Size Specific Distribution Analysis of Perfluoroalkyl Substances in Atmospheric Particulate Matter – Development of a Sampling Method and their Concentration in Meeting Room/Ambient Atmosphere

The international regulation of persistent organic pollutants (POPs) according to the Stockholm convention started in May 2001, and is intended to regulate the production and use of hazardous chemicals on a global scale. PFOS is one of the newly listed emerging POPs and only one of a diverse huge group of perfluoroalkyl substances (PFASs), which are known as a “super set” of chemical tracers that include more than ninety related chemicals. The comprehensive monitoring of PFASs is necessary to develop a reliable understanding of environmental kinetics related to these pollutants. However, the extent of atmospheric pollution by PFASs is still unclear because their distribution and sources are not fully understood. Hence, a reliable analytical method for precisely measuring the levels of PFASs in particulate matter is needed. In this study, in order to investigate the levels of PFASs in atmospheric particles including PM2.5, the use of new sampling equipment was evaluated by obtaining multiple samples of air from a stable meeting room environment. Meanwhile, by simultaneously obtaining samples from a roadside environment, the characteristics of PFASs from two different types of air samples were compared.


INTRODUCTION
Perfluoroalkyl substances (PFASs), known as a "super set" of chemicals, include more than 90 related derivatives, and are used in a variety of industrial and commercial applications, including as surfactants in pesticides, surface protectors in textiles, furnishings and food packaging.Perfluoroalkyl sulfonates (PFSAs) and perfluoroalkyl carboxylates (PFCAs) are typically dominated by the presence of eight-carbon perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) derivatives, representing the two main groups of PFASs, and are widely detected in wildlife and humans around the world (Delinsky et al., 2010;Zhang et al., 2010).In May 2001, a total of 164 countries and the European Union reached an agreement intended to regulate the production and use of hazardous chemicals on a global scale, and to protect the environment from excessive discharges of chemicals.Considering the current worldwide use of new types of hazardous chemicals, the previous list was updated and new chemicals were added recently.PFOS (and its salts) and perfluorooctane sulfonyl fluoride (POSF) were listed as new types of an emerging persistent organic pollutant (POP) and as the only member of the large group of perfluoroalkyl substances (PFASs).
Previous research indicated that PFASs, due to their persistence, water solubility, and measurability, could represent excellent tracers of the global circulation of oceanic currents (Yamashita et al., 2008).To that end, the comprehensive monitoring of PFASs is needed to enable a reliable understanding of the environmental kinetics of such a phenomenon.However, the extent of atmospheric pollution by PFASs is still unclear because their distribution in the environment is not well understood.
Regarding the concentration of PFASs in particulate matter, high volume air samplers and passive samplers have been frequently used in previous studies, which reported a dependency on the specific location and region in the world (Dreyer and Ebinghaus, 2009;Genualdi et al., 2010;Kim et al., 2012).Although the number of reports is limited, size segregated behaviours have been also discussed.PFOA sampled in the neighbourhood of a fluoropolymer manufacturing facility in the USA was shown to be mainly composed of fine particles < 0.28 µm (Barton et al., 2006) while PFOA and PFOS in respirable particles in urban areas in Japan were mainly coarse (> 3.3 µm) (Harada et al., 2006).In Germany, PFOA and MeFOSE were found to be predominant in an ultrafine fraction (< 0.14 µm) while the maximum mass fraction of PFOS was in a coarse fraction, with particle diameters between 1.38 and 3.81 µm (Dreyer et al., 2015).Particles and associated chemicals including POPs (i.e., PAHs, PCBs, PBDEs, OCPs, PCDD and PBDD) emitted from anthropogenic sources are likely to be enriched in ultra and fine fractions (e.g., Krecl et al., 2015;Odabasi et al., 2015;Zhang et al., 2015) and this may be similar to PFASs.However, their behaviour and status so far reported appear to be complicated and still not clear especially regarding size segregated behaviour in indoor environments, which has been reported as a major emission source for PFASs (Shoeib et al., 2011).Artefacts during the sampling and analysis may be another reason for uncertainty.Hence, further studies directed at size segregated PFASs with more reliable methods is clearly needed.
In this study, in order to investigate the levels of PFASs in atmospheric particles including PM 2.5 and to provide more information about size segregated PFASs, a new type of instrumentation was evaluated by collecting multiple samples from air in a meeting room.Meanwhile, by simultaneously collecting similar samples from a roadside location, the characteristics of PFASs obtained from two different types of atmospheric environments were compared.To increase the reliability of the analytical method, a protocol, in which mass labelled standards are spiked before sampling to eliminate concerns regarding chemical loss during sampling, is proposed and discussed.

