Reactive mercury flux measurements using cation exchange membranes 1

1 Reactive mercury flux measurements using cation exchange membranes 1 Matthieu B. Miller1*, Mae S. Gustin2, Grant C. Edwards1,3 2 1Faculty of Science and Engineering, Department of Environmental Science, Macquarie University, Sydney, NSW, 3 2113, Australia 4 2Department of Natural Resources and Environmental Science, University of Nevada, Reno NV, 89557, United 5 States 6 3Deceased 10 September 2018. 7 Correspondence to: Mae Sexauer Gustin mgustin@cabnr.unr.edu 8 ABSTRACT 9 A method was developed to measure gaseous oxidized mercury (GOM) air-surface exchange 10 using 2 replicated dynamic flux chambers (DFCs) in conjunction with cation exchange 11 membrane (CEM) filters. The experimental design and method was developed and tested in a 12 laboratory setting, using materials collected from industrial scale open pit gold mines in central 13 Nevada, USA. Materials used included waste rock, heap leach ore, and tailings, with substrate 14 concentrations ranging from 0.1 to 40 g g-1 total mercury (THg). CEM filters were used to 15 capture GOM from the DFC sample lines while a Tekran® 2537A analyzer measured GEM 16 concurrently. Previous and ongoing work demonstrated that CEM do not collect GEM and 17 efficiently collects multiple compounds of GOM. Positive GOM emission rates up to 4000 pg m18 2 h-1 were measured from tailings materials with high Hg substrate concentrations, and this has 19 significant implication with respect to air-Hg surface exchange. GOM flux was variable for 20 lower Hg concentration substrates, with both emission and deposition observed, and this 21 wasaffected by ambient air GOM concentrations. For substrates that experienced GOM 22 deposition, deposition velocities were in the range 0.01 – 0.07 cm s-1. 23


