Occurrence, Seasonal Variation, and Size Resolved Distribution of Arsenic Species in Atmospheric Particulate Matter in an Urban Area in Southeastern Austria.

Extensive information is available on total arsenic in particulate matter (PM), but little is known about the relative contribution of each individual species. Recent studies often focus on inorganic arsenic as arsenite and arsenate, neglecting the organoarsenicals, i.e., methylarsine, dimethylarsine, and trimethylarsine or the corresponding oxidized forms methylarsonate, dimethylarsinate, and trimethylarsine oxide, although they were already first detected in PM in the mid-1970s. This work presents results from more than 300 daily PM10 and further size-resolved atmospheric PM samples in the size range from 15 nm to 10 μm collected in an urban environment in Austria during the course of a year. An ion-exchange-HPLC (with anion and cation exchange columns) and an ICPMS/MS system were used to study the seasonal variations of total arsenic and all species known to exist in PM. Inorganic arsenic was present in significant amounts in all samples with highest concentrations during winter, but also all organoarsenicals were detected throughout the year. We show that their contribution cannot be ignored, as particles smaller than <1 μm can contain up to 35% of the water+H2O2 extractable arsenic as methylated species, but only dimethylarsinate showed a clear seasonal trend throughout the year.

Introduction

Arsenic (As) is a metalloid distributed ubiquitously in the earth’s crust (54th in abundance),[1] also found in water, air, soils, and even living organisms. The keen interest to study the element and its containing compounds is based on its complex toxicity.[2,3] Although air represents an important route of dispersion allowing global transport,[4] little is known about the behavior of arsenic containing species present in the atmosphere, especially regarding methylated As. In air, arsenic exists in the gaseous and the particulate matter phase, but it is predominantly absorbed on particles as a mixture of inorganic arsenic as arsenite (AsIII) and arsenate (AsV) and methylated organoarsenicals, namely methylarsonate (MA), dimethylarsinate (DMA), and trimethylarsine oxide (TMAO).[5]

Methylated As species in air are often considered to be of negligible importance except in areas with arsenic pesticide application or biotic activity, as for example mentioned in a review by Mandal and Suzuki.[6] This is despite the fact that Johnson and Braman, who studied alkyl- and inorganic As in air samples from a variety of urban, suburban, and rural areas, found approximately 20% of the total As (1.7 ng As m–3) to be present in the alkyl-arsenic form back in 1975.[7] Mukai et al. described the seasonality of organoarsenicals in airborne particulate matter, with changes in temperature for two sites in Japan in the mid-1980s.[8,9] DMA and TMAO were thought to have been formed via biomethylation, whereas trace amounts of MA detected in summer were attributed to contamination by methylarsonate containing pesticides used nearby. These results already provide a strong support for a significant contribution of biomethylation to the atmospheric arsenic cycle.

Interestingly, comparing these results to several more recent studies, TMAO was reported not to be present in PM, but AsIII was reported with an increased frequency.[10] Tziaras et al. hypothesized it could be possible, that TMAO is reported incorrectly as AsIII in several of these studies.[11] Since the introduction of inductively coupled plasma mass spectrometry (ICPMS) and its coupling to liquid chromatography (LC), LC–ICPMS is the tool of choice for arsenic speciation analysis in a variety of different matrices, but in connection with due diligence.[12,13] One common problem is to find a separation scheme, which will separate all the common As species, as they feature a wide variety of ionic characteristics and pH dependency. A general approach should be to use at least two complementary techniques to fully separate and characterize arsenic species in a sample, which for example can be achieved by use of anion- and cation-exchange chromatography. The Hamilton PRP-X100, an anion-exchange column for As speciation analysis often used in contemporary studies, does not retain AsIII or TMAO significantly (using mobile phases pH < 7), resulting in peaks observed near the column’s void volume. Therefore, it is not possible to distinguish between AsIII or TMAO without additional sample preparation (e.g., addition of hydrogen peroxide to oxidize AsIII to AsV) or changing chromatography. Tziaras et al. mentioned that most contemporary studies[5,14−17] did not address this point in their applied analytical approach.[11]

