ARTICLE pubs.acs.org/est
Evaluation of Passive Samplers for Assessment of Community Exposure to Toxic Air Contaminants and Related Pollutants J. Brooks Mason, Eric M. Fujita,* David E. Campbell, and Barbara Zielinska* Desert Research Institute Division of Atmospheric Sciences 2215 Raggio Parkway, Reno, Nevada 89512, United States ABSTRACT: The precision, accuracy, and sampling rates of Radiello and Ogawa passive samplers were evaluated in the laboratory using a flow-through chamber and under field conditions prior to their use in the 2007 Harbor Community Monitoring Study (HCMS), a saturation monitoring campaign in the communities adjacent to the Ports of Los Angeles and Long Beach. Passive methods included Radiello samplers for volatile organic compounds (benzene, toluene, ethylbenzene, xylenes, 1,3-butadiene), aldehydes (formaldehyde, acetaldehyde, acrolein) and hydrogen sulfide, and Ogawa samplers for nitrogen oxides and sulfur dioxide. Additional experiments were conducted to study the robustness of the passive sampling methods under variable ambient wind speed, sampling duration, and storage time before analysis. Our experimentally determined sampling rates were in agreement with the rates published by Radiello and Ogawa with the following exceptions: we observed a diffusion rate of 22.4 ( 0.1 mL/min for benzene and 37.4 ( 1.5 mL/min for ethylbenzene compared to the Radiello published values of 27.8 and 25.7 mL/min, respectively. With few exceptions, the passive monitoring methods measured one-week average ambient concentrations of selected pollutants with sensitivity and precision comparable to conventional monitoring methods averaged over the same period. Radiello Carbograph 4 VOC sampler is not suitable for the collection of 1,3-butadiene due to backdiffusion. Results for the Radiello aldehyde sampler were inconclusive due to lack of reliable reference methods for all carbonyl compounds of interest.
1. INTRODUCTION The Harbor Community Monitoring Study (HCMS) was conducted to characterize the spatial variations in concentrations of toxic air contaminants (TACs) and their copollutants within the the community of Wilmington and parts of Carson, West Long Beach, and San Pedro in California’s South Coast Air Basin (SoCAB). These communities were selected due to the close proximity of residents to the high density of emission sources in the area. The sources include the Ports of Los Angeles and Long Beach, petroleum refineries, intermodal rail facilities, and diesel trucks (high traffic volumes associated with the movement of goods from one of the busiest port complexes in the world). The HCMS consisted of three types of air pollution sampling: a high density (saturation) air monitoring network of 23 sampling locations operated by the Desert Research Institute (DRI), mobile sampling by the University of California, Los Angeles and California Air Resources Board (CARB), and a network of particle counters operated by the University of Southern California. HCMS was conducted during 2007 concurrently with ongoing monitoring programs in the study area by the South Coast Air Quality Management District (SCAQMD) and the Ports of Los Angeles and Long Beach. The saturation monitoring by DRI was designed to establish spatial variations in pollutant concentrations at the community scale. The network consisted of seven day time-integrated sampling at 20 sites for 4 consecutive weeks in four seasons during 2007. Measurements included NOX (NO þ NO2) and SO2 using Ogawa passive samplers, as well as volatile organic compounds (VOC) (benzene, toluene, ethylbenzene, xylene) and carbonyl compounds (formaldehyde, acetaldehyde, acrolein) using Radiello passive samplers. Additionally, 7 day integrated Teflon and quartz filters were collected with portable Airmetrics MiniVol samplers and analyzed for PM2.5 mass and r 2011 American Chemical Society