Introduction
On 16 August 2017, the United Nations Environment Program (UNEP) Minamata Convention came into force with a mission to protect human health and the environment from exposure to the toxic effects of mercury (Hg) and its various compounds (UNEP, 2013).The Convention proposes to fulfill this mission through a strategy of globally coordinated scientific research, and ongoing monitoring of Hg in the environment where possible.The mission poses a significant challenge as there are over 3000 known Hg contaminated sites worldwide due to mining and industrial activity, not including legacy and small scale operations involving Hg (Kocman et al., 2013;Krabbenhoft and Sunderland, 2013).Past emissions were dominantly from natural sources, but are now largely from anthropogenic activities and have led to large reservoirs of Hg in all environmental compartments (Pacyna et al., 2016;Pirrone et al., 2010;Streets et al., 2017).Even in the absence of new anthropogenic emissions, cycling of Hg between different spheres will continue at large scales and in possibly unexpected directions, driven by processes and mechanisms that are not fully understood, and susceptible to increasing human and climate perturbations (Obrist et al., 2018).
The atmosphere, a global common, is the dominant conduit for transport, transformation, emission, deposition, and re-emission of Hg and Hg compounds amongst Earth's ecosystems.
Mercury in the atmosphere is classified on the basis of three physiochemical forms: gaseous elemental mercury (GEM), gaseous oxidized mercury (GOM), and particulate bound mercury (PBM), with PBM and GOM defined together as reactive mercury (RM = GOM + PBM) (Gustin et al., 2015).Atmospheric concentrations of GEM can be measured with well calibrated analytical instruments, whereas the quantification of GOM and PBM has depended on operationally defined methods with demonstrably large uncertainty (Cheng and Zhang, 2017;Gustin et al., 2015;Gustin et al., 2013;Jaffe et al., 2014;Zhang et al., 2017).
Scientific understanding of the biogeochemical cycling of Hg is currently inadequate due to the large uncertainties in measurements, and general difficulty of research on atmospheric Hg chemistry (Jaffe et al., 2014).Atmospheric GEM is relatively inert, has a low deposition rate, and hence a relatively long atmospheric lifetime ranging from minutes to 1 year (Krabbenhoft and Sunderland, 2013;Zhang et al., 2009;Gustin et al., 2013).Observations of rapid depletion of GEM from the atmosphere suggests that deposition and re-emission on short time scales is an important process driving movement (Howard and Edwards, 2018;Lu et al., 2001;Schroeder et al., 1998).In contrast to GEM, GOM compounds have higher dry deposition velocities (Zhang et al., 2009).Knowledge of concentrations, chemistry, and processes forming atmospheric GOM is critical for understanding how Hg moves and impacts ecosystems globally.Recent research on GOM has demonstrated that compounds in air vary both spatially and temporally to a considerable extent, pointing to the need for extensive measurements of concentrations and identifying the specific compounds (Gustin et al., 2016;Huang and Gustin, 2015b;Huang et al., 2014;Huang et al., 2017).
Fluxes of total gaseous mercury (TGM) and GEM have been successfully measured in many environments, providing insights into air-surface exchange that is critical for understanding biogeochemical cycling of GEM (Yannick et al., 2016;Zhu et al., 2016).In contrast, there are few direct measurements of GOM air-surface exchange (Zhang et al., 2009).GOM fluxes reported in the literature are largely based on measurements made with the KCl denuder as the GOM collection method, as well as a passive Hg dry deposition sampler (Brooks et al., 2008;Castro et al., 2012;Engle et al., 2005;Lindberg et al., 2002;Lindberg and Stratton, 1998;Lyman et al., 2009;Lyman et al., 2007;Malcolm and Keeler, 2002;Poissant et al., 2004;Rea et al., 2000;Rothenberg et al., 2010;Sather et al., 2013;Skov et al., 2006;Zhang et al., 2005;Huang et al., 2015b).The few direct GOM-specific air-surface exchange measurements using KCl denuder-based approaches have been undertaken with mixed results (Brooks et al., 2008;Skov et al., 2006).Moreover, recent evaluation of KCl denuder approaches show these methods tend to underestimate GOM and are subject to interferences due to ozone and water vapor (Huang and Gustin, 2015a;Lyman et al., 2010;McClure et al., 2014).Understanding GOM air-surface exchange processes would facilitate the Hg scientific community's ability to move forward in understanding the biogeochemical cycling of atmospheric Hg.
In this study, we developed and applied a novel experimental approach for direct measurement of GOM fluxes using cation exchange membrane (CEM) filters.CEM filters have been successfully deployed to measure GOM in ambient air in previous studies (Gustin et al., 2016;Huang et al., 2017;Huang et al., 2013;Marusczak et al., 2017;Pierce and Gustin, 2017), and their use here was modified to allow for the determination of GOM air-surface exchange.A recent study (Miller et al., 2018) and ongoing work (unpublished data) have demonstrated that GEM is not collected by the CEM and it efficiently collects a variety of GOM compounds.Here is described a series of experiments that were conducted to optimize the CEM/GOM flux methodology, and then, the resulting method was subsequently applied to determine fluxes over both background waste rock material and Hg enriched mining materials.Our research hypothesis was the system developed would provide a means of determining if GOM is deposited or emitted from substrates, and this system would allow us to develop an understanding of GOM air-surface exchange.

