Alternate materials for the capture and quantification of gaseous oxidized mercury in the atmosphere

Methodologies for identifying atmospheric oxidized mercury (Hg^II) compounds, including particulate-bound Hg^II (${{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}$) and gaseous oxidized mercury (${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$), by mass spectrometry are currently under development. This method requires preconcentration of Hg^II for analysis due to high instrument detection limits relative to ambient Hg^II concentrations. The objective of this work was to identify and test materials for quantitative capture of Hg^II from the gas phase and to suggest potential surfaces onto which Hg^II can be collected, thermally desorbed, and characterized using mass spectrometry methods. From the literature, several compounds were identified as potential sorbent materials and tested in the laboratory for uptake of gaseous elemental mercury (Hg⁰) and ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ (permeated from a HgBr₂ salt source). Chitosan, α-Al₂O₃, and γ-Al₂O₃ demonstrated ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ capture in ambient air laboratory tests, without sorbing Hg⁰ under the same conditions. When compared to cation exchange membranes (CEMs), chitosan captured a comparable quantity of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$, while ≤90 % of loaded ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ was recovered from α-Al₂O₃ and γ-Al₂O₃. When deployed in the field, the capture efficiency of chitosan decreased compared to CEMs, indicating that environmental conditions impacted the sorption efficiency of this material. The poor recovery of Hg^II from the tested materials compared to CEMs in the field indicates that further identification and exploration of alternative sorbent materials are required to advance atmospheric mercury chemistry analysis by mass spectrometry methods.

1 Introduction

Mercury (Hg) is a toxic global contaminant that is introduced to aquatic and terrestrial ecosystems primarily from the atmosphere (Driscoll et al., 2013). Oxidized forms (Hg^II), including gaseous oxidized mercury (${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$) and particulate-bound mercury (${{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}$), are deposited from the atmosphere to ecosystems (Ariya et al., 2015) and may become available for transformation to and accumulation as methylmercury in food webs (Lyman et al., 2020a). A complete understanding of Hg behavior in the atmosphere is necessary to describe the fate of anthropogenic Hg pollution, assess health risks to humans and wildlife, and evaluate the effectiveness of the Minamata Convention. The mechanisms that govern the oxidation and reduction of Hg in the atmosphere are not well understood (Shah et al., 2021), and model results are uncertain because they do not consider the variety of Hg^II compounds that could be present (Gustin et al., 2023). For example, Shah et al. (2021) assumed that all Hg^II compounds volatilized from aerosols are HgCl₂. This is an assumption that has not been validated.

Currently, Br⁻ and Cl⁻ radicals are considered to participate in elemental Hg (Hg⁰) oxidation. This is based on both theoretical work (Holmes et al., 2010; Horowitz et al., 2017; Song et al., 2024) and experimental observations of atmospheric Hg depletion events in the Arctic (Steffen et al., 2008), as well as observations of Hg^II formation at the marine boundary layer (Laurier et al., 2003). The identity of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ compounds in the atmosphere is currently unknown, but mass spectrometry (MS) methods capable of observing atmospheric ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ speciation are in development (see Jones et al., 2016; Khalizov et al., 2020; Mao and Khalizov, 2021). These methods have been developed using HgBr₂ and HgCl₂ as model atmospheric ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ compounds, given the role of halogen radicals in atmospheric Hg⁰ oxidation. However, due to differences in ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ behavior, the use of a broad range of representative compounds is desirable in both MS method development and validation of preconcentration surfaces. MS methods will require preconcentration of ambient ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ for detection, but the preconcentration surfaces currently available for deployment in the field have limitations that prevent their use in MS methods. Commonly used materials include potassium chloride (KCl)-coated denuders in the Tekran 2537/1130/1135 speciation system (Landis et al., 2002) and membranes deployed in the Reactive Mercury Active System (RMAS) (Luippold et al., 2020b). These membranes include cation exchange membranes (CEMs) used for quantitative Hg^II measurement, polytetrafluoroethylene (PTFE) membranes used to quantify ${{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}$ (described in Luippold et al., 2020b, as PBM), and nylon membranes used for estimating Hg chemistry.

KCl denuders do not accurately measure Hg^II in ambient air due to ozone, humidity, and perhaps other interferences (Lyman and Jaffe, 2012; McClure et al., 2014; Huang and Gustin, 2015). PTFE membranes exposed to field conditions have also recently been found to sorb ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$, indicating that this membrane type cannot fully separate ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ and ${{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}$ measurements as intended and provides ${{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}$ measurements that are biased high (Allen et al., 2024). Although CEMs outperform KCl denuders for quantitative ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ capture (Huang et al., 2013) and are quantitative ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorbants under laboratory conditions (Miller et al., 2019; Dunham-Cheatham et al., 2020), recent work suggests that CEMs may not be fully quantitative under field conditions. For instance, Dunham-Cheatham et al. (2023) observed lower Hg^II concentrations collected on CEMs (discussed as RM by Dunham-Cheatham et al. (2023), which includes ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ and ${{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}$ as defined here) in the field relative to co-located dual-channel system (DCS) ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ measurements. However, it is possible that the discrepancy between CEM and DCS measurements observed by Dunham-Cheatham et al. (2023) was due to the 2 L min⁻¹ flow rate used to collect Hg^II on CEMs. Use of 2 L min⁻¹ flow rates has recently been found to decrease Hg capture efficiency relative to those sampled at a 1 L min⁻¹ flow rate (Allen et al., 2024). Increased breakthrough has been detected on downstream CEMs during field campaigns compared to CEMs exposed to ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ for short periods in the laboratory (Allen et al., 2024, and this work), and Hg^II loss from CEM during long campaign periods has yet to be quantified. CEMs are not appropriate for Hg^II sample introduction into MS systems because when heated they generate compounds that interfere with Hg quantification by cold-vapor atomic absorption spectroscopy (Gustin et al., 2019), and research done using high Hg^II concentrations has suggested that exchange reactions can occur with Hg compounds on the CEM surface (Mao and Khalizov, 2021).