organic and elemental carbon. NO2 (Ogawa) and H2S (Radiello) were also measured at three near-roadway sites, and full sets of passive measurements were made in triplicate at a quality assurance site in West Long Beach. The working principle of passive sampling is diffusion of gaseous pollutants across a surface to an adsorbing material onto which the pollutant of interest accumulates over time. The continual adsorption of the pollutant from the air maintains a concentration gradient near the surface that allows uptake of the pollutant to occur without any forced air movement (no pump or fan is required). Electricity demands, moving parts, and size make active and continuous monitoring technologies a challenge for personal exposure assessment, whereas passive monitors are unobtrusive and costs for sample collection are low. The ability of passive samplers to collect analytes over extended periods of time allows for the measurement of trace pollutants. Sensitivity is limited only by the amount of time for which a sampler can be exposed and the blank value of the analyte on an unexposed adsorbent surface. After sampling, the collected pollutant is desorbed from the sampling media by thermal or chemical means and analyzed quantitatively. The average concentration of the pollutant in the atmosphere to which the sampler was exposed is calculated by dividing the mass of pollutant, measured analytically, by the product of the diffusion rate and sampling time.1 The Ogawa NO2 and NOX sampler have been validated using chemiluminescence measurements in ambient environments.2,3 The Radiello sampler, developed over a decade ago for assessment of benzene exposure,4 has been evaluated for collection of VOC Received: August 12, 2010 Accepted: January 13, 2011 Revised: December 13, 2010 Published: February 15, 2011 2243
dx.doi.org/10.1021/es102500v | Environ. Sci. Technol. 2011, 45, 2243–2249
Environmental Science & Technology using Carbograph 4 and other adsorbents.5-8 PennequinCardinal9 reports BTEX sampling rates in line with those published by Radiello.10 The Radiello aldehyde and H2S samplers have been utilized in ambient measurement studies but little work has been done to independently evaluate the diffusion rates.11 Radiello10 and Ogawa and Company12 publish experimentally determined diffusion rates for a number of common pollutants. However, diffusivity can vary with environmental factors including temperature, relative humidity, wind speed, concentration of analyte, and the concentration of chemically or physically interfering species. Recent interest in 1,3-butadiene has spurred the development of a passive method for this suspected carcinogen.6,7,13 Measurement of 1,3-butadiene with adsorbents is difficult due to its volatility. Strandberg6 reports that 1,3-butadiene is stable on Carbopack X and Carbograph 5 for short sampling times (one day), but is susceptible to backdiffusion for increased periods (one week).
2. EXPERIMENTAL METHODS The diffusion rates of the Ogawa and Radiello passive samplers were evaluated in this study under controlled laboratory conditions, in the field during an initial pilot study and during the main HCMS. Measurement accuracy was assessed by comparison with reference methods, which included EPA-certified continuous gas monitors and time-integrated samples collected by active sampling methods. Precisions of the passive sampling methods were estimated from 2σ deviations of the absolute differences of the individual sample values to the mean of the triplicate samples (four sets of triplicates per season). 2.1. Laboratory Evaluation. Passive samplers were exposed in a 100 L flow-through chamber with known concentrations of analytes. The chamber is a 100 L half-cylindrical shape framed with Teflon rods and sheeting. An internal fan ensured a wellmixed atmosphere and emulated a wind speed of 1.0 m/s. Stainless steel ports are built into the base for test atmosphere inflow, exhaust, temperature and relative humidity (RH) probes, and sampling. The test atmospheres were created by diluting certified gas standards with zero air using an Environics 9100 Ambient Monitoring Calibration System (Toland, CT). Zero air was generated using an Aadco 737 (Cleves, OH) pure air generator outfitted with scrubbing filters. Zero air purity is regularly evaluated by the DRI Organic Analytical Laboratory (OAL) for use in other applications. The diluted test atmosphere was then split: half of the flow was humidified until saturation and half was diverted around the humidifier and mixed with the saturated air to form a 50% RH mixture. The humidified test atmosphere was fed directly to the chamber at 2.5 L per minute. Chamber tests were used to evaluate the published diffusion rates of Ogawa NOX and NO2 samplers and Radiello VOC, aldehyde and H2S samplers. The samplers were deployed in triplicate for one-week periods. A chemiluminescence monitor was used to measure NOX concentrations. VOC and aldehyde concentrations were measured with seven 24 h time-integrated canisters and DNPH cartridges, respectively, according to the EPA Method TO15 and TO11A,14,15 respectively. Nominal concentrations determined by the Environics 9100 were used for evaluating the diffusion rate of H2S. Additional experiments were used to evaluate the collection efficiency of the Radiello VOC sampler for 1,3-butadiene and to evaluate the published sampling rates for BTEX. If suitable for 1,3-butadiene, experimental calculation of a sampling rate is