Sample Collection
Sample collections were carried out at the National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba west campus between September 2014 and February 2016.The experiment was designed for two purposes, one was to establish a method for sampling and measuring the concentration of PM that contain PFSAs and PFCAs and the other was to validate the proposed method.
The first experiment was carried out in an air-conditioned meeting room with a controlled temperature (25°C) and relative humidity (45 ± 10%) to avoid influences of weather.This experiment was also considered a meeting room as a potential source of air pollution by PFASs.The latter experiment was carried out at the entrance gate of AIST, which is located a distance of 10 meters from a main road with heavy traffic, as a representative roadside environment.
Two Nanosamplers (Furuuchi et al., 2010), located 1.5 m above ground level, operating at a rate of 40 L min -1 , were used simultaneously to collect samples in parallel.Particles were collected in five size fractions, or, > 10, 10-2.5, 2.5-1, 1-0.5, and < 0.5 µm, respectively.The same sampling was conducted both in the summer and winter, resulting in the collection of a total of 10 sets of meeting room samples and ten sets of roadside samples.
Detailed information regarding the sampling protocols is shown in Table 1.

Sample Preparation
Quartz fibrous filters (φ55 mm, QFF, Pallflex, 2500QAT-UP) pre-baked at 350°C for 1 hour were used for the sampling to minimize possible contamination.All filters were conditioned in a weighing chamber with a controlled temperature (21.5 ± 1.5°C) and relative humidity (35 ± 5% RH) for 48 hours and the filter weight was measured using a microbalance (readability to 1 µg) before and after the sampling.
After the sampling, each QFF was placed in a polypropylene (PP) tube (15 mL) and extracted three times with 4 ml volumes of methanol in a sonication water bath (40°C) for 10 min.The supernatant was collected in a new PP tube and concentrated to 1mL using a stream of nitrogen gas.The extract was applied to Supelclean ENVI-Carb cartridges (100 mg, 1 mL, 100-400 mesh, Supelco, U.S.A.) to remove interfering compounds.The conditioning of the cartridges was carried involved three treatments with 1 mL of methanol.Afterward 1 mL of methanol was added to the cartridge and the extracted sample was directly collected in another PP tube.This procedure was repeated three times.Finally, the extract was concentrated to 1 mL under a stream of nitrogen and transferred to a vial for analysis using a high performance liquid chromatograph coupled with a tandem mass spectrometer (HPLC-MSMS).

Sample Analysis
An HP1100 liquid chromatograph (Agilent Technologies, Palo Alto, CA) interfaced with a Micromass® (Beverly, MA) Quattro Ultima Pt mass spectrometer was operated in the electrospray negative ionization mode to determine the levels of PFSAs and PFCAs in each extract.
A 10-µL aliquot of the sample extract was injected into a Betasil C18 column (2.1 mm i.d.× 50 mm length, 5 µm; Termo Hypersil-Keystone, Bellefonte, PA).The capillary was held at 1.2 kV.Cone-gas and desolvation-gas flows were maintained at 60 and 650 L h -1 , respectively.The source and desolvation temperatures were kept at 120 and 450°C, respectively.MS/MS parameters were optimized so as to transmit the A total of 20 chemicals (7 PFSAs and 13 PFCAs) were analysed.Detailed chemical information can be found in the supporting formation.

QA/QC
Laboratory blanks and recoveries were used to confirm the reliability of the sampling and analysis process.They were analysed along with each batch of samples to check for possible laboratory contamination and the presence of interfering substances.No interfering substances were found in the blanks.The recoveries of these 20 PFASs were been spiked onto the QFF and treated using the same analytical procedures.The overall recovery of the PFCAs and PFASs using the analytical procedure ranged from 71% to 102%.The reported PFASs concentrations were not corrected for recovery.Limit of quantification (LOQ) information is shown in Tables S1 and S2.
Table 2 shows the recovery of surrogate standards spiked on the first stage filters of the air sampler before sampling.Overall, the average recovery rate of the eight 13 C-labeled surrogate standards ranged from 76% to 98%, with standard deviations of less than 16%.For the meeting room samples, the average recovery rates for each of the surrogate standards ranged from 90% to 98% with standard deviations less than 15%.For the roadside samples, the average recovery rates for each of the surrogate standards ranged from 76% to 92% with a standard deviation of less than 16%.This result indicates that most of the PFASs were trapped on the collection filters and no obvious chemical re-suspension occurred during the air sampling.In addition, no significant loss of target chemicals was detected during the extraction procedure.
The relative percentage difference (RPD) for each parallel samples of particle mass concentration (µg m -3 ) and selected chemical concentration (pg m -3 ) is shown in Table 3.The overall average RPD for single chemicals varied from 6% to 34%, indicating that the sampler performance and chemical analysis procedures were stable and reliable.