Thermal desorption followed by peak deconvolution of Hg^II compounds from nylon membranes deployed in the RMAS is currently the only method available for estimating atmospheric Hg^II chemistry (Huang et al., 2013; Luippold et al., 2020a; Gustin et al., 2023). This method compares thermal desorption profiles of unknown Hg^II compounds to reference profiles developed from Hg^II salts permeated onto nylon membranes to identify potential compound constituents (e.g., −O, $-\mathrm{Br}/\mathrm{Cl}$, −N, −S, and –organic Hg compounds). The validity of thermal desorption interpretations depends on how well the desorption behavior of Hg salts represents unknown atmospheric Hg^II compounds. Exchange reactions involving HgBr₂ and HgCl₂ on CEMs and nylon membranes have also been observed at above-ambient concentrations (Mao and Khalizov, 2021), suggesting that Hg^II compounds desorbed from nylon membranes could be different from atmospheric Hg^II compounds initially captured. This may be true of any new ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorptive surface and should be considered during the validation of new materials.

A DCS circumvents the need for Hg^II preconcentration by converting Hg^II to Hg⁰ using a thermolyzer and measuring total gaseous Hg. The Hg^II concentration can then be calculated by subtracting the Hg⁰-only fraction of atmospheric air, obtained by scrubbing the ambient sample of Hg^II with a CEM, from the total gaseous Hg measurement. DCSs have been successfully calibrated for ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ measurement in the field but do not provide Hg^II chemistry data (Lyman et al., 2020b).

MS methods that can identify and quantify atmospheric Hg^II compounds could be an essential step towards describing Hg chemistry in the atmosphere, but unambiguous determination of the identity of Hg^II compounds via MS has not yet been achieved (see Deeds et al., 2015; Jones et al., 2016). Given the limitations of current Hg^II sorbents, new surfaces that can quantitatively capture Hg^II without compound-altering chemistry are needed to preconcentrate ambient samples to levels above MS detection limits. An ideal material for analysis of Hg^II compounds by MS will be inert to Hg⁰, capture and retain all Hg^II compounds with high efficiency, not promote compound-altering reactions occurring on the material surface, and release atmospherically representative Hg compounds by thermal desorption for downstream analysis.

Characteristics of promising materials include ion exchange with a porous or layered crystalline material (Manos and Kanatzidis, 2016), a high surface area, and a high melting temperature that would facilitate thermal sample recovery and analysis by MS. Materials functionalized with sulfur, such as thiol, thiosemicarbazide, sulfone, and sulfonamide groups, show promise due to their high affinity for Hg^II (Yu et al., 2016). Capture efficiency is increased in base materials by functionalization with active groups that interact with Hg through chemisorption, resulting in the formation of a covalent bond between the Hg atom and material (Ali et al., 2018). However, strong bonding between the material and Hg^II may cause the identity of the Hg compound to be lost upon collection. As a result of strong bonding, such a material may not be suitable for subsequent analysis by MS methods. Materials that capture Hg by physisorption processes (electrostatic interactions) may be desirable if Hg compounds do not undergo chemistry on the sorbent surface, as has been observed for CEMs and nylon membranes at high concentrations (Mao and Khalizov, 2021).

The objective of this work was to explore chitosan, α-Al₂O₃, γ-Al₂O₃, poly(1,4-phenylene sulfide), and perfluorosulfonic acid as candidate materials for preconcentration of atmospheric Hg^II that would be suitable for subsequent analysis by MS. A custom-built Hg^II permeation calibrator was used to load candidate materials with a known quantity of Hg^II for comparison to CEM. Hg^II capture by these materials was also compared under field conditions. It was hypothesized that chitosan, α-Al₂O₃, and γ-Al₂O₃ would quantitatively sorb Hg^II under ambient conditions. Chitosan sorbs Hg^II through both chelation and electrostatic interactions in liquid matrices via amino and hydroxyl groups (Vieira and Beppu, 2006). α-Al₂O₃ and γ-Al₂O₃, which are alumina polymorphs, were materials of interest because they are polar compounds but not acidic and may thus attract ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ compounds without capturing Hg⁰ (Zheng et al., 2019). Alumina polymorphs are stable at high temperatures (Baronskiy et al., 2022), making them ideal for reuse following thermal desorption. α-Al₂O₃ and γ-Al₂O₃ differ in thermal stability and specific surface area and may thus perform differently in terms of capture efficiency and reusability.