ARTICLE
possible by measuring the chamber test atmosphere with an established method. Previous studies have shown that 1,3butadiene is susceptible to backdiffusion on passive adsorbents.6,7 In order to check the stability of 1,3-butadiene and BTEX on the cartridge during and after exposure, samplers were deployed in triplicate in six groups. Triplicate groups were removed from the chamber and analyzed immediately after 1, 4, and 7 day exposures. The other three groups were exposed for the full seven days, transferred to a freezer at -18 °C for storage, and analyzed after 1, 7, and 14 days to evaluate the effect of storage time before analysis. 2.2. Pilot Study. A pilot study was conducted during a week in August 2006 at the SCAQMD North Long Beach monitoring station to determine the replicate precision of the passive samplers for NO2, NOX, SO2, H2S, BTEX, formaldehyde, acetaldehyde, and acrolein under field conditions. The passive measurements were compared with the SCAQMD continuous NOX, NO2, and SO2 data, and with 24 hr time-integrated canisters and carbonyl samples, as discussed in Section 3.1. Sampler inlets and passive samplers were located on the station’s rooftop instrument platform. Passive samplers were exposed for a oneweek period. They were deployed at a height of approximately 2 m above the instrument platform on the roof of the station, along a line running parallel to Long Beach Blvd. approximately 10 m from the street side roofline of the building. Another objective of the pilot study was to understand the effect that stagnant air might have on diffusion rates since the Long Beach/Wilmington area is characterized by low nocturnal winds. The sampling rates are invariant from 0.1 to 10 m/s, according to specification published by Radiello.10 In order to determine the potential influence of airflow on diffusion rates, the samplers were deployed in two groups with an oscillating electric fan providing constant easterly airflow at approximately 3 mph (1.3 m/s) across one group. 2.3. Harbor Community Monitoring Study Quality Assurance. Quality assurance during the HCMS was conducted at a site in West Long Beach for two weeks in the summer and winter seasons. During these periods, Ogawa passive samplers for NO2, NOX, and SO2 and Radiello passive samplers for VOC and aldehydes were deployed in triplicate to determine replicate precision. SO2 and NOX were measured with colocated continuous analyzers. Other passive diffusion rates were tested by comparison with seven consecutive 24 h active canister and DNPH cartridge samples as mentioned in Section 3.1. These comparisons were not valid for the summer season due to loss of one or more 24 h active samples for several sets of seven consecutive sampling days. 2.4. Equipment and Analysis. Ogawa passive samplers were used for monitoring NOX, NO2, and SO2. NOX and SO2 were collected over weeklong periods using precoated 14.5 mm sampling pads, deployed in personal sampling bodies. NO concentrations were calculated by subtracting NO2 from NOX concentrations. Sampling and analysis were performed according to manufacturer protocols.12 The Ogawa NO2 and NOX pads were extracted and mixed with a solution of sulfanilamide and N-(1-naphthyl)-ethylenediamine dihydrochoride to produce a colored nitrite solution which was analyzed on a Technicon (Tarrytown, NY) TRAACS 800 Automated Colorimetric System. The Ogawa SO2 pads were extracted in 8 mL of deionizeddistilled water and combined with 1.75% hydrogen peroxide. The solution was measured for sulfate by a Dionex 2020i (Sunnyvale, CA) ion chromatograph (IC). Certified commercial standards 2244
dx.doi.org/10.1021/es102500v |Environ. Sci. Technol. 2011, 45, 2243–2249