The only exceptions are their own work and a study by Jakob et al.,[18] who collected and analyzed PM10 samples from two geographical different locations in different seasons (San Nicolás, Argentina in winter, as an urban area with a temperate climate and Luján de Cuyo, Argentina in summer, as an area with industrial activity with an arid climate). MA, DMA, and TMAO were identified in more than 90% of 49 analyzed PM10 samples, with the conclusion that methylated arsenicals do occur as background chemicals in the atmosphere without a seasonal impact. Conversely, Tziaras et al. reported no seasonal changes for MA and DMA, but for TMAO, which showed higher concentrations during colder months.[11]

In view of the contrasting results found in the literature, the objective of the present work was an in depth characterization of arsenicals found in atmospheric particulate matter in terms of occurrence and seasonal trends. To achieve this, 325 daily PM10 samples were collected in an urban environment during a whole year and analyzed for the arsenic speciation with anion- and cation-exchange chromatography with an ICPMS/MS as an element-selective detector. Additional atmospheric PM samples in the size range from 15 nm to 10 μm (in 14 fractions) were collected to gain new insights into the arsenic species distribution in such particles with a high size resolution.

Materials and Methods

Chemicals and Standards

Nitric acid (HNO3, 65%), hydrogen peroxide (H2O2, ≥ 30% for trace analysis), and ammonia solution (≥25% p.a.) were purchased from Carl Roth GmbH + Co. KG and used as-received, except for the nitric acid, which was sub boiled prior to use. Phosphoric acid (≥85% for trace analysis), pyridine (≥99.9% p.a.) and formic acid (≥98%) were purchased from Sigma-Aldrich and used as-received. Ultrapure water (18.2 MΩ cm) was produced by a Merck Millipore Milli-Q water purification system. Arsenic, germanium, and indium single element ICPMS standards (1000 mg L–1), dibasic sodium arsenate heptahydrate (AsV), nitrocellulose membrane filters (1.2 μm pore size, 47 mm in diameter), and polystyrene Petrislides were purchased from Merck KGaA and used as-received. Sodium dimethylarsinate (DMA) was purchased from Fluka. Trimethylarsine oxide (TMAO) was synthesized in-house according to Merijanian and Zingaro,[19] as well as methylarsonic acid (MA) according to the Meyer reaction.[20] Cellstar tubes (polypropylene, 15 and 50 mL) were purchased from Greiner Bio-One International GmbH. Polycarbonate foils (25 mm in diameter) were purchased from Whatman Nuclepore. DS-515 collection substrate spray was bought from Dekati, Ltd.

Sampling Location

Sampling of airborne particulate matter was carried out over more than a year at one site in Graz, Austria. The sampling location present an urban background, which was located on the premises of the University of Graz (N 47 04.660 E 15 26.935, altitude 350 m). Weather data and other ambient conditions at the sampling site, including air and soil temperature, wind speed, relative humidity, and atmospheric pressure were provided by the University of Graz, Institute of Physics, Institute’s Geophysics, Astrophysics and Meteorology (IGAM), Universitaetsplatz 5/II, 8010 Graz. PM10 mass concentrations were collected from the nearest air quality monitoring station owned by the Land Steiermark (data source: CC-BY-3.0: Land Steiermark–data.steiermark.gv.at).

Collection of Daily PM10 Samples

Medium volume samplers (MVS 6, Sven Leckel Ingenieurbuero GmbH) and one sequential sampler (SEQ 47/50, Sven Leckel Ingenieurbuero GmbH) were used for the collection of PM10 samples. Each device was equipped with an inlet with inserted jets for the collection of particulate matter (PM10) on nitrocellulose membrane filters, in turn placed in filter holders out of polyoxymethylene. Airflow was set to 2.3 m3 h–1 according to EN 12341:2014 Ambient air–Standard gravimetric measurement method for the determination of the PM10 or PM2.5 mass concentration of suspended particulate matter. One sample was collected per day (from midnight to midnight), sampling about 55.2 m3 of air during these 24 h. Filters were stored (average storage time was typically 2–3 months, species stability was examined in pre-experiments, results are shown in Supporting Information Table S1) in Petrislides without incident light at 20 °C. In total, 325 daily PM10 samples were collected between February 01, 2017 and February 25, 2018.