2 Methods

2.1 Materials

CEMs, polyethersulfone membranes that are proprietarily treated, were purchased from Pall Corporation (0.8 µm pore size; Mustang-S, P/N MSTGS3R) as sheets and cut to 47 mm diameter disks. PTFE membranes were purchased from Sartorius Stedim Biotech (0.2 µm pore size; P/N 1180747). Chitosan (85 % deacetylated; P/N J64143.18) and α-Al₂O₃ (<1 µm, powder; P/N 0452572.22) were purchased from Thermo Scientific. γ-Al₂O₃ was obtained from Alpha Aesar as a 40 µm powder (P/N 043266.22). α-Al₂O₃ used in this study had a specific surface area of 2–4 m² g⁻¹ and thermal stability of 1200 °C, while γ-Al₂O₃ had a specific surface area of 100 m² g⁻¹ and thermal stability up to 500 °C. Perfluorosulfonic acid membrane sheets (PFSA-M) were purchased from Sigma Aldrich (trade name Aquivion E98-05, PFSA equivalent weight 980 g mol⁻¹ SO₃H, 50 µm film thickness; P/N 802697). Poly(1,4-phenylene sulfide) was also purchased from Sigma Aldrich (P/N 182354). Activated carbon was acquired from Aldrich Chemical Company (P/N 292591) as 4-14 mesh granules and crushed for use in experiments as a powder.

A set of glass tubes used to test PFSA-M were sent to SilcoTek for inert coating with deactivated silica (SilcoNert^® 2000) to test if this improved Hg^II recovery. Based on the results of this test (discussed below), glass tubes without inert coating were used in laboratory and field tests. PTFE frits that were 10–30 µm in pore size, 2.5 mm thickness, and cut to fit 6.35 mm diameter (4 mm inner diameter) tubing were acquired from Savillex (P/N: 730-0065), as were perfluoroalkoxyalkane filter packs used to house membranes (47 mm diameter, P/N: 403-21-47-22-21-2). Optima™ HCl (A466-500), KBr (P205-500), NH₂OH⋅HCl (H330-500), and SnCl₂ (T142-500) were obtained from Fisher Scientific. KBrO₃ was purchased from Acros Organics (268392500). All reagents were American Chemical Society-grade or higher and made with 18.2 MΩ cm type 1 water. Ultrahigh-purity argon (Linde Gas and Equipment Inc.) was used in sorption tests described in the Appendix. Gas-tight syringes were purchased from Hamilton Company (Reno, Nevada, USA). Flow rates in laboratory tests and field campaigns were controlled with critical flow orifices obtained from Teledyne API (P/N 941100).

Candidate materials that could potentially sorb ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ without capturing Hg⁰ were identified from the literature and preliminary work was performed to select promising materials for further experimentation. Of the five new materials tested, only three were considered for further investigation. Preliminary work indicated poor recovery of Hg²⁺ from two liquid-spiked poly(1,4-phenylene sulfide) samples when analyzed using a modified EPA Method 1631 digestion (United States Environmental Protection Agency, 2002). This suggested a matrix interference, and this material was not tested further. Investigation of PFSA-M was also discontinued after poor performance in ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ laboratory tests (discussed below). Preliminary work with poly(1,4-phenylene sulfide) is detailed in the Appendix, with additional data for chitosan, α-Al₂O₃, and γ-Al₂O₃ (Appendices A and B). An alternative digestion method (United States Environmental Protection Agency, 2001) for recovering Hg^II from CEMs, chitosan, α-Al₂O₃, and γ-Al₂O₃ was attempted, and these data are also available in the Appendix (Appendix C).

2.2 Laboratory loading of Hg⁰ and ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ onto candidate materials

Chitosan, α-Al₂O₃, γ-Al₂O₃, and PFSA-M were tested for quantitative Hg⁰ and ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorption in the laboratory. To test for sorption of Hg⁰ to candidate materials, laboratory air was drawn at 1 L min⁻¹ through traps containing test material. A syringe was used to inject 1.2 ng Hg⁰ from a bell jar into the trap (n=9). As a control, activated carbon was loaded by the same method. Traps containing 30 ± 5 mg of chitosan, α-Al₂O₃, γ-Al₂O₃, or shredded PFSA-M were constructed with glass tubing containing a single PTFE frit. The glass tube (6.35 mm outer diameter) was slightly pinched at one end to prevent the frit and test material from being pulled through the trap during loading. Replicates (n=9) of each material were exposed to laboratory air for 3 min, at the end of which the Hg⁰ injection was made. Additional traps (n=9 per material type) that were exposed to laboratory air, but not loaded with Hg⁰, were used to blank-correct loaded samples. The mass of loaded Hg⁰ was calculated based on the Dumarey equation (Dumarey et al., 2010). Hg⁰ recovered from candidate materials was compared to both the calculated mass loaded and Hg⁰ recovered from the method control. The syringe (Hg⁰) tip was placed as close as possible to the materials during loading to minimize loss to the atmosphere or glass tubing.

Hg⁰ was recovered by combustion (EPA Method 7473) using a direct Hg analyzer (Nippon Instruments Corporation, MA-3000) (United States Environmental Protection Agency, 2007). This instrument was calibrated (r²≥0.999) with a primary liquid Hg²⁺ standard (Inorganic Ventures, MSHG-1PPM). At the beginning of each analytical run, three aliquots of a secondary standard, National Institute of Standards and Technology Standard Reference Material 1547, were analyzed to demonstrate instrument performance. Two additional aliquots of 1547 were run every 10 samples or fewer to confirm ongoing instrument performance. A recovery within ±10 % of the certified value was considered acceptable for analysis. This instrument has a detection limit of 0.10 ng Hg⁰.