Environmental Science & Technology (purchased from Dionex and ERA, Arvada, CO) were used for the calibration of the IC and colorimeter. Benzene, toluene, ethylbenzene, and xylenes (BTEX) were passively collected over weeklong periods using Radiello VOC samplers consisting of stainless steel mesh cylinders (3 8 um mesh, 4.8 mm diameter 60 mm length) packed with Carbograph 4 (350 mg). The cartridges were deployed in the diffusive sampling bodies according to manufacturer’s instructions.10 Collection of 1,3-butadiene was also evaluated. Radiello VOC cartridges were analyzed on a Varian 3800 gas chromatograph with Saturn 2000 mass spectrometry (MS) detection equipped with a Gerstel TDSA-3 thermal desorption unit. Initial desorption was set for five minutes at 300 °C before transfer to a Tenax TA trap cooled to -150 °C. Sample was split 15:1 in order to reduce analytical loading in the MS. After preconcentration on the trap, the sample was injected at 240 °C onto a 60 m, widebore, Phenometrix ZB-1 for separation before MS detection. The instrument was calibrated by thermally desorbing glass tubes packed with Carbograph 4. The calibration tubes were loaded with a certified standard (1,3-butadiene and BTEX in nitrogen, Scott Specialty Gases, 5% accuracy) by measuring the volume passing through the tubes. Down stream tubes were analyzed on a weekly basis to rule out breakthrough. Radiello diffusive samplers were used to passively collect carbonyl compounds. Stainless steel net cartridges filled with 2,4-dinitrophenylhydrazine (2,4-DNPH) coated florisil were used. Carbonyl compounds react with 2,4-DNPH forming corresponding dinitrophenylhydrazones. Twenty-four hour time averaged Waters Sep-Pac DNPH cartridges and Radiello aldehyde cartridges were eluted with 2 mL of acetonitrile (ACN) and filtered before analysis. The samples were then separated and analyzed on a Waters 2695 equipped with a Waters 996 photodiode array detector. The mobile phase was water and acetonitrile run on a Varian Polaris 3μ C18-A 150 4.6 mm column according to EPA method TO-11A.14 The HPLC was calibrated using a certified liquid standard (carbonyl compounds as DNPH derivatives, ACCU Standards, Inc.). Radiello chemiadsorbing cartridges were used for passive sampling of H2S. The cartridge is made of microporous polyethylene and impregnated with zinc acetate. H2S is chemiadsorbed by zinc acetate and transformed into stable zinc sulfide. Radiello H2S samples were eluted with a 10.5 mL ferric chlorideamine solution to yield methylene blue, which was analyzed with a Bausch and Laumb Spectronic 20 visible spectrometer at 665 nm. Calibration was completed using a calibration kit from Radiello: RAD-171.10 Passive VOC samples were compared to corresponding canister samples analyzed according to the EPA Method TO1515 using a Varian 3800 gas chromatograph interfaced to a Varian Saturn 2000 ion trap mass spectrometer (MS) and flame ionization detector (FID). Canisters were preconcentrated using a Lotus Ultra Trace Toxics System-MS-TO15 before injection onto a Varian CP fused silica 60 m widebore column with MS detection for BTEX and an Agilent Alumina 30 m megabore column with FID detection for 1,3-butadiene . Calibration of the system was done using a 74 component VOC mixture containing BTEX and 1,3-butadiene (Apel-Riemer Environmental, Inc., Denver, CO). Passive carbonyl samples were compared to corresponding samples collected with Sep-Pak cartridges that have been impregnated with an acidified 2,4-dinitrophenylhydrazine (DNPH) reagent (Waters, Inc.), according to the EPA Method TO-11A.14 The cartridges were analyzed with a Waters
ARTICLE
2690 high performance liquid chromatograph (HPLC) equipped with a photodiode array detector for separation and quantification of the hydrazones. Acrolein is known to rearrange on DNPH cartridges to an unknown degradation product: acrolein-x. The process of rearrangement is sufficiently rapid that the majority of acrolein may convert to acrolein-x, unless the sample is analyzed within a few hours. The problem is compounded by the fact that acrolein-x coelutes in our HPLC spectra with another common carbonyl compound, butyraldehyde. The spectra from the photodiode array detector show that there is substantial overlap in the chromatographic retention time of acrolein-x with butyraldehyde. Thus, the sum of acrolein and butyraldehyde represents an upper-bound estimate of acrolein that was originally present in the sample. In order to circumvent this problem, the Organic Analytical Laboratory (OAL) at DRI recently performed experiments, as part of a separate study, to determine if a more accurate measurement of acrolein concentration could be obtained by postanalysis reprocessing of the HPLC spectra.16 This procedure was used to estimate total acrolein for analysis of Radiello passive samples and DNPH cartridges. Laboratory and field blanks (10% of samples) were analyzed as described above. Average field blank values (in general, below 10% of ambient measured values) were used for blank corrections.
