Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Appendix B Additional Field Study Information
Additional Field Study Information PM Sample Collection Protocols Airborne PM samples were collected using Model PQ100 portable samplers (BGI, Waltham, MA). The PQ100 is a U.S. EPA Federal Reference Method (FRM) for PM10 sampling and for this study was used with BGI Very Sharp Cut Cyclones (VSCC) to achieve PM2.5 cutpoints. A louvered inlet with PM10 impactorâthe standard configuration for ambient PM10 samplingâwas used upstream of the PM2.5 cyclone. Total Suspended Particulate (TSP) samples were collected using PQ100 samplers with BGI TSP inlets. Samples were collected onto 47mm, 2.0 µm pore size Teflon® filters (PT47, MTL Corp., Minneapolis, MN). Prior to use, all filters were visually inspected for puncture holes or other impurities. Filter holders and screens were cleaned daily with de-ionized water. After sampling, the filters were again visually inspected for puncture holes and other irregularities and were then placed in a freezer for storage. All filters were transported using ice packs and insulated shipping containers. The BGI samplers were operated at an actual flow rate of 16.7 LPM, which is the design flow rate for the aforementioned inlets. Flow rates were calibrated using a NIST- traceable flow meter (deltaCal, BGI, Waltham, MA). Flow rates were checked nominally every 3 days and with recalibration performed if the deviation from setpoint exceeded 5%. In such cases, the flow rate was rechecked after the end of the sampling event. ICP-MS Analysis for Total PM-Pb and Pb Isotopes Exposed filters were digested and analyzed for Pb by ICP-MS. Filters were first cut from their support rings and placed into polypropylene vials. Particulate matter on the filters was extracted using a two-stage digestion process. In the first stage, nitric and hydrofluoric acid were added to the sample vials and the samples were digested for 2 hours using a hot-block (ModBlock, CPI International, Santa Rosa, CA) at 90°C. After the first stage, samples were allowed to cool and then boric acid was added to enhance recovery and complex the excess hydrofluoric acid. The second-stage digestion was then performed, again with the hot-block at 90°C for 2 hours. QA/QC measures for each set of digestions included (1) a reagent blank, prepared using the same amount of acid that was added to a vial which was subjected to the digestion process; (2) a spiked reagent blank, prepared identically to the reagent blank and spiked with a known amount of the multi-element standard before the first hot-block digestion stage; (3) a filter blank that used an unexposed filter that was subjected to the same digestion method as the exposed filter samples; and (4) a digested acid matrix matched blank (DAMMB), prepared identically to the reagent blank but in greater volume and using different glassware for the purpose of making ICP-MS calibration and concentration verification standards. Digested samples and QA/QC blanks were filtered using 0.45 µm pore size Acrodisc filters (Pall Corp., Port Washington, NY) and then B-1
diluted to a final volume of 15 ml with de-ionized water to have a nitric acid content of 5% (v/v). Samples were analyzed using an ELAN DRC II Inductively Coupled Plasma Mass Spectrometer (PerkinElmer, Waltham, MA). The ICP-MS nebulizer flow rate and lens voltage was optimized prior to each analysis batch. The instrument was calibrated using a number of DAMMB samples with known amounts of Pb. Calibration concentrations ranged from 0.001 to 50 ppb. For each analysis batch, the performance of the ICP-MS was re-evaluated after every 10 samples using blank DAMMB solution and a concentration verification solution made from DAMMB and 1ppb Pb. Total Pb (and select other elements) signal intensities were corrected using rhodium and rhenium internal standards, with rhodium used to correct the Pb intensities. Calibration curves were used to determine Pb concentration in solution, and the ambient air volume sampled was used to calculate ambient PM-Pb mass concentrations. NIST Standard Reference Material (SRM) were digested and analyzed for PM-Pb content. Recoveries were 100±1 % (N=3) for SRM 1648a (urban particulate matter) and 102% (N=1) for SRM 2783 (urban PM2.5 collected on polycarbonate filter media). Lead isotope ratios were analyzed using a second ICP-MS analysis with a thallium internal standard. Thallium isotope ratios were first calibrated using a NIST Pb standard solution with quantified isotopic composition. The thallium ratios were then used as internal standards to correct Pb isotope ratios for the digested filter samples. The consistency of thallium internal standard isotopic ratios during ICP-MS analysis was evaluated by running a verification solution made from the NIST Pb isotopic standard after every 10 digested filter samples. Soil and Avgas Analysis Protocols Four soil samples were collected at each airport, one from each of the four sampling locations. Topsoil was collected into glass jars, transported from the field sites to the laboratory using ice packs and insulated shipping containers, and stored in a freezer. A portion of each sample has been analyzed total Pb