Sorption of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ to candidate materials was performed with the same procedure, using a custom-built HgBr₂ calibrator (Allen et al., 2024; Gačnik et al., 2024) that releases a constant stream of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ from a salt-based permeation source. The permeation rate is tightly controlled by maintaining constant temperature, pressure, and He flow over the permeation source. The permeation rate of the calibrator can be determined either gravimetrically or by measurement with CEMs (Lyman et al., 2016; Gačnik et al., 2024). These methods have been demonstrated to be equivalent by Elgiar et al. (2024). The CEM method was used to calculate the permeation rate of the calibrator in this study. Briefly, permeation rates are calculated by exposing CEMs to calibrator output, then digesting the CEMs by EPA Method 1631. The recovered mass of Hg is then divided by the exposure time to provide a pg s⁻¹ permeation rate (blank-corrected). Although this calibrator can be used to calibrate ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ measurements in other systems, it was used here to compare sorptive properties of candidate materials to CEMs by delivering consistent, measurable quantities of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$. Chitosan, α-Al₂O₃, γ-Al₂O₃, nylon (n=9 replicates each), and PFSA-M (n=6 replicates) were exposed to the calibrator output for the same duration to ensure equal loading of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ for each replicate. Triplicate blanks of each material were also exposed to laboratory air for an equivalent period of time. Initial experiments loading CEMs in this study measured a 1.76 ± 0.18 pg s⁻¹ permeation rate (mean ± standard deviation, n=9). The permeation rate of this calibrator was reported as 2.2 ± 0.2 pg s⁻¹ in experiments performed concurrently, discussed elsewhere (Gačnik et al., 2024). The difference in observed permeation rate between these two studies is of significance for the use of this system for calibrating Hg^II measurements and should be studied further before it is broadly employed by the research community. A possible explanation for the difference may be the positioning of the calibrator tip at a distance of 2 cm from the CEM during loading (Gačnik et al., 2024) versus at the filter pack inlet (5.5 cm in this work), as HgBr₂ is more likely to come in contact with the filter pack when loaded at the inlet. Work by Allen et al. (2024) suggests that less than 5 % of atmospheric Hg^II is sorbed to the PTFE filter packs after field deployment.

Table 1 Outcomes of materials tested.

Permeation, as measured by CEMs, dropped significantly to 0.42 ± 0.03 pg s⁻¹ during experiment replication. It was suspected that the chamber containing the permeation tube overheated and shut off, cooling the HgBr₂ salt. Returning the heated chamber to 50 °C restored the measured permeation rate to 1.77 ± 0.06 (data available in Table I1). Due to the change in permeation rate across replicate experiments, results are reported as a percent of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered from CEMs rather than a percent of the expected recovery based on the perm rate that was calculated as the mass of Hg^II recovered from CEMs divided by exposure time (pg s⁻¹). ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ was quantified by following a modified version of EPA Method 1631. Briefly, BrCl solution (1.8 % KBr and 1.2 % KBrO₃ in 32 %–35 % $w/w$ Optima™ HCl) was added to candidate materials and membranes in 1 % HCl at a ratio of 3 mL BrCl to 50 mL 1 % HCl and digested at room temperature overnight (see the SI of Dunham-Cheatham et al., 2023, for additional details). Samples were analyzed by cold-vapor atomic fluorescence spectroscopy (Tekran 2600-IVS). The instrument was calibrated (r²≥0.999) at the beginning and end of analysis. A check standard was analyzed every 10 samples and the instrument was recalibrated after a maximum of 30 samples were run. Data were considered acceptable if check standards were recovered within ±15 % of the true value, as per EPA Method 1631. Blanks for each tested material were collected and analyzed with samples during each experiment and were used to correct the analyzed value of samples. These blanks were exposed to laboratory air only (no HgBr₂) during laboratory HgBr₂ exposure tests or were not exposed to air during field campaigns. Mean recovery on CEM blanks (exposed to laboratory air or not) was 0.03 ± 0.01 ng per membrane (±1 standard deviation) across 28 replicates. Mean recovery of α-Al₂O₃, γ-Al₂O₃, and chitosan was 0.01 ± 0.02, 0.01 ± 0.01, and 0.02 ± 0.01 ng per target mass (30 mg), respectively (n=27 for each material).

2.3 Field comparison of candidate materials and CEM

Candidate materials were deployed for three 1-week sampling campaigns in the summer (late July through mid-September 2023) at the University of Nevada, Reno College of Agriculture, Biotechnology & Natural Resources Agricultural Experiment Station, Valley Road Greenhouse Complex (39.5375, −119.8047; 1370 m a.s.l.). This sampling location is within 100 m of Interstate 80 and is impacted by vehicle emissions and long-range transport of pollutants (Gustin et al., 2021; Luippold et al., 2020a). Environmental conditions varied between campaigns, with weekly mean temperatures falling from 25.0 ± 0.5 °C in the first campaign to 22.0 ± 3.4 and 20.8 ± 0.6 °C in the last two campaigns. Relative humidity was 20 % ± 4 %, 48 % ± 16 %, and 35 % ± 5 % for the first, second, and third campaigns, respectively. Temperature, relative humidity, solar radiation, and precipitation information was obtained from the Western Regional Climate Center (https://raws.dri.edu/, last access: 25 October 2024) measurement station located at the test site (39.53917, −119.806; 1370 m a.s.l.) and is available in the Appendix (Table D1).