Materials
Substrate materials used for measuring Hg flux were acquired from industrial scale, open pit gold mines in central Nevada, and include waste rock, heap leach ore, and tailings.Materials were obtained from 4 ongoing mining operations: Twin Creeks (TC) and Lone Tree (LT) operated by Newmont Mining Corporation, and Cortez Pipeline (CP) and Gold Strike (GS) operated by Barrick Gold Corporation (Fig. 1; Table 1).There were 3 types of materials: waste rock called cap (C), heap leach material (L), and tailings (T or tails).(Waste rock is nonmineralized low Hg overburden consisting of alluvium or hard rock that covered the ore body prior to mining.For this study, materials were sampled from the waste rock piles specifically set aside for future capping and site reclamation, and as such is referred to as cap material.Heap leach is low-grade ore blasted from the mine wall and "heaped" on an impoundment for irrigation with dilute cyanide leach solution for gold extraction.Tailings are the waste remnant of high-grade ore that has been pulverized by mechanical ball milling and undergone thermal/chemical treatment to extract gold.All mining substrate materials were collected in September 2010, for measurements of GEM flux under controlled conditions (Miller and Gustin, 2013).
Each material was divided into replicate trays (50 x 50 x 7 cm 3 plywood lined with 3mil polyvinyl sheet) and stored inside a greenhouse bay at the University of Nevada Reno

Methodology
GOM flux was measured by modifying our existing GEM flux method consisting of a dynamic flux chamber (DFC) and an ambient air Hg analyzer (Tekran ® 2537A) after the methods of Eckley et al. (2010;2011a) and Miller et al. (2011).
Briefly, the GEM flux system used a 2537A in conjunction with a cylindrical DFC (footprint 0.036 m 2 ) made of molded Teflon film (0.19 mm thick) over a rigid Teflon frame (1.5 mm thickness), with a total internal volume of 2.0 Liters (Fig. 2).Air enters the chamber through 24 inlet holes (1 cm diameter) spaced 2.5 cm apart around the perimeter and 2.0 cm above the bottom edge.The sample outlet is 0.625 cm diameter PTFE tubing at the top-center of the chamber, and sample inlet air is measured through equivalent tubing at the height of the chamber inlet holes.A Tekran ® Automated Dual Switching (TADS) unit was used to cycle the sample flow (1.0 Lpm) between the chamber inlet and outlet lines in sequential 10 min intervals (two 5 min samples on each line).GEM in each 5 min sample volume was quantified automatically by the 2537A analyzer using pre-concentration on gold traps followed by thermal desorption (500 where F is the net Hg flux (ng m -2 h -1 ), Q is the flow rate through the chamber (m 3 h -1 ), Co is the mean concentration of two consecutive 5 min outlet air samples (ng m -3 ), Ci is the mean concentration of inlet air in the samples before and after Co, and A is the area of the substrate under the chamber (m 2 ).The sign of ΔC indicates the direction of flux with positive being emission and negative being deposition.
The two modified systems included fitting the DFC inlet and outlet sample lines with 2-stage disc filter assemblies (Savillex © 47 mm PFA Teflon Filter Holder) holding two inline polysulfone cation exchange membranes (CEM 0.8 m, Mustang ® S, Pall Corporation) (Fig. 2).
CEMs preferentially capture GOM compounds while allowing GEM to pass freely (Miller et al. 2018; unpublished data).The first upstream CEM served as the primary collection filter, while the second downstream CEM captured any Hg escaping from or missed by the first filter (referred to as breakthrough).With CEM filters deployed at the front of the sample lines, in conjunction with the 0.2 m particulate filter at the rear sample inlet of the Tekran ® , GOM was scrubbed from the sample flow and all Hg measured downstream on the 2537A is in GEM form.
An additional set of CEM filters was deployed simultaneously, using sample pumps and mass flow controllers set to match the 2537A sample flow rate (1.0 Lpm, Fig. 2) at a height of 2 m With both the Tekran ® and external pump controlling CEM sample lines at 1 Lpm, total flow through the DFC was 2.0 Lpm that provides a chamber turn over time (TOT) of 1 minute.At this flow rate, flow velocity through the chamber is laminar with no turbulent eddy formation, as determined through computational fluid dynamic (CFD) modelling of the chamber geometry performed by Eckley et al. (2010).Thus the chamber design of Eckley et al. (2010) is optimal in terms of not disturbing the substrate being studied.With low velocity, non-turbulent flow, particle entrainment from the substrate surfaces is not expected, especially for particle sizes greater than the CEM filter pore size of 0.8 m.