and Pb isotopes. Soil samples were first sieved to remove small rocks and then ground into a coarse powder. The powdered soil was resuspended using a custom-made resuspension chamber generally based on the designs of Dobrzhinsky et al. (2012) and Martuzevicius et al. (2011). Resuspended soil was sampled onto Teflon filters using a MetOne SASS filter canister and PM2.5 cyclone. Deposited samples were analyzed for total Pb and Pb isotopes using the same methods as the ambient PM filter samples. A total of 15 avgas samples were collected from the three airports. Samples were collected in tin-plated steel cans, with caps sealed with Teflon tape to prevent volatile losses. A small portion of each sample was withdrawn and sent to Intertek Caleb Brett for total Pb analysis. Additional samples were sent to Washington University for Pb isotopes analysis using ICP-MS. The gasoline samples cannot be directed injected into the ICP-MS and thus Pb was extracted using the methodology presented by Lord (1994). A 3% m/v iodine in toluene solution was added to 1 mL of avgas. The lead was allowed B-2
to react with the iodine and then an ICP-MS standard bismuth solution was added as an internal standard. A 10% nitric acid solution was then added and thoroughly mixed. The solution was allowed to separate with the reacted Pb, partitioning from the organic phase to the aqueous phase. The aqueous phase was diluted to form a 5% nitric acid solution with expected Pb concentrations within the calibrated range of the ICP-MS. The diluted extracts were then analyzed by ICP-MS using rhodium as an internal standard to quantify total lead and thallium as an internal standard to quantify Pb isotope ratios. Thallium isotope ratios were first calibrated using a NIST Pb standard solution with quantified isotopic composition. The consistency of thallium internal standard isotopic ratios during ICP-MS analysis was evaluated by running a verification solution made from the NIST Pb isotopic standard after every 10 extracted avgas samples. PM-Pb Data Validation Field operator and laboratory analyst comments were entered into a database that tracks each sample. This information was used to flag or invalidate samples for reasons such as specks observed on the filter after sampling, sampler flow rate or duration out of range, or laboratory contamination. Table B-1 summarizes the samples that were invalidated. Table B-1 PM-Pb Samples That Were Invalidated Airport Date Site Sample Type Reason RVS 03/30/13 East PM2.5 collocate Sample duration < 9 hours (75%) RVS 04/06/13 North PM2.5 collocate Low Pb isotope ratios, poor agreement with collocated sample, large bias compared to XRF Pb RVS 04/08/13 South PM2.5 Filter fell in dirt during retrieval RVS 04/13/13 West PM2.5 Sample duration < 9 hours (75%) APA 05/15/13 Central Sec. PM2.5 Sample duration < 9 hours (75%) APA 06/07/13 Central PM2.5 collocate Sample duration 10 hours with poor collocated precision APA 06/07/13 East PM2.5 collocate Sample duration < 9 hours (75%) SMO 07/13/13 Central PM2.5 Contamination identified by Pb isotopes analysis (Figure B-1) SMO 07/13/13 East PM2.5 Contamination identified by Pb isotopes analysis (Figure B-1) SMO 07/17/13 South PM2.5 Contamination identified by Pb isotopes analysis (Figure B-1) B-3
Analysis of reagent, filter, and spiked blanks identified contamination in two batches of ICP-MS sample runs accounting for 55% of the SMO samples. Blank concentrations were high as 5 ng/m3 (reported as effective ambient Pb concentrations). The source of contamination was traced to the acid bath used to clean the glassware for sample digestion and extraction. Ten new blank samples were made by processing DAMMB through the glassware used for normal samples. The acid bath container was then cleaned and a new acid bath was made. All the glassware was cleaned with the new acid bath and three additional blank samples were prepared. The 13 blanks were then analyzed using ICP-MS. The median effective Pb concentrations before and after cleaning the acid bath were 2.8 ng/m3 and 0.2 ng/m3, respectively. Pb levels after cleaning the bath were equal to the MDL. Based on this analysis, 2.8 ng/m3 was subtracted from all Pb sample concentrations measured in these two SMO sample batches. Lead isotope ratios were then examined to determine if there were any samples with extreme levels of contamination. Figure B-1 shows the 208Pb/206Pb isotope ratio versus measured Pb concentration for the samples in the two contaminated SMO batches. Three samples, denoted by the solid black circles, were determined to have extreme levels of contamination as evidenced by the combination of high concentrations and low 208Pb/206Pb ratios. Even if these samples were not contaminated by the sample digestion process, they indicate high levels of Pb with isotopic composition that does not correspond to TEL-Pb. Figure B-1 208Pb/206Pb Ratio versus PM-Pb Concentration for Contaminated SMO Digestion Batches B-4