Traps, constructed as described for Hg⁰ and ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorption tests above, were deployed in inverted RMAS shields (Fig. E1). Three replicates of each material were deployed with a PTFE membrane upstream (to separate ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ and ${{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}$), and three replicate traps were deployed without an upstream PTFE membrane (providing a total Hg^II measurement). A CEM was deployed behind each candidate material to capture breakthrough ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$. Filter packs containing two consecutive CEMs, both with and without an upstream PTFE membrane (n=3 for each configuration), were co-deployed with candidate materials. Critical flow orifices controlled the flow across all collection materials at 1 L min⁻¹, and the flow rate through each trap or filter pack assembly was measured as standard flow at the beginning and end of each campaign using a volumetric airflow calibrator (BGI tetraCal). Measured masses of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ in each trap or filter pack were divided by the volume of air sampled, calculated as the average of flow rates measured at the beginning and end of deployment (L min⁻¹) multiplied by the total sampling time in minutes, to calculate a ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ concentration sampled in ambient air. All membranes and candidate materials were digested by the modified EPA Method 1631 procedure described above.

2.4 Statistical analysis

One-way ANOVA and Tukey's honestly significant difference tests (α≤0.05), comparing the recovery of Hg^II between material types, were performed using R (R Core Team, 2023, version 2023.06.2+561). Means, standard deviations, and t tests were calculated using Excel 2016.

3 Results and discussion

3.1 Laboratory tests for Hg⁰ and ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorption

In this work, chitosan, α-Al₂O₃, and γ-Al₂O₃ were tested for sorption of Hg⁰ and Hg^II in the laboratory and the field. Additionally, preliminary ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorption tests with poly(1,4-phenylene sulfide) and PFSA-M were performed, but these materials were abandoned when a matrix interference was identified for poly(1,4-phenylene sulfide) and PFSA-M demonstrated poor ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovery. A summary of the materials tested and outcomes is available in Table 1, and the details of preliminary work are available in Appendix A and B. No quantifiable Hg⁰ was recovered from blanks or Hg⁰-loaded chitosan, α-Al₂O₃, or γ-Al₂O₃, except for one Hg⁰-loaded γ-Al₂O₃ trap that had low recovery (0.11 ng). Hg⁰ recovery from activated carbon was 1.3 ± 0.4 ng (mean ± standard deviation), indicating reasonably good agreement with the expected recovery of 1.1 ng based on the Dumarey equation (Fig. F1).

Chitosan traps recovered 99 % ± 36 % of the loaded ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ compared to the CEM, while α-Al₂O₃, γ-Al₂O₃, nylon, and PFSA-M recovered less (86 % ± 15 %, 69 % ± 21 %, 81 % ± 7 %, and 26 % ± 7 %, respectively) (Fig. 1). Due to the smaller quantities of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ loaded following the drop in permeation, a small variation in the mass of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered resulted in a larger percent variation. Variation in mass ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered from both the candidate material and CEM also contributed to a larger percent variation. Given the low recovery of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ from PFSA-M, it was not tested further. As breakthrough was not quantified for this material, it is unclear if the low recovery is due to low capture efficiency of the material itself or if the geometry of the trap (shredded membrane packed into glass tubing) permitted greater breakthrough. If this membrane material could be made porous or functionalized onto a porous substrate, it may demonstrate increased capture of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ due to the presence of sulfonic acid functional groups. A comparison of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recoveries was made between PFSA-M loaded with ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ in glass tubing coated with deactivated fused silica and PFSA-M in uncoated glass tubing. Although it was hypothesized that the coating would reduce ${\mathrm{Hg}}_{\text{(g)}}^{\text{II}}$ sorption to the glass tubing (Jones et al., 2016), ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovery was not statistically different between PFSA-M samples loaded in deactivated fused-silica-coated or uncoated tubes (p>0.05; Fig. G1). ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered on CEMs downstream of other candidate materials provided a measurement of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ breakthrough, calculated as a percent of the sum of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered from the candidate material plus the CEM (n=6 replicates for chitosan, α-Al₂O₃, γ-Al₂O₃, and nylon). An average of ≤5 % ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ was recovered downstream of candidate materials, and no quantifiable ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ was recovered on the second-in-line CEM behind a CEM.

Figure 1 (a) ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered from candidate materials (n=9 each for chitosan, α-Al₂O₃, γ-Al₂O₃, and nylon, and n=6 for PFSA-M) loaded with HgBr₂ in laboratory air as a percentage of HgBr₂ recovered on CEMs. (b) ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ breakthrough from candidate materials (n=6 each) as a percent of HgBr₂ collected on CEMs. Error bars represent ± 1 standard deviation from the mean.

The relatively low recovery of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ from α-Al₂O₃ and γ-Al₂O₃, as well as the minimal breakthrough, indicated that ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ was either lost during loading, possibly to the glass tubing, or not recovered from the material matrix by BrCl digestion. Al₂O₃ has been previously used to capture and thermally reduce Hg^II to Hg⁰ with high efficiency in an inert atmosphere (Gačnik et al., 2022), suggesting that EPA Method 1631 digestion is not sufficient to recover Hg^II sample from this matrix. Chitosan, an organic compound, is more easily decomposed, and thus, sorbed Hg is made more available for analysis by acid digestion compared to oxide crystals like α-Al₂O₃ and γ-Al₂O₃. This may explain why a higher percent recovery was observed from chitosan compared to α-Al₂O₃ and γ-Al₂O₃ and highlights the need to consider alternative digestion methods, or possibly thermal desorption, and utilize matrix-matched certified reference materials when considering new surfaces for quantitative ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ capture.

