Dimethylmercury in natural waters—analytical and experimental considerations

Mono‐ and dimethylmercury (MMHg and DMHg, respectively) are the two primary organic forms of mercury (Hg) found in natural waters. While experimental approaches to characterize the environmental behavior of MMHg and inorganic forms of Hg are widely used today, few laboratories conduct experimental studies entailing the use of DMHg. In this paper, we have evaluated and developed different analytical and experimental approaches to quantify and use DMHg in laboratory studies. We demonstrate that DMHg can be analyzed from samples where MMHg is derivatized using sodium tetraethyl borate and where the matrix effects of dissolved sulfide are masked using copper sulfate. Tests, where the calibration curves of MMHg and DMHg were used, showed that MMHg may be used to calibrate for DMHg. For the pre‐concentration of DMHg, both traps filled with Tenax® TA and Bond Elut ENV were found suitable. We observed good recoveries of DMHg added to different types of natural waters or purified water containing aquarium salt, sodium chloride and dissolved sulfide, iron sulfide, and cadmium sulfide at DMHg : sulfide molar ratios > 10−6. In addition to evaluating these analytical aspects, we present suitable subsampling techniques for DMHg‐containing solutions, the recovery of DMHg when filtering DMHg through different types of filters, and experimental data on the long‐term stability of DMHg added to different types of waters and stored at different temperatures. Finally, we present and discuss a new synthetization protocol for preparing aqueous solutions containing DMHg free of organic solvents and where handling DMHg in a pure form is prevented.

Mercury (Hg) is released from the bedrock through natural and anthropogenic processes and bioaccumulates in aquatic systems to levels that concern human and wildlife health (e.g., AMAP 2021).The biogeochemical pathways of Hg, linking the Hg released from the bedrock to the pool of Hg in aquatic food webs, are complex and involve several chemical forms and species of Hg.Inorganic divalent Hg (Hg II ) and elemental Hg (Hg 0 ) are the two primary forms of Hg released from natural and anthropogenic sources and the main forms of Hg that cycle in the abiotic compartments (air, water, sediment, and soil).In 1969, it was demonstrated that Hg II can be methylated in natural environments to mono-and dimethylmercury (hereon referred to as MMHg and DMHg [Jensen and Jernelov 1969]).Two decades later, these organic forms of Hg were also detected for the first time in marine waters (Mason and Fitzgerald 1990).Today, it is well documented that MMHg is the primary form of Hg that accumulates to concentrations of concern in aquatic food webs (e.g., AMAP 2021).However, the role DMHg plays in Hg's biogeochemical cycling is largely unknown due to a lack of mechanistic studies.Several reasons explain the current lack of experimental studies entailing the use of DMHg.First of all, most of the studies conducted in the past 50 years have focused on understanding Hg cycling in freshwater systems, of which only a handful of studies have reported significant concentrations of DMHg (Bloom and Effler 1990;Alanoca et al. 2016;Wang et al. 2019).Second, handling and conducting laboratory studies with DMHg comes with a range of challenges, one major challenge being its extreme toxicity.In 1997, accidental exposure to concentrated DMHg, through a few drops spilled on her glove, resulted in the death of a professor at Dartmouth College (Avid et al. 1998).Additional challenges in dealing with this toxic compound come from its high volatility (Henry's law constant of 0.145 at 0 C [Talmi and Mesmer 1975]) and its ability to adsorb to, and diffuse through, different materials (including commonly used laboratory gloves and Teflon) (Parker and Bloom 2005).Furthermore, DMHg decomposes through acidolysis, which means that both MMHg and DMHg will be detected as MMHg in samples preserved with acid (Black et al. 2009).
We have recently conducted laboratory studies with DMHg to study its stability in natural waters (unpublished data) and to explore potential degradation pathways (West et al. 2020(West et al. , 2022)).To facilitate these experiments, we have also evaluated different analytical and experimental approaches and developed a new synthetization protocol for producing aqueous DMHg solutions free from organic solvents where dealing with concentrated DMHg solutions is avoided.The analytical aspects presented in this paper include an evaluation of the use of MMHg for external calibration of DMHg, a comparison of the use of Bond Elut ENV with Tenax ® TA as the solid adsorbent for the pre-concentration of both MMHg and DMHg, and finally, potential effects of matrices and interfering compounds.To assess experimental approaches, we evaluated different transferring techniques for subsampling of DMHg-containing water samples (gas-tight syringes and automated pipettes), sample filtration techniques, and tested for adsorption of DMHg to flask walls.Finally, we tested the long-term stability of DMHg under dark conditions in different types of waters.