PM-Pb Field Blanks Eight PM2.5 and four TSP field blanks were collected at each airport (three field blanks per sampler per airport). Table B-2 presents the effective ambient field blank concentrations. One extreme value was observed (38 ng/m3) and cannot be explained. Summary statistics are presented in the main body of the report. Table B-2 Effective Ambient Pb Concentrations for the Field Blanks Collection Date Airport Inlet PM-Pb (ng/m3) 04/11/2013 RVS 2.5 0.21 04/11/2013 RVS 2.5 0.25 04/11/2013 RVS 2.5 38.66 04/11/2013 RVS 2.5 0.13 04/18/2013 RVS 2.5 0.09 04/18/2013 RVS TSP 0.42 04/18/2013 RVS 2.5 0.08 04/18/2013 RVS TSP 1.78 04/27/2013 RVS 2.5 0.13 04/27/2013 RVS TSP 0.17 04/27/2013 RVS 2.5 0.16 04/27/2013 RVS TSP 0.42 05/27/2013 APA 2.5 0.5 05/27/2013 APA TSP 0.3 05/27/2013 APA 2.5 -0.1 05/27/2013 APA TSP 0.3 05/29/2013 APA 2.5 -0.2 05/29/2013 APA 2.5 0.0 05/29/2013 APA 2.5 -0.1 05/29/2013 APA 2.5 0.4 06/05/2013 APA 2.5 0.3 06/05/2013 APA TSP -0.2 06/05/2013 APA 2.5 -0.2 06/05/2013 APA TSP 0.3 07/10/2013 SMO 2.5 0.2 07/10/2013 SMO TSP 0.9 07/10/2013 SMO 2.5 0.0 07/10/2013 SMO TSP 1.6 07/16/2013 SMO 2.5 -1.1 07/16/2013 SMO 2.5 0.0 07/16/2013 SMO 2.5 -0.8 07/16/2013 SMO 2.5 -0.3 07/19/2013 SMO 2.5 1.3 07/19/2013 SMO TSP -2.0 07/19/2013 SMO 2.5 0.3 07/19/2013 SMO TSP 2.0 B-5
Collocated Precision Given the importance of characterizing data quality at high PM-Pb concentrations, PM2.5- Pb precision was further examined. Collocated sample pairs were ranked by the mean concentration for each pair and the precision was repeatedly calculated adding one pair at a time, starting with the highest concentration and adding sample pairs with decreasing concentration. Figure B-2 shows the evolution of relative precision using this approach. Relative precisions when including only a few sample pairs can be noisy and the values for N ⤠3 pairs are excluded from the figure. With increasing sample pairs (i.e., moving from right to left on the plot), including a large concentration gap between 8 and 16 ng/m3, the relative precision is relatively constant. This asymptotic behavior is consistent with the relative precision being a constant value at high concentrations. Further increasing the number of sample pairs to include lower concentrations yielded a monotonic degradation in relative precision with a maximum value of 17% when including all data. In this region of lower concentrations, the additive contribution to precision is also important. The upper tertile mean concentration value is 16.1 ng/m3 and the maximum mean concentration value is 48.8 ng/m3. Precision was calculated 25 times by re-sampling with replication the ten sample pairs in this concentration range. The mean relative precision was 12% with 1Ï standard deviation of 3%. This result demonstrates that 12% is a stable estimate of the relative precision at high concentrations. Figure B-2 Ripening of the PM2.5-Pb Collocated Precision B-6
PM-Pb Analysis by XRF with Comparison to ICP-MS Twenty-two airborne PM2.5 samples from each airport were sent to Cooper Environmental Services (CES) for elemental analysis by XRF. Fourteen field blanks, including at least four from each airport, and six laboratory blanks were also analyzed, with the latter used to develop the spectral blank correction for the specific make and model of filters used in this study. Samples were analyzed by CES Protocol C, which is the most sensitive of the three routine protocols offered with a Pb MDL of 0.24 ng/m3 effective ambient concentration. Pb effective ambient concentrations for each of the 14 field blanks were less than 0.5 ng/m3. Figure B-3 compares ambient PM2.5-Pb measured by XRF and ICP-MS. Samples with ICP-MS PM2.5-Pb less than three times the ICP-MS MDL of 0.2 ng/m3 were excluded. The data are highly correlated, with r2 = 0.99 (N = 57). The regression intercept is statistically indistinguishable from zero, but from the regression slope the XRF data are biased 20% high compared to the ICP-MS data. The quantitative SRM recoveries provide compelling evidence for the accuracy of the ICP-MS data. Measurement differences in Pb are not unusual. For example, in the South Coast Air Quality Management District airport study (SCAQMD 2010), XRF and ICP-MS data were highly correlated (r2 = 0.97) with regression slope of 1.06 and intercept of 25.6 ng/m3. Compared to this study, the SCAQMD-reported slope is closer to unity, but the intercept is much larger. Figure B-3 PM2.5-Pb measured by XRF and ICP-MS Note: Regression coefficients including 95% confidence intervals are from a constant variance Deming regression. The solid line is the 1:1 line and the dashed line is the regression line. B-7
References Dobrzhinsky, N., E. Krugly, L. Kliucininkas, T. Prasauskas, M. Kireitseu, A. Zerrath, and D. Martuzevicius. 2012. âCharacterization of desert road dust aerosol from provinces of Afghanistan and Iraq.â Aerosol and Air Quality Research 12: 1209â1216. Lord, C.J. 1994. âDetermination of lead and lead isotope ratios in gasoline by inductively coupled plasma mass spectrometry.â Journal of Analytical Atomic Spectrometry 9: 599â603. Martuzevicius, D., L. Kliucininkas, T. Prasauskas, E. Krugly, V. Kauneliene, and B. Strandberg. 2011. âResuspension of particulate matter and PAHs from street dust.â Atmospheric Environment 45: 310â31. South Coast Air Quality Management District (SCAQMD). 2010. General Aviation Airport Monitoring Study, Final Report. Prepared for the U.S. Environmental Protection Agency. B-8