Digestions with HF and HNO₃ have been used to recover metals from refractory silicates and oxides (Zimmermann et al., 2020), and EPA Method 3052 (United States Environmental Protection Agency, 1996) is an established method for Hg. It aims to completely decompose and dissolve the sample by microwave digestion with HF and HNO₃ but also offers alternative matrix-specific reagent mixtures with HCl and H₂O₂. This method may not be suitable for some oxides, including Al₂O₃ and TiO₂, among others, and target analytes (including Hg) can be sequestered by undecomposed sample, leading to low recovery. Re-adsorption of Hg by residual sample matrix during digestion is also noted for activated carbon matrices in the appendix to EPA Method 1631 and could explain low ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovery from α-Al₂O₃ and γ-Al₂O₃ in this study. Microwave digestion of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$-loaded carbon and γ-Al₂O₃ with HBF₄ was attempted as a safer alternative to digestion with HF (Zimmermann et al., 2020), but high background Hg in analytical-grade reagents made data inconclusive. Direct Hg analyzers conveniently overcome matrix interferences by combusting the sample, but atmospheric samples will need to be preconcentrated over 2-week campaigns to collect enough Hg to exceed analytical detection limits, limiting the utility of this method for analyzing Hg^II trends over short timescales.

Figure 2 (a) Hg^II recovery from PTFE + candidate materials deployed in the field for three 1-week campaigns. Hg^II recovery from traps with first-in-line PTFE membranes (${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}+{{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}$) was combined with traps without PTFE (Hg^II) for statistical analysis (n=6 per campaign). Data do not include ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered on breakthrough CEMs. Error bars are ± 1 standard deviation from the mean. Asterisks (^∗) indicate a statistically different recovery of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ on candidate materials compared to CEMs (ANOVA, α≤0.05). (b) The percent of Hg^II recovered from each portion of the trap assembly, including from the PTFE, candidate material, first-in-line CEM (CEM A), and/or breakthrough CEM. Hg^II traps with no upstream PTFE are shown separately from ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ + PTFE traps.

3.2 Hg^II recovery from candidate materials in the field

Candidate materials were deployed with or without a PTFE membrane upstream and with a CEM downstream. Of the total Hg^II recovered from PTFE + Al₂O₃ traps, 66 % was recovered from the PTFE portion of α-Al₂O₃ traps and 55 % of Hg^II was recovered from the PTFE on γ-Al₂O₃ traps, indicating that half or more of the Hg^II recovered was particulate-bound or ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorbed to particles (see below). More ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ was captured on chitosan behind PTFE compared to the equivalent α-Al₂O₃ and γ-Al₂O₃ traps (an average across all three campaigns of 7 ± 3 pg m⁻³ α-Al₂O₃ vs. 28 ± 43 pg m⁻³ γ-Al₂O₃ and 46 ± 36 pg m⁻³ chitosan), but not as much as compared to CEMs (91 ± 45 pg m⁻³). This agrees with the laboratory tests that show poor ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovery from α-Al₂O₃ and γ-Al₂O₃ and relatively greater ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovery from chitosan. Total Hg^II recovery from the entire trap assembly (PTFE membrane (if present) + candidate material or first-in-line CEM + breakthrough CEM) was not statistically different between chitosan and CEM traps during any campaign. Recovery was statistically lower for α-Al₂O₃ and γ-Al₂O₃ in the first and third campaigns and higher for α-Al₂O₃ during the second campaign (Fig. H1a). Field measurements included a downstream CEM that captured Hg^II not sorbed by candidate materials (i.e., “breakthrough”), if present. These data also suggest that either BrCl digestion was not sufficient to recover Hg^II from Al₂O₃ matrices or reduction was occurring on the material surface during the week-long sampling period and Hg^II was lost as Hg⁰.

Recent work by Allen et al. (2024) demonstrated that ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ can be sorbed by particulates on PTFE filters, suggesting that ${{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}$ measurements are biased high with additional ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorbed. For this reason, Hg^II recovered on PTFE filters was added to ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered from downstream candidate materials to yield a total Hg^II measurement, and these data were combined with Hg^II measurements from candidate materials without upstream PTFE. The sum of ${\mathrm{Hg}}_{\text{(p)}}^{\text{II}}$ + ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered from PTFE  + CEM, respectively, has been well correlated with Hg^II measurements on CEMs in previous work (Gustin et al., 2019, 2023). CEMs (n=6, 3 of which included Hg^II recovery from PTFE + CEM) recovered the highest mass of Hg^II m⁻³ air sampled of all materials tested (Fig. 2a). More than 20 % of the Hg^II recovered from the entire trap (PTFE membrane, candidate material, and breakthrough CEMs) was recovered downstream of candidate materials (Fig. 2b), indicating that chitosan, α-Al₂O₃, and γ-Al₂O₃ did not quantitatively measure Hg^II under field conditions. There was a decrease in the Hg^II measured by all materials in the second and third campaigns that coincided with periods of rain and increased humidity, which is consistent with observations of Hg^II washout during rain events (Kaulfus et al., 2017). Of the candidate materials, chitosan performed the best during the first campaign, recovering a similar quantity of Hg^II as CEMs, but decreased in relative recovery during the second and third campaigns. Chitosan is highly hygroscopic (Szymańska and Winnicka, 2015), and the amino functional groups on chitosan are easily protonated at pH<6; thus, it is possible the increased humidity led to a decrease in ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorption capacity due to electrostatic repulsion between protonated amino groups and ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ (Vieira and Beppu, 2006).