Collection of Size-Resolved PM Samples

In addition to PM10 samples described above, several times a year further size-resolved samples were collected using an electrical low-pressure impactor (ELPI+, Dekati Ltd.). This impactor is a real-time particle spectrometer consisting of 15 stages (from 6 nm to 10 μm), including 14 impactor stages (from 15 nm to 10 μm) plus a filter stage. The 50% cut off diameters (D50%) of the individual stages were 9.83, 5.34, 3.64, 2.46, 1.62, 0.942, 0.599, 0.379, 0.254, 0.154, 0.095, 0.053, 0.029, 0.015, and 0.006 μm. During sample collection, the setup and device settings as recommended by the manufacturer were as follows: air flow was set to 10 L min–1, particles were captured on polycarbonate foils, which were greased with collection substrate spray to prevent bouncing of the particles during collection and the dryer DD 603 (Dekati Ltd.) was used to remove humidity from the samples. Further details can be found in an earlier publication.[21] All data discussed in this work are from two sampling campaigns from July 28, 2017 to August 09, 2017 (171.1 m3 of air were sampled) and from February 16, 2018 to February 26, 2018 (141.4 m3 of air were sampled).

Sample Preparation

PM loaded filters and foils were cut in two halves. 50% of each nitrocellulose filter (PM10 samples) or polycarbonate foil (ELPI+ samples) was used to determine total arsenic. The other half of each filter was used to quantify the arsenic species.

For total arsenic determination, filters and foils were digested in an ultraCLAVE III microwave digestion system (MLS GmbH Mikrowellen-Labor-Systeme). Furthermore, the Standard Reference Material (SRM) 1648a Urban Particulate Matter (National Institute of Standards & Technology), a set of digestion blanks and a set of filter/foil blanks were digested with each set of samples. One filter/foil, filter/foil blank, or ∼10 mg of SRM 1648a, respectively, was mixed with nitric acid (3 mL in case of the PM10 loaded filters and filter blanks, 2 mL in case of the loaded ELPI+ foils and foil blanks and 5 mL in case of the digestion blanks and SRM 1648a) in 12 cm3 precleaned quartz tubes and closed with lids made of PTFE. Each batch consisted of a set of digestion blanks (only HNO3, n ≥ 3), filter/foil blanks (n ≥ 3), SRM 1648a (n ≥ 3) and a set of samples. Argon (>99.9990%) was used to create an initial pressure of 40 bar inside the closed microwave digestion system. After the finished digestion at 250 °C held for 30 min, solutions were transferred into 50 mL Cellstar tubes and filled up with water to contain 10% (v/v) of nitric acid. The SRM 1648a had to be further diluted 1 + 9 with ultrapure water (acidity was adjusted with HNO3) prior to measurement.

For the determination of TMAO, DMA, MA, and total inorganic arsenic as arsenate a combined extraction and oxidation technique was used.[22] Extracts were filtered through 0.2 μm Nylon filters and transferred into HPLC vials.

Determination of Total Arsenic

An Agilent 8800 triple quadrupole inductively coupled plasma mass spectrometer (ICPMS/MS) (Agilent Technologies) was used for the determination of total arsenic. For details regarding the instrument configuration refer to Tanda et al. (2019).[21] The ICPMS/MS was tuned for suitable sensitivity and robustness using a tuning solution containing 1.0 μg L–1 each of Li, Co, Y, Ce and Tl in 2% (v/v) HNO3. Intensity and relative standard deviations (RSDs) for masses 7Li, 59Co, 89Y, and 205Tl were checked together with oxide ratios (156CeO+/140Ce+) and doubly charged ion ratios (140Ce2+/140Ce+). Typical performance is given as Supporting Information (SI Table S2).

Arsenic was determined in reaction mode (scan type = MS/MS) with oxygen (0.25 mL min–1) as the cell gas and the mass-shift m/z 75 → 91 (integration time/mass: 1 s), additionally germanium on-mass m/z 74 → 74 and indium on mass m/z 115 → 115 (integration time/mass: 0.1 s) were monitored. Germanium was used to correct for matrix effects and instrumental drifts. External calibration solutions (0.01 to 10 μg As L–1) were prepared in water with 10% (v/v) HNO3.