Analyses
After flux measurements, CEM filters were collected into sterile 50 mL polypropylene centrifuge tubes, and frozen at -20 °C until analyses (within 14 days of collection).Filters were analyzed for total Hg by aqueous digestion and cold vapor atomic fluorescence spectrometry (CVAFS, EPA Method 1631, Rev. E) using a Tekran 2600 system, with total Hg operationally equivalent to total GOM.The blank Hg mass that can be expected on an unused CEM filter (median = 68 pg, n = 56) was determined from clean filters collected with every set of measurements, and the median blank value was subtracted from all sample values.Breakthrough is defined as the amount of Hg on the secondary filter as a percent of the total Hg collected on both filters (after blank correction), and overall median breakthrough was low (4.2%, n = 222).(pg h -1 ).Reactive Hg flux (pg m -2 h -1 ) was calculated with Eq. 1 using ΔCGOM values, the flow rate (1.0 ± 0.005 Lpm) and the chamber footprint (0.036 m 2 ).
The GOM flux detection limit was determined by the minimum statistically resolvable difference between Co and Ci, i.e. the smallest meaningful ΔCGOM that could be measured (Fig. 3).The Co and Ci concentrations were based on two measurements: total Hg on the CEM filters as determined by analysis on the Tekran ® 2600 system, and total sample volume as determined by the mass flow-controlled sample rate and time.The MFC precision was ± 0.5% (± 0.005 Lpm at 1.0 Lpm) and the detection limit of the 2600 was 1 ppt, or ~ 53 pg per reagent blank in a clean 50 mL collection tube.The median mass of Hg on the blank CEM filters was 68 pg and was used as the practical detection limit for the method.As the distribution of CEM blank values was nonnormal and skewed heavily to the right (Fig. 3a), the 95% confidence interval around the median (58 -73 pg) was used to define a minimum detectable GOM concentration (Fig. 3b).For example, in a 24 h sample, the minimum detectable GOM concentration would be ~ 3.5 pg m -3 to exceed the upper 95% confidence limit of 73 pg.In a 24 h sample the minimum resolvable ΔCGOM would be 13.5 pg m -3 (sum of upper + lower confidence intervals, + 3 pg m -3 to account for flow precision, converted to 24 h sample volume concentration).This was taken as the upper minimum ΔCGOM value.immediately following chamber cleanings (24 h acid wash, 10% HNO3), and then between substrate types (i.e.cap, leach, tailings).For the summer measurement period, the median chamber blank ΔCGOM for both Chambers A and B (A = 13 pg m -3 , B = 14 pg m -3 , n = 12) was at the detection limit, so chamber blanks were not subtracted from material fluxes.This was due to high ambient GOM concentrations in the greenhouse.For the following winter measurement period, the median chamber blank ΔCGOM for both chambers was significantly negative (A = -60 pg m -3 , B= -60 pg m -3 , n = 6), and the median chamber blank was -215 pg m -2 h -1 , indicating that the chamber and the blank Teflon sheet were acting as depositional surfaces for GOM.For these measurements, the chamber blanks were subtracted.For materials that demonstrated net GOM deposition, the deposition velocity (Vd cm s -1 ) was calculated using Eq.2:

Development of the method
Two Tekran ® 2537A analyzers with associated GOM filter systems were used to simultaneously measure flux from two replicate trays (one for each A and B system) of each sample material.
The duplicated systems allowed a total of four GOM flux measurements to be collected each time a material was tested, two from the Tekran ® sample lines and two from the external pump lines.The intention of the duplicate systems was to evaluate consistency and repeatability of the measurements.However, during the initial method testing it became apparent that the position of the trays and the inlet sample lines was an important variable.For example, strong GOM emission from an "upwind" tray could substantially increase the inlet GOM concentration of the "downwind" tray, resulting in a negative ΔCGOM and apparent deposition.Such contradictory results compelled us to test a variety of orientations for the two systems.The best results were achieved in the final design by placing both trays all the way against the intake wall of the greenhouse bay, and separating them laterally by 1.5 m, with all equipment located downwind.
This configuration provided the most uniform inlet air concentrations for both systems.

