Introduction

    The CEMRC aerosol sampling program for the WIPP EM is designed to study the pathway that is unarguably the most likely route by which contaminants from the WIPP site could become rapidly dispersed in the environment. A major objective for these studies of aerosols is to investigate relationships among trace metals, ions, and radionuclides in aerosol samples collected from the vicinity of the WIPP. Detailed documentation of these parameters of pre-operational conditions is necessary to accurately assess any suspected changes occurring after the disposal operations commence. Data for trace metals and aerosol ion concentrations also contribute to an understanding of the natural variability and the sources and sinks for various types of pollutants, which enhances identification of causes for any observed changes in radionuclide activities.

Methods

    A summary of the sampling design for the aerosol studies is presented elsewhere in this report (Aerosol section). Aerosol samples are currently collected from four sites, but only data from three sampling stations (On Site, Cactus Flats and Near Field) are reported in this summary. Low-volume samplers (10 L min^-1) were used to collect aerosol TSP, PM₁₀ and PM_2.5 for non-radiological analyses reported here.

    The aerosol samples were analyzed by CEMRC for suites of major and trace elements by atomic absorption (AA) spectrometry and inductively-coupled plasma atomic emission spectrometry (ICP-ES), and the concentrations of selected anions and cations were determined by ion chromatography (IC). A subset of the aerosol samples was also analyzed by an X-ray fluorescence (XRF) technique through a subcontract with the Desert Research Institute (DRI). The XRF studies are a continuation of work begun in December 1996; the first set of XRF data was presented in the CEMRC 1997 Report.

    Aerosol filters from low-volume samplers were prepared for the elemental (AA and ICP-ES) analyses using a CEM 2100 Microwave digestion system. A combination of microwave energy and strong acids is needed to destroy the mixed-cellulose ester filter matrix (Gelman MetricelÒ ) and to solubilize the more refractory materials collected on the filter. The aerosol-laden filters were processed in sealed microwave vessels using a combination of HNO₃, HCl, HF, and H₂O₂.

    For the IC analyses, individual Gelman Teflo® PTFE Teflon® filters were extracted in clean polyethylene bags after first wetting the filters with isopropyl alcohol. The filters were extracted in three steps with de-ionized water in sealed polyethylene bags, and the extractions were done in an ultrasonic water bath to facilitate the process. The same aqueous extracts of the aerosol samples were used for both anion and cation analyses. In some cases the aqueous extracts not consumed in the IC analyses also were analyzed by ICP-ES to provide intra-laboratory comparisons between instruments.

    Standard operating procedures have been developed for the CEMRC analyses, and where possible these are based on applicable standard U.S. Environmental Protection Agency (EPA) procedures. A summary of the analytical procedures used for the AA, ICP-ES and IC analysesof WIPP EM samples is presented elsewhere in this report (Appendix K).

    Standard operating procedures have been developed for the CEMRC analyses, and where possible these are based on applicable standard U.S. Environmental Protection Agency (EPA) procedures. A summary of the analytical procedures used for the AA, ICP-ES and IC analyses of WIPP EM samples is presented elsewhere in this report (Appendix K).

    Gravimetric determinations of aerosol mass were completed at CEMRC. Prior to placement in the aerosol samplers, all filters were preconditioned in a dessicator, equilibrated to ambient conditions, and then weighed using a microbalance
(1 m g resolution). At the completion of each sampling period, filters were removed from the samplers and placed in petri dishes for transport and storage. Loaded filters were re-conditioned, re-equilibrated, and re-weighed to determine total mass accumulation. The total air volume for each sampling period was calculated based on an integrated total during each sampling interval. The mass accumulation divided by the total air volume drawn through the sampler was used to calculate the aerosol mass concentrations.

    Elemental analyses by XRF were performed by DRI in Reno, Nevada. A Kevex Corporation Model 0700/8000 and a Kevex 0700/IXRF energy dispersive X-ray fluorescence analyzer were used for the analyses. Details of the XRF analytical procedure and the quality assurance/quality controls (QA/QC) used for the analyses were presented in the CEMRC 1997 annual report. Briefly, two protocols were used for the analyses of the aerosol-laden filters: Protocol C for the TSP and PM₁₀ samples and Protocol D for the PM_2.5 samples. These protocols differ in data acquisition times, with longer counting times used for Protocol D. The more sensitive analytical scheme was used for the PM_2.5 samples because the mass of material collected in that size fraction was expected to be lower than in either the TSP or PM₁₀.

    The XRF analyses can generate data for the following elements, providing their concentrations are sufficiently high: Al, Si, P, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ga, As, Se, Br, Rb, Sr, Y, Zr, Mo, Pd, Ag, Cd, In, Sn, Sb, Ba, La, Au, Hg, Tl, Pb, and U. However, not all elements were detected in the WIPP EM aerosol samples. The following discussion is based on the XRF analysis of randomly selected sets of samples (two per week), most of which were collected in 1997. Each set consisted of a TSP, PM₁₀ and PM_2.5 low-volume sample collected over a nominal 24-hr period. All of the samples analyzed by XRF were from the Near-Field station. If one or more samples from a set were missing, the others were analyzed anyway.