Arsenic Speciation Analysis

An HPLC (high performance liquid chromatography) system was coupled to the ICPMS/MS for the arsenic speciation analysis. Chromatographic conditions and changes in the ICPMS/MS setup are listed in an earlier publication.[22] In addition to the detection of m/z 75 → 91 (integration time/mass: 0.3 s) for arsenic, m/z 53 → 53 (integration time/mass: 0.1 s) was monitored to ensure a constant carbon load during chromatography as carbon dioxide was added as optional gas to increase the arsenic signal.[22,23] Standard solutions of TMAO, DMA, MA, and arsenate with concentrations between 0.01 and 10 μg L–1 were prepared externally in ultrapure water with 10% (v/v) H2O2.

Quality Control

Column recovery rates and extraction efficiencies were calculated after determination of total arsenic concentrations with ICPMS/MS in the diluted extracts (1 + 9 with water (1% (v/v) HNO3). Trueness of the methods was evaluated using the SRMs 1648a Urban Particulate Matter and 1640a Trace Elements in Natural Water (National Institute of Standards & Technology (SI Table S3).

Results and Discussion

General Approach for the Determination of As and its Species in Atmospheric PM

Detailed information describing our analytical approach has already been published,[22] but it can be briefly summarized as follows: Each individual sample is analyzed four times after suitable sample preparation for each type of measurement. After a microwave-assisted acid digestion, total arsenic concentrations are determined by ICPMS/MS. Another part of the sample is treated with a mixture of water and hydrogen peroxide to extract the arsenic species and directly oxidize AsIII to AsV. Separation of species is done twice, once on an anion-exchange column to determine DMA, MA, and AsV and once on an cation-exchange column to determine TMAO, which is not retained on the anion-exchange column (SI Figure S1a). Finally, total arsenic is determined in the extracts to calculate extraction efficiencies and column recovery rates.

By reanalyzing the extracts on a cation-exchange column, it is ensured that TMAO is sufficiently retained and correctly identified when present in the sample (SI Figure S1b). The agreement of the two methods in terms of TMAO concentrations determined is shown in the SI (Figure S2). Comparing our analytical approach to the recently published dual column HPLC–AG–ICPMS method (anion- and cation-exchange column in series) developed by Tziaras et al.,[11] chromatographic runtimes are very similar, although two independent measurements are performed. The determined LODs are 3.7 ng L–1 for the determination of total arsenic and 8 ng As L–1 for MA, 9 ng As L–1 for DMA, 17 ng As L–1 for TMAO, and 8 ng As L–1 for AsV. These LODs can be converted to 2 pg m–3 of air for total arsenic and 0.46, 0.50, 0.91, and 0.45 pg As m–3 of air for MA, DMA, TMAO, and Asv, respectively. Calculation is based on a typical volume of 55 m3 of sampled air with the medium volume samplers for the collection of PM10 during 24 h. Limits of detections are lower for the samples collected with ELPI+ due to the higher volume of air sampled. Column recovery rates for anion- and cation-exchange chromatography were determined to be 98 ± 11% and 90 ± 16%, respectively (n = 325).

Determination of Total Arsenic and Extraction Efficiency

In total, 325 daily PM10 samples were analyzed between February 01, 2017 and February 25, 2018 and an average of 25 samples were collected per month. Total arsenic concentration determined in the samples ranged from 0.06 to 3.32 ng As m–3 (arithmetic mean: 0.50 ng As m–3) and were in all samples above the limit of detection of 0.002 ng As m–3. This mean concentration did not exceed the target value (yearly average) for total arsenic in ambient air of 6 ng m–3 in PM10 established in the European Union.[24] Furthermore, there has not even been one single day during the entire sampling period at which this target value was reached/exceeded. Values are in good agreement with typical values for urban background areas found in the European Union ranging from 0.5 to 3 ng As m–3.[25] The sequence of monthly mean values (Figure ) exhibits variations with increased total arsenic concentrations during winter, in particular during February 2017 and 2018. It could be argued that the citizens of Graz use their private cars more frequently during colder months, which might be a reason for the increase during winter. However, a study conducted by the Land Steiermark showed that there are typically no significant variations in traffic volume during summer and winter (by comparing NO concentrations for different seasons as ground level NO is mainly traffic related).[26] It was especially cold during both months (average temperatures: 3.3 and 0.6 °C). Therefore, it is very likely that higher arsenic concentrations can be attributed to increased household heating (burning of fossil fuels). This is also reflected by the two highest monthly mean PM10 concentrations of 48 and 33 μg m–3.