Limitations of method
The use of filter membranes in conjunction with a DFC to measure GOM flux has several limitations.Given the low concentrations of GOM, a relatively long sampling time of at least 24 h is required to capture a sufficient mass for quantification on low Hg containing substrates.
Despite the limitations, this method moves us a step further for understanding GOM flux.The low temporal resolution limits analysis of the factors controlling GOM flux, since there will be changes on a diel cycle similar to observations of GEM flux (c.f.Gustin et al., 2013).In addition, flow rate and surface area could be increased, but in order to change this method a number of proof of concept tests would have to be done.
A necessary condition of the DFC method is the placement of flux chambers directly on a material, resulting in artificial modification of surface conditions and a presumptive influence on the magnitude of flux.In our study, this limitation is immaterial, as the entire experimental setup constitutes an artificial environment, and we are more interested in 1) our ability to experimentally detect measurable GOM flux, and 2) the qualitative direction of flux versus absolute quantification.As the method develops, it will become necessary to more thoroughly evaluate the effects of the DFC on GOM flux.
Lastly, CEM were analyzed within two weeks of collection.Pierce et al. ( 2017) demonstrated that little HgCl2 was lost from the membranes over this time period.This needs to be tested for other GOM compounds thought to be present in the atmosphere (Huang et al., 2013;Gustin et al., 2016).

GOM flux measurement repeatability
The first test of the optimized configuration was a set of replicate measurements made on a single material, to assess method repeatability.The material used for this test was a heap leach ore of intermediate total Hg concentration (TCL, 13.2  2.0 g g -1 ).Filters were deployed in three consecutive sample sets of approximately 72 h each, for a total of 12 replicate filter flux measurements (i.e. two GOM flux measurements on two separate replicate trays, three times).
The mean GEM flux from all samples was 167  57 ng m -2 h -1 (n = 6).The duplicate replication displayed by the system, both in general and in detail, for multiple substrate types, increases confidence that these measurements represent a real net surface exchange of GOM.
The mean ambient GOM in the greenhouse during this period was 130 ± 55 pg m -3 (n = 24); however, there was a distinct trend of increasing ambient GOM over the course of measurements.During the TCC measurements, ambient GOM concentration in the greenhouse was 70 ± 20 pg m -3 (n = 8).This increased to 150 ± 20 pg m -3 (n = 8) during the LTL measurements, and up to 180 ± 40 pg m -3 (n = 8) during the TCT measurement.