3. RESULTS 3.1. Laboratory Evaluation. The chamber concentrations measured by Ogawa and Radiello passive samplers during 7 day exposures were in agreement with reference values (Table 1). The Ogawa NOX and NO2 and Radiello aldehyde and H2S samplers had measurement accuracy above 95% and replicate one σ precision within 2% for target compounds. BTEX concentrations measured by the Radiello VOC sampler were within 18% of the canister measurements with the exception of ethylbenzene. Replicate precision was within 8% of the mean in all cases. Additional experiments for the Radiello VOC sampler demonstrate increasing precision with exposure time (Figure 1). BTEX chamber concentrations were measured in the 0.4-3 ppbv range for all compounds. Benzene, toluene, and ethylbenzene show increasing precision and a decreasing diffusion rate as exposure time increases from 1 to 7 days, evidence that slight backdiffusion takes place during sampling and that sampling rates may need to be adjusted depending on exposure time. Furthermore, the ethylbenzene sampling rate is higher than the Radiello published rate in all replicate groups, however, the diffusion rate is stable among 7 day exposure times. Xylene precision and measurement accuracy were poor for 1 day exposures, but were excellent for longer sampling and storage durations. The low one-day values are likely due to lower xylene concentrations in the standard mixture and thus lower mass of xylenes collected on the passive samplers compared to other BTEX species. Storage tests demonstrated good reproducibility for samples stored up to 14 days at -18 °C. 1,3-butadiene chamber concentration during the 7 day period was measured by canister sampling to be 1.8 parts per billion. The mean passive sampling rate for 1, 4, and 7 day exposures, calculated from this concentration, was 4.9, 1.3, and 0.7 mL/min, respectively. The sampling rate reveals marked backdiffusion with sampling rates exponentially declining by 73% and 86% from the 1 day value for 4 and 7 day exposures. The calculated 2245
dx.doi.org/10.1021/es102500v |Environ. Sci. Technol. 2011, 45, 2243–2249
Environmental Science & Technology
ARTICLE
Table 1. Concentrations (ppbv) Measured by Passive and Reference Methods during Chamber Evaluations compounds
n
passive Sample (ppbv)a
passive RSD (%)
reference value (ppbv)b
passive-ref %Δc
NOX
3
39.8 ( 0.6
1.6%
39.00
2%
NO2
3
21.5 ( 0.3
1.4%
21.80
-1% -2%
formaldehyde
3
5.08 ( 0.36
2.0%
5.20
H2S
3
1.99 ( 0.04
2.0%
2.10
-5%
benzene
3
2.10 ( 0.24
4.9%
2.57
-18%or(l%)d
toluene
3
2.24 ( 0.11
6.7%
2.37
-5%
ethylbenzene
3
1.80 ( 0.12
4.5%
1.28
41% or (-6%) d
m,p-xylene o-xylene
3 3
0.89 ( 0.04 0.38 ( 0.02
5.3% 7.1%
1.02 0.43
-13% -12%
a Mean value ( standard deviation. b Reference method is by Horriba NO/NOX analyzer for NO, NO2, by 24 h time-integrated canisters for BTEX, and by dilution of standards by Environics 9100 for formaldehyde and H2S. c Percent difference of the passive result compared to the reference result. d Using our experimentally determined sampling rates of 22.4 and 37.4 mL/min (in parentheses) rather than 27.8 and 25.7 mL/min published by Radiello for benzene and ethylbenzene, respectively.