Figure 1

Determined total arsenic concentrations and sum of species (MA+DMA+TMAO+total inorg. As as AsV) in daily PM10 samples combined with PM10 concentrations as monthly averages.

Extraction efficiency for arsenic in PM10 samples was determined to be 58 ± 20% (n = 325). In a previous work, in which we studied the arsenic speciation in PM10 samples inside and outside a respiratory therapeutic cave,[22] an extraction efficiency of 88 ± 26% (n = 52) was reported. In the same study, it was shown that the extraction efficiency decreases with increasing aerodynamic diameter of particles, very likely because of different particle sources and arsenic species present in them. More arsenic is extractable with water+H2O2 during the colder season, in which predominantly smaller particles are produced by the additional burning of fossil fuels. This is shown by the fact that the difference in concentration for total arsenic and the sum of species is smaller during these months (Figure ).

As Speciation Analysis in PM10

Total inorganic arsenic as AsV was found to be the dominant As species in each individual analyzed daily PM10 sample (n = 325), corresponding to 89.1 ± 6.9% of the total water+H2O2 extracted species in terms of As concentration. TMAO, DMA, and MA make up the missing 11%, corresponding to 7.9 ± 5.5, 2.6 ± 2.2, and 0.4 ± 0.2%, respectively. Total inorganic arsenic as arsenate, TMAO and DMA were detected in all 325 PM10 samples. MA was detected in 277 out of the total 325 samples. For further calculations, MA values below the LOD were substituted with the LOD divided by two (LOD/2).

Total Inorganic Arsenic as Arsenate

Daily inorganic arsenic concentrations ranged from 29 to 3076 pg As m–3 (mean: 265 ± 362 pg As m–3). Lowest monthly mean concentration was found in July (90 ± 37 pg As m–3), highest concentration in February 17 (718 ± 863 pg As m–3) (Figure ). Mean inorganic arsenic concentration was nearly twice as high in the winter half-year 17/18 (308 ± 328 pg As m–3) as it was in the summer half-year 17 (163 ± 133 pg As m–3). As explained before, additional burning of fossil fuels during winter act as an important source for PM and inorganic arsenic, as arsenic is generally present in flue gas as oxides or chlorides condensed onto the surface of particles.

Figure 2

Seasonal variation of MA, DMA, TMAO, and total inorg. As as AsV in daily PM10 samples. Total number of samples was 325 with at least 21 samples per month. Box range = 25th and 75th pecentiles, bars = values within 1.5 IQR.

Trimethylarsine Oxide

TMAO was the dominant organoarsenical in the analyzed extracts. Concentrations ranging from 0.92 to 80 pg As m–3 (mean: 17 ± 13 pg As m–3) were determined between February 17 and February 18. Lowest monthly mean concentration was determined for April (10 ± 9.2 pg As m–3) and highest monthly mean for February 17 (29 ± 15 pg As m–3) (Figure ). Calculated half-year means were 14 ± 12 and 19 ± 12 pg As m–3 for summer 17 and winter 17/18, respectively. Upon comparison with other studies, these mean concentrations in PM10 are similar to values stated by Jakob et al. (20.3 ± 15.8 and 16.2 ± 11.5 pg As m–3 for two locations in Argentina during summer and winter, respectively)[18] and Tziaras et al. (36 ± 25 pg As m–3 in Crete, Greece).[11] In a study from 1987, Mukai and Ambe reported TMAO concentrations up to 600 pg As m–3 during summer, but comparable results for the winter months (from 6 to 66 pg As m–3).[8] Regarding the seasonal variation, they suggested temperature and humidity dependent biomethylation processes as the most probable origin of TMAO in PM. However, Tziaras et al. observed the opposite trend, as TMAO was found at about 3 times higher concentrations during autumn and winter compared to spring and summer and Jakob et al. could not find any significant seasonal variation at all. On the basis of these and our own results, it is shown that the presence of TMAO in PM is ubiquitous but the reported divergent seasonal patterns might be the result of sampling periods chosen too short in earlier studies, giving an incomplete picture. Our results (Figure ) show significant day-to-day variations, but no clear seasonal trend.