GOM flux measurement over complete range of materials: Winter
A full set of 24 h GOM flux measurements including all the available mining materials was made in the following winter period (January -March, 2016).Ambient GOM concentration in greenhouse air was 50 pg m -3 (n = 16), much lower than in the summer months.Mean RH and solar radiation were similar between the summer and winter periods, due to the attenuating effect of the greenhouse.However, mean air and substrate temperatures were significantly lower in the winter (Table 1).
All measured fluxes were above the ΔCGOM detection limit (Fig. 7a).The relationship between GOM flux measured on the A and B systems was slightly less than 1:1 (B = 0.87*A) and not as strong (r 2 = 0.74) as during the summer measurements.Without chamber blank correction, GOM fluxes were uniformly negative for all materials except the very high Hg TCT (Fig. 7b, orange shades).However, with the chamber blank correction applied, GOM flux from the cap materials became ambiguous (i.e. both deposition and emission observed) and positive for all leach and tailings materials except the LTT (Fig. 7c, green shades).
Three replicated materials had consistent fluxes, with or without blank correction: GS cap (GOM deposition, Vd = 0.02 -0.04 cm s -1 , corrected), LTT (GOM deposition, Vd = 0.01 -0.05 cm s -1 , corrected), and TCT (strong GOM emission).The GS mine exploits a predominantly carbonaceous ore deposit, and much of the mine and surrounding areas are subject to carbon loading from aerial dust deposition.The carbon in the GS waste rock and ore likely facilitate deposition of Hg and GOM (c.f.Miller et al, 2013 ;Eckley et al., 2011), .The LTT material is from a non-active tailings impoundment that was partially revegetated at the time of collection, versus the entirely barren surface typical of active tailings.Although the LTT Hg concentration was high (11 g g -1 ), the surface was old, and it is possible this material behaved more as a background substrate.The TCT material was collected from an actively filling tailings impoundment that collected process waste from a variety of ore types, from multiple mine sites, and had the highest Hg concentration of the materials used in this study, which likely explains its tendency to by a strong emitter of both GEM and GOM, a conclusion also made by Eckley et al. (2011a).
An interesting point is that background (2 m and inlet GOM concentrations were very similar during the summer measurements (Fig. 8a), but inlet concentrations were 3x higher than background during the winter measurements (Fig. 8b).The lower winter background GOM (50 pg m -3 ) compared to summer (110 pg m -3 ) may partially explain this, however the distance between the inlet and background sampling heights was only ~ 1 m, which implies a strong vertical gradient in GOM concentration in the greenhouse air.The explanation for this is less well mixed air in the greenhouse bay during the winter, as the greenhouse circulation was shut down at temperatures approaching 13 °C, setting up stratified conditions.The relatively high inlet level GOM concentrations during the winter also caused the higher deposition observed in the chamber blank measurements during this time.

Conclusions
This study presents direct GOM flux measurements using a CEM filter technique and provides an analysis of whether the necessary measurements of small differences in GOM concentration were possible using a DFC method.Measurements of GOM flux were above calculated detection limits in most cases, and both intra-(Tekran vs Pump sample) and inter-(A vs B) system replicate measurements showed very good agreement.After initial trials at 72 h and 48 h, a 24 h sample time was found to be generally sufficient for detecting GOM flux on the mining materials used in this study, some of which were similar to background soils.It would be possible to operate the system at higher flow rates, to decrease sample time and improve the temporal resolution of the flux measurements.However, sample deployment and collection require 20-30 min and to some extent perturbs the system, so ultimately the flux resolution is limited by practical operational constraints.The 24 h measurement at least serves to capture net flux over a full diel cycle without continuous interruptions.
We specifically refer to our measurement as GOM.This is because particulate Hg is not emitted from a surface in the volatile sense, and particle entrainment was negligible at the low flow Here we have demonstrated that GOM can be deposited and emitted from a surface.It is unlikely GOM was produced by oxidation of GEM given short time of air moving through the chamber and the fact that strong oxidants were likely removed as air moved through air handlers into the greenhouse.This is a unique scientific finding.Flux measured from low Hg, non-mineralized cap materials for both GEM and GOM were low positive and negative values oscillating around a net zero flux (Table 1).The observed GEM deposition velocities were typical of non-vegetated surfaces, while GOM deposition occurred at Vd values (0.01 -0.07 cm s -1 ) on the low end of the suggested range (Zhang et al., 2009).The higher Hg concentration leach and tailings material (except LT tails) showed net GEM and GOM emission in the summer, while deposition was observed for both for most materials in the winter.The highest GOM emissions were consistently observed for the highest Hg concentration substrate, TCT, in both summer and winter conditions with fluxes higher in the summer during which time the greenhouse was better mixed.The implications of these results are that GOM fluxes may be directly measured and that contaminated areas such as mine tailings impoundments can act as a direct emission source of GOM compounds to the atmosphere.Note TCT data not graphed, as fluxes were an order of magnitude higher and skew the regression r 2 towards 1.
Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2018-360Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 30 November 2018 c Author(s) 2018.CC BY 4.0 License.Agricultural Experiment Complex.At the onset of the GOM flux experiments described in this paper, substrates were undisturbed for ~3 years, and were completely dry and well compacted/consolidated from previous watering experiments.A circular chamber footprint impression previously existed from prior experiments provided an excellent existing contact for the chamber base.Comparing GEM flux measurements made in this study with the previous measurements, an overall trend of decreasing GEM flux over time was observed.This fits a hypothesis suggested by Eckley et al. (2011b) of a long term reduction in GEM evasion from a mine substrate as time from disturbance increases.
Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2018-360Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 30 November 2018 c Author(s) 2018.CC BY 4.0 License.The GOM concentration in the inlet/outlet sample air was calculated by combining the blankcorrected primary and secondary filters into a total Hg mass per sample line and dividing by the respective sample air volume.The difference in GOM concentration between the inlet and outlet lines provided a ΔCGOM, with this multiplied by sample flow providing the GOM emission rate Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2018-360Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 30 November 2018 c Author(s) 2018.CC BY 4.0 License.Chamber blanks were determined for each Teflon DFC (referred to as Chamber A and Chamber B) by measuring flux over a clean Teflon sheet.Chamber blank emission rates were measured Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2018-360Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 30 November 2018 c Author(s) 2018.CC BY 4.0 License.
Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2018-360Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 30 November 2018 c Author(s) 2018.CC BY 4.0 License.velocitiesgenerated in the flux chambers.Thus GOM concentrations measured at the chamber outlet CEM filters are dominated by GOM.If ambient air GOM concentrations were primarily PBM, at a size fraction large enough to be captured by the CEM filters (i.e.> 0.8 m), it will undeniably be captured at the chamber inlet filters.However, PBM at this size would likely deposit to the substrate surface within the chamber, thus deposition is GOM loading on the Co filters versus the Ci filters.