PFASs Concentration in Meeting Room and Roadside Air
Fig. 1 shows the average concentrations (pg m -3 ) of PFASs in air samples collected from the meeting room and the roadside locations (all stages combined).The LOQ for each chemical varied from 0.01 to 0.84 pg m -3 depending on chemical and sample variation, where detailed information can be found in supporting information Table S3.
The meeting room concentration of PFASs in the particle phase (pg m -3 ) profile was mainly dominated by neutral PFASs -FOSA (5.7 pg m -3 ), N-EtFOSAA (4.7 pg m -3 ), N-EtFOSA (3.0 pg m -3 ) and one ionic PFASs (PFNA, 4.1 pg m -3 ), while the roadside air profile was dominated by ionic PFCAs -PFHxA (7.3 pg m -3 ), PFHpA (3.7 pg m -3 ), PFOA (3.1 pg m -3 ) and PFNA (2.6 pg m -3 ).The concentrations of three neutral PFASs (FOSA, N-EtFOSA, N-EtFOSAA) in the meeting room were substantially higher than the values for the roadside air samples.The meeting room/roadside ratios for these three compounds were 10, 11, 5, respectively, highlighting the importance of indoor exposure.Similar monitoring results were reported in 2011 (Shoeib et al., 2011).In both the meeting room and roadside air, the concentrations of PFCAs were significantly higher than the values for PFSAs.Harada et al. (2006) also reported a 7-50 times differential between PFOA and PFOS in a sampling done in Fukushima.The concentrations of PFOA and PFHxA were about 3, 10 times those of PFOS and PFHxS in both the meeting room air and the roadside air.As a representative ionic PFCA, the concentration of PFOA in indoor air has been reported by other studies (Shoeib et al., 2011;Goosey and Harrad, 2012), while reports concerning the concentration of other ionic PFCAs are limited.In this study, the dominant  PFCAs was PFNA, the level of which was nearly double that for PFOA.Generally, there are two sources for emitted PFNA.One is indirect input, a degradation product derived from FTOs e.g., 8:2 FTOH (C8F17C2H4OH) can decompose into C 5 -C 9 PFCAs through OH radical oxidation (Ellis et al., 2004).The other source is direct input.It was reported that PFNA production is primarily in Japan and that 14 of the world's 33 fluoropolymer production sites are located in Eastern Asia (Prevedouros et al., 2006).Similarly, a higher ratio of PFNA/PFOA in the Western Arctic compared to the Eastern Arctic was reported, which is consistent with the manufacture and use of PFNA/APFN in Eastern Asia (Smithwick et al., 2005).Moreover, the concentration of carboxylate (C 4 -C 8 ) in roadside air increased compared with those of meeting room particles.The meeting room/roadside ratios for these compounds varied from 0.26 (PFHpA) to 0.76 (PFOA).The fact that the meeting room/roadside particle mass concentration ratio was about 0.5 suggests that emission sources other than those in the indoor environment can contribute to roadside contamination.Other sources of contamination in the roadside environment are also possible, which may be responsible for the concentrations of carboxylate PFASs in particles.Also, in both the meeting room and roadside air, the average concentration of PFHxA exceeded that of PFOA, which is consistent with the hypothesis that the use of PFHxA is increasing in response to restrictions imposed on the use of PFOA.

Size Segregated PFASs Concentration (pg m -3 )
Fig. 2 shows the size segregated chemical concentrations (pg m -3 ) of some selected PFASs and particle mass concentrations (µg m -3 ) in the meeting room and roadside air, which represent the first reported values for size segregated PFASs concentrations, except for PFOA and PFOS which were previously reported by Harada et al. (2006).Detailed information is listed in Tables S4 and S5.
The average particle mass concentration in the roadside air was 46 µg m -3 , about 2 times larger than that for the meeting room air (27 µg m -3 ).The particle size distributions in the meeting room and roadside air were different.In meeting room air, the particle mass concentration gradually decreased with increasing particle size.To the contrary, particle size distribution in the roadside environment showed a bi-model behaviour with the main peak appearing at 2.5-10 µm and the second peak at 0.5-1 µm.The main peak indicated a natural source while the second peak indicated a human source, most likely automobile exhaust.
Size segregated concentrations of PFASs of FOSA were enriched in particles sizes less than 0.5 µm in the meeting room air, while roadside air was enriched in particles between 2.5-10 µm and 0.5-1 µm.Such a size dependency of these chemicals is very similar to that of particle mass.Concerning to ionic PFCA, in meeting room air, the concentration of PFHxA and PFHpA was dominated by < 0.5 µm particles, 34% and 38%, respectively.Regarding PFOA and PFNA, the < 0.5 µm stage fraction was decreased to 28% and 19%.On the other hand, the 1-10 µm fraction was increased from 36%, 39% of PFHxA, PFHpA to 44%, 51% of PFOA and PFNA.