Dimethylarsinate

Determined daily DMA concentrations ranged from 0.27 to 61 pg As m–3 with a mean concentration of 5.0 ± 4.8 pg As m–3 over the whole sampling period. Calculated monthly mean values were between 1.70 ± 0.72 pg As m–3 for December and 8.7 ± 4.3 pg As m–3 for October. High average concentration was also determined during February 17 with 8.7 ± 12.2 pg As m–3. Average concentrations were 5.4 ± 3.3 pg As m–3 for the summer half-year and 4.1 ± 3.5 pg As m–3 for the winter half-year. Mukai and Ambe found DMA concentrations between 3 and 53 pg As m–3 for the two sites studied in Japan with a similar seasonal variation as they found for TMAO.[27] Jakob et al. reported mean concentrations of 4.5 ± 2.1 pg As m–3 (summer) and 4.9 ± 3.8 pg As m–3 (winter) at their two sites in Argentina.[18] DMA was detected by Tziaras et al. in 11 out of 33 PM10 samples analyzed at an average concentration of 6.8 ± 6.0 pg As m–3.[11]

Methylarsonate

Methylarsonate concentrations were in 85% of all samples above the limit of detection. Highest daily mean concentration was 5.2 pg As m–3 during Feburary 2017, the average of all measured values was 0.77 ± 0.59 pg As m–3. Similar to DMA, a slightly higher half-year average was calculated for the summer half-year than the winter half-year with 0.74 ± 0.46 pg As m–3 and 0.65 ± 0.38 pg As m–3, respectively. The oldest works on organoarsenicals by Mukai and co-workers reported significantly higher MA concentrations than any other publications. During two consecutive summers MA concentrations up to 100 and 300 pg As m–3 were measured at a specific site in Japan.[9] In a follow up study even elevated levels of MA up to 1.4 ng As m–3 were found to be present in PM at the same site.[27] They attributed the high levels of MA to contamination by an alkylarsenic pesticide (iron methane arsonate) spread as a sterilizer on rice fields near this sampling location. Jakob et al. reported MA concentrations with 2.3 ± 2.1 and 2.9 ± 1.8 pg As m–3 at their two examined Argentinian sites during summer and winter, respectively.[18] Tziaras et al. were only able to detect MA in 6 out of 33 PM10 samples collected, with concentrations above the LOD between 2.0 and 5.3 pg As m–3.[11] They could not detect MA in any of their additionally collected PM2.5 or PM2.1 samples. They assumed that MA might be preferentially present in the coarse fraction (in particles with an aerodynamic diameter between 2.5 and 10 μm), but had no further evidence to support their theory. Results presented below will show that all organoarsenicals are typically present in particles <1 μm in size. Therefore, the reported trend in MA’s distribution found by Tziaras et al. is very likely caused by concentrations close to their method’s LOD. On the basis of these and our results, it seems reasonable to conclude that MA is typically present in the lower pikogram per cubicmetre range, as long as industrial products such as herbicides are not used or other natural point sources are not present near the sampling site.

Seasonal Variation of Organoarsenicals in PM10

As shown in Figure , we were not able to observe a clear trend for the seasonal variation of methylarsonate and trimethylarsine oxide in contrast to dimethylarsinate in airborne PM sampled in Graz between February 2017 and February 2018. Furthermore, the sum of the individual species showed no definite correlation with the mean air temperature during the sampling period (Figure ). Furthermore, all monitored meteorological parameters including air and soil temperatures, wind speed, relative humidity, and atmospheric pressure were correlated with the daily concentrations determined for each arsenic species, but no meaningful relationships became obvious. However, monthly mean concentrations tended to be slightly higher between June and October (compare Figure ).