4 Conclusions

CEMs outperformed chitosan, α-Al₂O₃, and γ-Al₂O₃ for ${\mathrm{Hg}}_{\left(\mathrm{g}\right)}^{\text{II}}$ measurement in the laboratory and the field, indicating they do not quantitatively capture Hg^II. Candidate materials did not collect Hg⁰. Low recoveries of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ from α-Al₂O₃ and γ-Al₂O₃ may be due to insufficient digestion methods, demonstrating a need to use matrix-specific methods with certified reference materials when testing alternative materials in the future. Promising materials should be tested for sorption of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ and Hg⁰ capture efficiency for a broad range of representative Hg compounds (Dunham-Cheatham et al., 2020); the potential for chemical transformation on the material surface; potential reactions between the Hg sample and other atmospheric constituents, including interferences with humidity (Huang and Gustin, 2015) and ozone (McClure et al., 2014); and performance under both laboratory and field conditions.

Appendix A: Preliminary assessment of gaseous elemental mercury sorption in an argon atmosphere to poly(1,4-phenylene sulfide), chitosan, perfluorosulfonic acid, and α-Al₂O₃

Cold-vapor atomic fluorescence spectroscopy was used to characterize the loss of Hg in an argon (Ar) carrier gas following injection of gaseous elemental mercury (Hg⁰) through a trap containing poly(1,4-phenylene sulfide) (PPS), chitosan, or a shredded perfluorosulfonic acid membrane (PFSA-M). The testing apparatus (Fig. A1) consisted of an Ar cylinder that provided carrier gas and pressure to the sample line, a PTFE sample line into which a trap containing a test material could be inserted with an upstream Hg⁰ injection port, a thermolyzer (>650 °C) to convert gaseous oxidized mercury (${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$) to Hg⁰, a gold cartridge to collect Hg⁰ for analysis, and a Tekran 2500 to measure Hg⁰. Materials were loaded with a gas-tight syringe by injecting Hg⁰ from a temperature-stabilized source through the injection port upstream of a trap containing a sorbent material. The quantity of Hg⁰ loaded was calculated based on the Dumarey equation (Dumarey et al., 2010). Sorbent traps containing 29.4 mg PPS, 29.4 mg chitosan, or a half of a 47 mm diameter PFSA-M (one replicate each) were constructed as described in the main text with 6.35 mm internal diameter uncoated glass tubing and quartz wool plugs.

To test for sorption of Hg⁰ to PPS, chitosan, and PFSA-M, the peak area of 0.4 ng Hg⁰ detected downstream of traps containing material was compared to a baseline peak area of 0.4 ng Hg⁰ injected through a glass trap without sorbent material. Peaks observed following an injection indicated no or partial sorption of Hg⁰ to the candidate material. A Student's t test (α=0.05) was used to assess if the peak area of Hg⁰ detected downstream of the candidate material (n=5 injections of 0.4 ng Hg⁰ each, relative standard deviation – RSD <10 % for PPS) was statistically different from the Hg⁰ peak area detected downstream of the empty trap (n=5 injections, RSD=7 %). The absence of a peak following an injection indicated complete sorption of Hg⁰ to the trap, while significantly lower peak areas indicated partial sorption, and peak areas equivalent to peaks detected downstream of an empty trap indicated no Hg⁰ sorption to the candidate material. PPS and chitosan did not sorb any Hg⁰. More Hg⁰ was recovered downstream of the PFSA-M compared to empty glass traps (RSD of injections <5 %, p<0.01). The reason for this was unclear and this test was repeated a second time with the same outcome (a total of 20 injections through each empty and PFSA-M trap over 2 d). It was concluded that PFSA-M did not sorb Hg⁰.

A trap containing 14 mg α-Al₂O₃ was tested using a similar procedure with a few minor differences. The glass tubing was pinched at one end and contained PTFE frits, rather than quartz wool plugs, to prevent material movement, and the Tekran 2500 was calibrated with Hg⁰ using the bell jar method (r²>0.998) to quantify sorption by α-Al₂O₃. The quantity of Hg⁰ injected was also increased from 0.4 to 1.1 ng Hg, so analysis was performed on Hg mass in the middle of the calibration range. α-Al₂O₃ sorbed 40 % of the injected Hg⁰, demonstrating that significant sorption of Hg⁰ is possible in an Ar atmosphere.

Figure A1 Experimental setup for Hg⁰ and ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorption to candidate materials in an argon atmosphere. During Hg⁰ sorption testing, the line containing the ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ permeation tube was disconnected. Lines leading into the permeation source and to the Tekran 2500 were heated to a nominal 83 °C with heat tape and insulated with aluminum foil to encourage ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ movement through the system and reduce photo-effects. Components in the figure are not to scale.

Appendix B: Preliminary assessment of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorption to PFSA-M and PPS in an argon atmosphere

To test ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorption to candidate materials, the experimental apparatus included a heated (nominally 90 °C) impinger containing a HgBr₂ permeation tube (Fig. A1) that could provide controlled injections of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ through a candidate material trap (35.6 mg PPS or a half of a 47 mm diameter PFSA-M). For Hg^II tests, the sample line was also kept warm with heat tape (∼83 °C) and insulated with aluminum foil. The same injection procedure as described above was used to determine ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ sorption to PPS and PFSA-M, although the empty trap contained PTFE plugs rather than quartz wool. To confirm ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ source stability over time, peak area was observed through an empty trap before and after ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ injections, and a t test was used to check that peak area from injections did not differ from the beginning of the test to the end of the test. Beginning and end injections were not statistically different in peak area.

