are <0.20 and compare favorably with variabilities typically observed in interlaboratory multi-element analysis of air samples (22–25, 27)

In carrying out this work it was observed that some CFCs, as provided by the manufacturer, were found not to fit tightly together, or else the cassettes fit tightly but sealed above the edge of the internal capsule. Also, about 3% of the assembled cassettes with support pads and cellulosic filter capsules had very high pressure drops and were thus unusable. This was believed to result from regions within the cellulose ester membrane filter with limited porosity. A similar problem with MCE filters has been reported previously,⁽²⁸⁾ and such matters highlight the need for ongoing quality assurance. Even if these (and other) quality issues are corrected by the manufacturer, it is recommended that all assembled CFCs with or without internal capsule inserts be checked by the laboratory assembling the cassettes (or by the user in the field) to ensure that (a) the internal capsule is securely sealed within the cassette (this can be checked by pulling air through a condensation nucleus counter in line with the CFC, as has been proposed for all sampling cassettes)⁽²⁹⁾; and (b) that the CFC inlet plug fits both the entry orifice of the cassette and the insert so as to properly seal the internal capsule. Assembled capsule inserts that do not meet these checks should be rejected for further use. Also, to ensure that the seal remains good after sampling, it is further recommended that the specific plug/cassette combination be retained for reassembly by, for example, placing the plug in a secure labeled container to match the cassette while sample collection is underway.

To ensure that personal exposures to aerosols are fully quantified, it is necessary to account for all particles crossing the plane of aspiration of the sampler, whether they eventually reside on the filter or elsewhere within the sampler. Several methods have been proposed for accounting for internal non-filter deposits, including within-cassette dissolution, rinsing and/or wiping of the surfaces, and an internal capsule that will contain all particles. Through the experiments described here, we have demonstrated the potential of acid-soluble cellulose acetate capsules bound to mixed-cellulose ester filters for the collection and analysis of aerosols containing metal particles entering the 37-mm CFCs. These particular capsules are not suitable for gravimetric analysis because of high variability in mass associated with changes in humidity.

We have shown that the background levels of many metals important in industrial hygiene, together with any matrix effects from the capsule material, are sufficiently low to allow repeatable and reproducible analyses down to levels approaching typical limits of quantitation for MCE filters. Internal capsules have advantages over rinsing or wiping in that they are less labor intensive and less likely to show variability due to variation in individual manual technique. Should acid-soluble cellulose acetate capsules bound to MCE filters of appropriate quality become commercially available, they can be used for sampling and analysis of many elements according to NIOSH procedures. It is anticipated that the results from this and subsequent studies will be used in methods that are or will be published in the NIOSH Manual of Analytical Methods and through consensus standards bodies such as ASTM International and the International Organization for Standardization (ISO).