Figure 3

Concentration of methylated arsenic species present in PM10 samples and air temperature as monthly averages.

High monthly trimethylarsine oxide and dimethylarsinate concentrations determined in February 2017 must be discussed separately in this context. Elevated monthly averages are caused by just a few days with very high daily concentrations. Measurements of soil temperature at a depth of 10, 20, and 50 cm showed simultaneous thawing of the upper Earth layers after a long cold period (compare SI Figure 3). Similar observations were reported for grounds in a Boreal forest in Hyytiälä, Finland, when the snow cover began to melt.[28] Measurements going in Pallas, Finland for nearly two decades, also show a small arsenic spike each June at the same time when the snow melts there.[29] The existence of a snow cover appears to be of minor importance, as no snow covered the ground during the investigated period in Graz, Austria. However, springtime thawing and the coupled increase in soil moisture affect arsenic release by increased solubility and microbial activity. Low O2 concentrations in the soil lead to the reductive dissolution of iron and other metals, which can also free arsenic. As a result, this arsenic is converted into volatile forms through methylation by bacteria, fungi and certain eukaryotes when the soil redox potential decreases.[30−32] In addition, increasing soil temperature can promote microbial metabolism to overcome the ∼12 kcal mol–1 barrier to form methylated arsenic species.[9] A more unlikely scenario, but one which is still possible, is larger climatic changes leading to a long-range transport and deposition of organoarsenicals biovolatilized from seawater. A recent study by Savage et al. showed a significant flux of trimethylarsine oxide from the ocean to the atmosphere.[33]

Size Resolved Arsenic Distribution

In addition to the PM10 samples described above, several size-resolved samples were collected using an electrical low-pressure impactor separating particles with an aerodynamic diameter between 15 nm and 10 μm. Results from two sampling campaigns during summer 2017 (July 28 to August 09, 171.1 m3 of air were sampled) and winter 2017/18 (February 16 to February 26, 141.4 m3 of air were sampled) are shown in Figure . Highest arsenic concentrations for all arsenicals were determined in particles with an aerodynamic diameter between 250 and 600 nm. A monomodal distribution was observed in both seasons. Inorganic arsenic was distributed over the whole size range, whereas organoarsenicals featured narrow distributions below 1 μm. Particles between 150 and 500 nm are part of the so-called accumulation mode, typically formed by condensation and coagulation, producing a surface area maximum in the given size range.[34] An explanation for the enrichment of organoarsenicals in this size range could give the experiments by Jakob et al.[18] It was shown that methylated arsines exhibit smaller lifetimes in air than AsH3 and half-life times decrease linearly with the number of As–C bonds. Furthermore, the stability of all arsines is decreased rapidly, when comparing daytime with nighttime conditions. Most likely reactions include oxidation of the arsenic from the trivalent to the pentavalent form, which in turn forms water-soluble nonvolatile oxo-acids.[35,36] In the case that volatile organoarsenicals are formed, their oxidation products find their way into small water droplets and are adsorbed onto atmospheric particles.

Figure 4

Size resolved arsenic distribution in atmospheric PM in the size range from 15 nm to 10 μm collected in different seasons. (a) Sum of species collected during Summer 2017 (left) with percentage share in water + H2O2 extracts (right). (b) Sum of species collected during Winter 2017/18 (left) with percentage share in water + H2O2 extracts (right).

Total inorganic arsenic (sum of arsenite and arsenate) was determined as the dominant species in all size fractions, with higher concentrations during winter, because of the additional anthropogenic input. In contrast, higher concentrations of all organoarsenicals were found during summer. These results show that methylated species can contribute substantially to the total arsenic concentration in particles with aerodynamic diameters <0.25 μm. During summer, organoarsenicals made up to 35% of the total arsenic in the given size range. At present, only total arsenic in PM10 is regulated by law, with an air quality standard set at 6 ng m–3 by the European Union.[37] However, studies on fine and especially ultrafine particles gained enormously in importance, due to the particles’ ability to reach the deepest parts of our lungs or bodies by crossing the air–blood barrier. Our results show that these particles in particular contain the highest diversity of arsenic compounds.