Figure 2 .
Figure 2. Diagram of one GOM filter-based flux system, deployed in duplicate as Systems A and B..Filter packs indicate the location of the CEM samples.

Figure 3 .
Figure 3. Determination of ΔCGOM detection limit.a) Distribution of Hg mass on unused "blank" CEM filters (median = 68 pg) b) Hypothetical example of statistically detectable GOM flux criteria: shaded boxes represent the maximum uncertainty in concentration, based on 95% confidence interval around the median filter blank (58 -73 pg), Flux1 represents an insufficiently resolvable ΔCGOM in which the 95% confidence intervals around the median blank-corrected Co and Ci values overlap, Flux2 represents the minimum detectable ΔCGOM (13.5 pg m -3 ), and Flux3 represents an obviously resolvable ΔCGOM.

Figure 4 .
Figure 4. Test 72 h replicate measurements of TCL material.a) ΔCGOM values: grey points indicate chamber concentration (Co), blue points indicate inlet air concentration (Ci), and the numeric value of ΔCGOM is shown above b) GOM flux from TCL material in three consecutive 72 h measurements, no chamber blank correction.Sample line labels: AP = Pump A,PB = Pump B, AT = Tekran A, BT = Tekran B.

Figure 5 .
Figure 5. Expanded set of GOM flux measurements for select materials during summer.ΔCGOM was below detection for 5 of 8 LTL measurements, and 2 of 8 TCC measurements, and these values were excluded.

Figure 6 .
Figure 6.Comparison of GOM fluxes measured by Tekran controlled flow sample lines and external pump flow controlled sample lines, for a) System A and b) System B using TCL, TCC, LTL, and TCT summer measurements.

Figure 8 .
Figure 8.Comparison of ambient background GOM concentrations measured at 2 m height in the greenhouse, vs GOM concentrations measured at the chamber inlet, for a) Summer 2015, and b) Winter 2016.Background vs inlet concentrations were comparable during Summer measurements, but inlet concentrations were much higher relative to background in the winter.