Results and Discussion

    A time-series plot of Al (mineral aerosol) concentrations in the TSP, PM₁₀ and PM_2.5 samples is presented in Fig. 11 (note that the Y-axis for this figure is a logarithmic scale). The mineral aerosol concentrations of Al did not vary strongly with season. However, several moderate dust events were evident, including the two most prominent in early mid-March and July of 1997. The concentrations of Al in the TSP and PM₁₀ fractions exhibited a similar pattern of variation, which is not unexpected because most of the dust mass normally is associated with particles less than 10 mm, and therefore the dust concentrations in the PM₁₀ samples should be similar to those in the TSP fraction. It is also interesting to note that although the Al concentrations in the PM_2.5 fraction were only about 10% of those in TSP or PM₁₀, the small particle fraction tracked the PM₁₀ and TSP samples rather well, suggesting that the proportion of 2.5 mm dust particles relative to the total mineral dust concentrations stayed relatively constant throughout the year.

    Mineral aerosol made up a significant fraction of the TSP mass, but the mass of particulate material in the atmosphere clearly cannot be accounted for by windblown dust alone. If the TSP mass were solely due to mineral dust, all of the points in Fig. 12 would have fallen close to the "crustal ratio" line, which is based on an average Al concentration of ~8% in crustal weathering products (Taylor, S. R., and S. M. McLennan, 1995, Reviews of Geophysics, 33, 241). Some points were at or above the crustal line, indicating that under some circumstances the bulk of the TSP could be accounted for by dust concentrations, but this only seemed to occur when the TSP mass concentrations were > 10 mg m^-3. To evaluate the dust contribution to the total mass concentrations, one can calculate the percentage of mass accounted for by dust:
 
 

    For the TSP samples (N = 60), the mean percentage (+ SD) of mass attributable to dust was 46% (+ 33%). The corresponding value for PM₁₀ (N = 67) was essentially the same (47 + 36%). In contrast, the percentage of aerosol mass attributable to mineral dust for the PM_2.5 fraction (N = 64) was only 9% (+ 12%), clearly indicating the dominance of non-crustal materials in the small particle fraction.

    Compared with Al (see Fig. 11) the atmospheric S mass concentrations followed a similar pattern of variation in all three particle fractions (TSP, PM₁₀ and PM_2.5) (Fig. 13). Presumably, this occurred because most of the S occurred as sulfate aerosol which is mainly formed via gas-to-particle conversion (see below). This heterogeneous process favors the formation of submicrometer (often called fine) aerosols, and therefore the bulk of the sulfate mass was carried by the PM_2.5 fraction. Zn and Pb were also substantially enriched with respect to a crustal source and exhibited patterns more similar to S than to Al, in that their mass tended to be concentrated in the PM_2.5 fraction.

    A comparison of S concentrations as a function of mass (Fig. 14) showed that the calculated sulfate concentrations (derived from the XRF sulfur data) at times accounted for ~40% to 45% of the total mass in the PM_2.5 fraction. This conclusion is based on the assumption that sulfate was the dominant form of S in the aerosol; evidence supporting this assumption is presented below. Further examination of this relationship for the XRF data indicated that sulfate more commonly accounted for ~20% of the PM_2.5 mass.

    The IC data were used to directly calculate the percentages of mass accounted for by sulfate at Near Field and at Cactus Flats. These calculations showed that the mean percentages of total mass (+SE) accounted for by sulfate in the PM_2.5 fraction were comparable to estimates from the XRF data: 18% (+ 1.9%) at Near Field (N = 19) and 19% (+ 2.6%) at Cactus Flats (N = 14) (Table 4). The percentage of mass due to sulfate was higher in the PM_2.5 fraction than in either PM₁₀ or TSP, which were similar, at about 10% at both sites. The concentrations of nitrate in the samples were much smaller than sulfate, amounting to only a few percent in any of the size fractions. These results indicate that major ions such as sulfate and nitrate are important, but at least in terms of mass, not dominant components of the background fine aerosol in the area of the WIPP.

    The results of this comparison are shown in a plot of the XRF S data compared to the IC sulfate data (Fig. 15). The diagonal line shown in this figure is the predicted relationship if all of the S were in the form of sulfate (3:1 ratio, MWT(molecular weight) sulfate/MWT of S). For the TSP samples, all of the points fall on the diagonal line or very close to it, supporting the hypothesis that sulfate is the dominant form of S in the aerosol. Clearly some of the PM₁₀ and PM_2.5 points fall to the right of (or below) the line as concentrations increase; this could most easily be explained by the existence of some S species other than sulfate. Another possibility is that one set of measurements is biased. However, even if this were true, the non-sulfate fraction would be small in terms of mass (typically 10% to 15% of the S mass concentration).

    A comparison of sulfate and nitrate concentrations (Fig. 16) showed little difference between the Near Field and Cactus Flats stations. However, there appeared to be a difference among size fractions, with a higher sulfate to nitrate ratio in the PM_2.5 samples compared to the TSP and PM₁₀ samples. A possible mechanism for this is sorption of gas-phase nitric acid onto pre-existing aerosols (i.e. condensation of nitric acid onto dust and other large particles).