Figure 1. Experimentally determined sampling rates for BTEX and 1,3-butadiene using Radiello samplers.
sampling rate was consistent for samples stored for up to 14 days at -18 °C, indicating 1,3-butadiene stability during storage at this temperature. 3.2. Pilot Study. Concentrations of air toxics were low at the Long Beach AQMD station during the week of the pilot study measurements (Table 2). Wind data showed a strong, consistent diurnal pattern dominated by westerly winds mid-day and stagnant air at night. The following section describes the result of comparisons between active time-integrated or continuous sampling methods and corresponding passive measurements, as well as evaluations of measurement precision. The 1-σ replicate precisions for benzene, toluene, ethylbenzene, xylenes, formaldehyde, acetaldehyde, NOX, NO2, and SO2 were evaluated to be less than 16% of the mean in all cases (Table 2). It should be noted that the percent standard deviations for acrolein and H2S (37 and 40%, respectively) were measured under ambient concentrations that were below or near the published10 limits of detection (Table 3). A discernible difference was not observed between the constant airflow samplers and those exposed to ambient winds. An anemometer positioned directly above the ambient passive samplers confirmed that winds were undetectable for a significant period of time every night during exposure.
Nevertheless, there was no correlation between wind speed and sampling rate observed. The comparison of the passive samplers to the active reference methods produced variable results (Table 2). The BTEX compounds, with the exception of ethylbenzene, were within 20% of the canister measured concentrations. For ethylbenzene, a 33% higher sampling rate than that reported by Radiello was observed. However, when adjusted with our experimentally determined sampling rate, only 8% difference was observed. Xylenes were measured passively to be within 2% of the reference method. The Radiello aldehyde sampler performed well for formaldehyde (12%) but acetaldehyde concentration was underestimated by 43% in comparison with 24 h integrated active sampling. It should be noted that the presence of ozone is a concern when sampling with DNPH due to aldehyde scavenging and, as such, active DNPH media is typically ozone denuded. Since passive ozone denudation is infeasible due to the diffusive collection mechanism, active DNPH cartridges were not denuded to compare the effect of ozonolysis on passive versus active media. The mean ozone concentration during the pilot study was calculated from hourly average data as 26.9 parts per billion. Radiello10 has published data that suggests acetaldehyde is 2246
dx.doi.org/10.1021/es102500v |Environ. Sci. Technol. 2011, 45, 2243–2249
Environmental Science & Technology
ARTICLE
Table 2. Concentrations (ppbv) measured by passive and reference methods during the pilot study compounds
n
ambient winds (ppbv)a
fan-induced winds (ppbv)a
passive RSD (%)
reference value (ppbv)b
amb-fan % Δc
amb-ref % Δd
NOX
3
22.3 ( 0.8
22.9 ( 0.6
2.2%
28.0
-3%
-20%
NO2
3
14.1 ( 0.5
14.4 ( 1.2
5.9%
17.2
-2%
-18%
SO2
3
1.4 ( 0.2
1.2 ( 0.2
15.5%
1.7
16%
-18%
formaldehyde
3
1.23 ( 0.04
1.27 ( 0.12
6.4%
1.10
-3%
12%
acetaldehyde
3
0.59 ( 0.01
0.59 ( 0.03
3.4%
1.04