Statistically different peak areas (p<0.05) were observed between an empty glass trap and both PFSA-M and PPS, indicating these materials sorbed ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ (n=5, RSD≤10 % for injections through an empty trap; n=5, RSD<10 % for PFSA-M; and n=5, RSD=14 % for injections through the PPS trap).

Appendix C: Appendix to EPA Method 1631

An alternative digestion method was attempted to improve recovery of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ from α-Al₂O₃ and γ-Al₂O₃. The appendix to EPA Method 1631 is a similar digestion procedure to Method 1631, but with an additional leaching step using aqua regia (3:1 HCl:HNO₃) before digestion with BrCl. A ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ calibrator (described in the main text) was used to load materials with a known mass of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$, in lieu of an appropriate certified reference material. An expected 0.25 ng ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ was loaded onto each material (30 ± 5 mg chitosan, α-Al₂O₃, or γ-Al₂O₃; n=3 traps each), which was then digested by the appendix to EPA Method 1631 and analyzed by cold-vapor atomic fluorescence. Activated carbon (30 ± 5 mg) was used downstream of chitosan, α-Al₂O₃, and γ-Al₂O₃ to measure ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ not captured by the candidate material. A second-in-line cation exchange membrane (CEM) captured breakthrough from a first-in-line CEM. All measurements were blank-corrected with the appropriate material. Results were highly variable for α-Al₂O₃ (0.13 ± 0.12 ng ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered on α-Al₂O₃ and 0.17 ± 0.15 on breakthrough carbon), and no ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ was recovered from γ-Al₂O₃, with little ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovered from downstream carbon (0.03 ± 0.06 ng ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$). CEM reasonably recovered the expected loaded mass (0.23 ± 0.06 ng ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$), with no quantifiable breakthrough (Fig. C1). The expected 0.25 ng of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ was reasonably recovered from traps containing chitosan and breakthrough activated carbon (0.13 ± 0.06 ng recovered on chitosan, 0.1 ng recovered on downstream activated carbon), indicating that the appendix method worked to recover the mass balance from CEM, chitosan, and carbon matrices. The lack of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovery from α-Al₂O₃, γ-Al₂O₃, and downstream carbon suggests that α-Al₂O₃ and γ-Al₂O₃ may be sorbing Hg^II, but this digestion method is insufficient to quantify it. The appendix to EPA Method 1635 was chosen as a digestion procedure because it is intended for recalcitrant matrices, including coal; however, aqua regia has a matrix-dependent leaching efficiency (Zimmermann et al., 2020). A certified reference material matrix matched to Al₂O₃ may conclusively demonstrate this.

Figure C1 ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ (ng) recovered by the appendix to EPA Method 1631 on chitosan, α-Al₂O₃, and γ-Al₂O₃ compared to an equivalent mass of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ loaded on CEMs. Masses include ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ measured on downstream activated carbon. Error bars represent 1 standard deviation from the mean.

Appendix D

Table D1 Environmental conditions during Reactive Mercury Active System (RMAS) campaigns. These data were downloaded as shown from the Western Regional Climate Center (https://raws.dri.edu/, last access: 25 October 2024). The measurement station was located at the test site (39.53917, −119.806,; 1370 m a.s.l.). Means and standard deviations (highlighted in italics) were calculated by the authors from the presented data.

Appendix E

Figure E1 (a) Field campaign of candidate materials and CEM in inverted RMAS shields. (b) A close-up of an inverted RMAS shield holding (1) glass traps with candidate materials, (2) PTFE membranes, and (3) breakthrough CEMs.

Appendix F

Figure F1 Recovery of Hg⁰ from activated carbon and candidate materials loaded in ambient air by syringe injection. The black line indicates the calculated mass (1.1 ng Hg⁰) loaded based on the Dumarey equation.

Appendix G

Figure G1 ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ recovery from PFSA-M using deactivated fused-silica-coated and uncoated glass tubes compared to recovery from CEMs and chitosan. The recovery of ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ from PFSA-M was not statistically different between coated and uncoated glass traps (p>0.05, two-sample t test).

Appendix H

Figure H1 (a) Total Hg^II recovery from the entire trap assembly (PTFE membrane (if present) + candidate material or first-in-line CEM + breakthrough CEM). Data combine traps with and without upstream PTFE membranes for each material type (n=6). Error bars represent 1 standard deviation from the mean. Asterisks (*) indicate a statistically different recovery of Hg^II on candidate materials compared to CEMs (ANOVA, α≤0.05). (b) Total Hg^II recovered from candidate materials or first-in-line CEMs without upstream PTFE membranes (Hg^II; n=3) shown adjacent to downstream candidate materials or CEMs in traps with upstream PTFE (${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$; n=3) and Hg^II recovered from PTFE membranes (${{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}$). In theory, ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}+{{\mathrm{Hg}}^{\text{II}}}_{\text{(p)}}={\mathrm{Hg}}^{\text{II}}$.

Appendix I

Table I1 Calibrator permeation rates as measured by CEMs.

^∗ ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$ source was not heated during loading.

Appendix J

Table J1 Recoveries (pg m⁻³) of Hg^II from traps following field deployment. Columns labeled Hg^II and Hg^II breakthrough belong to traps deployed without PTFE membranes. Recoveries (pg m⁻³) of Hg^II from PTFE are indicated as ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$, recoveries from CEMs or candidate material downstream of PTFE are indicated by ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g)}}$, and breakthrough from traps with PTFE membranes is indicated by ${{\mathrm{Hg}}^{\text{II}}}_{\text{(g\hspace{0.17em}+\hspace{0.17em}p)}}$ breakthrough.