PFASs Concentration in Meeting Room and Roadside Particulate (ng g -1 )
Fig. 3 shows the particle-bound PFNA and PFOA concentrations (ng g -1 ) in meeting room and roadside air.Note that, due to a lack of particle weight information, data for the two samples collected in 2014 are not included and eight meeting room samples and nine roadside samples were evaluated.Detailed information is listed in Table S6.Particle-bound PFNA and PFOA concentrations in the meeting room air were up to 4 times larger than those for roadside air.PFOA and PFNA levels were significantly positively correlated (r > 0.88, p < 0.01) for all size ranges of meeting room particles.PFOA and PFNA were positively correlated in the < 0.5 µm stage (r > 0.92, p < 0.01) in roadside particles.
In the meeting room environment, the PFNA concentration reached about 800 ng g -1 , in particle size ranges of < 0.5 µm, 1-2.5 µm and > 10 µm.This is twice as large as that in size ranges between 0.5-1 µm and 2.5-10 µm, or, 350 ng g -1 .The tendency for the PFOA concentration is quite similar and reached about 350 ng g -1 in the case of size rages of < 0.5 µm, 1-2.5 µm and >10 µm while the value was 150 ng g -1 in size ranges between 0.5-1 µm and 2.5-10 µm.The corresponding PFNA/PFOA ratio varied between 1.8 and 3.5.This indicates the existence of similar emission sources, especially for particles larger than 0.5 µm in which the PFNA concentration was about 3 times higher than that for PFOA.To the contrary, in the roadside environment, the particle size dependency for PFOA and PFNA concentration was not clear although it was up to about 100 ng g -1 , except for 200 ng g -1 in the case of < 0.5 µm.The PFOA/PFNA ratio (1.5) was similar to that for the meeting room air (1.8), indicating similar emission sources for these two chemicals in both the meeting room and the roadside environment.
Concerning the other stages (except > 10 µm), the lower R-value and PFOA/PFNA ratios (up to about 1) than meeting room air (3-3.5)suggests that multiple chemical sources contribute to the formation of coarser particles.The above results can be rationalized by the following hypothesis: The pathway for ionic PFASs from the meeting room to outside is through a particle phase, especially through fine particles.Since the doors and windows were rarely opened during daily use and kept closed during the sampling period, the main air exchange between the meeting room to outside is the air ventilation system, which is equipped with air filters.Such a process would eliminate the coarse particles.Thus, this might explain the high correlation in PM < 0.5 µm and the low correlation of larger particles.Table 4 summarizes the concentrations (pg m -3 ) of PFASs in air samples obtained from a meeting room (representative of an indoor environment) and a roadside (representative of an outdoor environment) (this study), and in indoor and outdoor samples from different cities in other selected studies.Indoor air concentrations in Canada (Shoeib et al., 2011) and the United Kindom (Goosey and Harrad, 2012) were more than an order of magnitude higher than roadside concentrations and indoor air was considered to be a source of ambient air.While given the differences in concentrations and compounds between the indoor and roadside samples observed in this study, meeting room air in Tsukuba city did not appear to be the only main contributor to the levels in roadside air.Similar observations have been reported for cities in Korea (Kim et al., 2012).A study conducted in the Kyoto area, in Japan, also reported that the concentration of PFOS and PFOA was 3-20 times as high that for a busy traffic national route compared with a local road, indicating that automobile exhaust may also be a contributor (Harada et al., 2006).Further studies will clearly be necessary to elucidate the presence of PFAS emission sources in outside air.

CONCLUSIONS
This study confirms the reliability and stability of air sampling and chemical analysis for studies of the atmospheric concentration of PFASs.The present method provides new information regarding the concentrations of size segregated PFASs while confirming the recovery rate throughout air sampling and analysis.Using the present method, the concentrations of PFASs in two typical indoor and outdoor environments were evaluated and clear size dependencies were found to exist.

Table 4 .
Summary of the concentrations (pg m -3 ) of PFASs in the meeting room (representing indoor environment) and roadside (representing outdoor environment) air samples (this study), and in indoor and outdoor samples from different cities as reported in other selected studies.Location (# samples)(homes) (n = 59.except for PFOS and PFOA n = 39).(Shoeibet al., Birmingham UK (n = 10).(Goosey and Harrad, 2012

Table 1 .
List of samples collected in a meeting room (representing indoor air) and the main gate at AIST (representing roadside air).The RS5 result is lacking in reliability and only surrogate standard recovery results were included in the discussion.