    The comparison of sulfate and nitrate concentrations also showed that some of the aerosol samples had unusually high sulfate/nitrate ratios (> 50). These anomalous ratios occurred when sulfate loadings were higher than average, and the nitrate concentrations were lower than average. Furthermore, the samples with high sulfate/nitrate were not restricted to one site, nor to a particular size fraction (Fig. 17). It is possible that some artifacts were introduced during sampling or sample preparation, but a more interesting possibility is that the anomalous sulfate/nitrate ratios reflected changing environmental conditions. The most intriguing scenario is that these anomalous chemical signals were related to the very large volumes of smoke transported from Mexico to the southwestern United States in the spring of 1998.

    It is interesting to note that some unusually high sulfate/nitrate ratios were observed at or around the same time as the high mass concentrations. At Near Field, the nitrate concentration in the 23 March PM_2.5 sample was below detection, but if the detection limit for nitrate for a nominal 24-hour sample (0.0088 mg m^-3) were substituted, the sulfate to nitrate ratio would be 81, a value much higher than the typical ratios of 5 to 10. Furthermore, the sulfate/nitrate ratios for the prior two PM_2.5 samples (collected starting 9 and 18 March) were 31 and 84, respectively, indicating that the 23 March sample was not simply an outlier. This is further substantiated by the data from Cactus Flats where a high sulfate/nitrate ratio (66) was observed in the 13 March PM_2.5 sample. However, the sulfate/nitrate ratios during this period are not uniformly high; the 18 March PM_2.5 sample from Cactus Flats had a more typical value (6.1).

    As mentioned above, the high ratios were more the result of lower nitrate rather than higher sulfate concentrations, suggesting the re-volatilization of nitrate from the aerosol. Tsai and Peng (Tsai, C-J., and S-N. Peng, 1998, Atmospheric Environment, 32, 1605) suggest that sampling losses of volatile species, including nitrate, can occur as a result of gas-particle and particle-particle interactions during sampling, and these may apply in this case. Although outside the scope of the WIPP EM, these data raise an interesting question concerning what chemical reactions were responsible for the loss of nitrate during collection of some samples but not others.

    Another step in investigating possible effects from the fires was to examine the data for fine-particle K, a commonly used chemical tracer for biomass burning (Andreae, M. O., 1983, Science, 220, 1148). One would expect strong peaks in the PM_2.5 K concentrations during the smoke events, but only one sample, collected on 18 May 1998, appeared to have an unusually high fine particle K concentrations (one collected on 4 May and the other on 18 May 1998). High K concentrations were not observed on 23 March, when the high mass concentrations and the high sulfate/nitrate ratios were observed (Fig. 19). However, the sample with the high concentration of fine particle K was collected when the effects of the smoke from the Mexican fires were readily visible locally, confirming that the long-range transport of smoke aerosol can at times be detected via chemical changes in the Carlsbad aerosol.

    Although many residents of the Carlsbad area probably were aware of the smoke, the impacts of the fires were more subtle than one might expect. For example, a plot of solar radiation at Carlsbad (Fig. 20) does not show any extended perturbations caused by the smoke. Even though some reductions of the solar flux were evident, these cannot be directly linked to the fires from the data currently available. As noted above, the mass concentrations did not appear to be significantly affected by the fires, but some chemical alterations of the aerosol may have occurred as a result of the fires.

    This was the first year in which a relatively complete set of inorganic data were obtained for the WIPP EM, and more information on the effects of Mexican fires on the Carlsbad aerosol will be obtained in future monitoring studies if the situation arises again. In addition, studies by other groups have shown that the smoke aerosols have elevated levels of several carbonyl compounds (R. Dixon, New Mexico Institute of Mining and Technology, personal communication). Methods are being developed by CEMRC for the determination of acetate and formate, two substances that also may be affected by fires.

    To compare results between the XRF and AA/ICP-ES analyses, we calculated relative percent difference for the mean concentrations of those elements determined by both laboratories. The relative percent difference is calculated as

where RPD is the relative percent difference, C₁ is the concentration determined by XRF analysis, C₂ is the concentration determined by AA/ICP-ES analysis.

    Results of this comparison between analytical techniques are presented in Table 7. In general the results all agreed to within a factor of 3 (this corresponds to a relative percent difference of 100%). This comparison showed the largest differences between techniques for Zn and Ni, but these large differences were driven by a few samples with very high concentrations of these elements. For example, the Zn concentration measured by ICP-ES in the TSP sample collected at Near Field on 6 October 1998 was 0.37 mg m^-3. Another sample collected on 22 June 1998 and analyzed by XRF had a Zn concentration of 0.0056 mg m^-3. One might speculate that the high Zn concentration resulted from contamination during sampling or sample preparation, but this would only be speculation, and there is no a priori reason to discard this sample as problematic. Follow-up examinations of the in-house data with respect to the XRF results will continue as more samples are analyzed. At the same time, both inter- and intra-laboratory comparisons will be conducted to further validate the analytical methods.