Reagents and standards
DMHg used in the tests conducted at the Department of Environmental Science, Stockholm University (Stockholm, Sweden), was synthesized in tetrahydrofuran (THF) according to a previously described method (Snell et al. 2000;West et al. 2020).For DMHg synthetization without THF, a 1000 mg L À1 Hg II in 12% vol/vol HNO 3 (AAS standard; Sigma-Aldrich) and methylcobalamin (MeB 12 ; 98%; purchased from Ambeed) were used.The method for the synthetization of a pure DMHg standard involving MeB 12 and preferential diffusion of DMHg into a larger volume of water has been previously described (West et al. 2022).Sodium tetraethyl borate (NaTEB; 1% wt/vol in 2% wt/vol KOH), buffers, MMHg calibration stock solutions, Na 2 S solutions, disordered mackinawite (FeS(s)), cadmium sulfide CdS(s), copper sulfate (CuSO 4 ) and sodium oxalate (Na 2 C 2 O 4 ) were all prepared as previously described in West et al. (2020).For NaCl solutions, ACS reagent NaCl (Merck) was used.The aquarium salt solution (Instant Ocean ® ; 29.2‰) was the same prepared and used in previously published experiments (West et al. 2022).Acids and bases used in experimental procedures were of trace metal grade purity.
For the tests conducted in the laboratory at Tekran Instruments Corporation, a DMHg standard was obtained from the chemistry department of Trent University in 2013 (concentration of 348 mg mL À1 ).
This standard was diluted by pipetting 100 μL of the stock solution to 100 mL of methanol and another 100 μL to 100 mL of isopropanol.Both bottles with the diluted stocks were capped, double-bagged, and transported in a mobile mini fridge from Peterborough, Canada.Upon arrival in the Tekran laboratory in Toronto, Canada, they were immediately put in a À20 C freezer.Before further tests, both standards were then diluted (in methanol and isopropanol, respectively).To determine the concentrations of the standards received, two replicates of each standard (with a concentration of 10 ng L À1 Hg) were reacted with 1 and 2% BrCl for 3 h at room temperature.Then, total mercury was analyzed by cold vapor atomic fluorescence spectroscopy (CVAFS) using the Tekran 2600 system.The Tekran 2600 analyzer was calibrated using the HgCl 2 solution Nist 3177 Standard Reference Material.The recoveries of the total mercury in both 1% and 2% BrCl standards were 100%.No difference was found between the dilutions of the standards in methanol or isopropanol.The stability of the standard over the years since it was obtained was also confirmed in tests where it was analyzed together with a methylmercury (II) standard solution (1000 mg L À1 in H 2 O; Alfa Aesar ® , product ID 33553), diluted to the same concentrations and analyzed together in the same vials on the Tekran ® Model 2700 Automated Methylmercury Analyzer System (hereafter referred to as Tekran 2700) (Supporting Information Fig. S1).Thus, our tests confirmed that the Hg concentration in the DMHg standard did not change over the years and agreed with the value found at Trent University.

Collection of natural waters
Arctic waters were collected from 5 m water depths on different days in the autumn of 2018 during the SWEDARCTIC 2018 expedition per previously described procedures (West et al. 2022).Baltic waters were sampled onboard R/V Electra from Landsort Deep (58 60 0 N, 18 23 0 E) on 16 March 2020, from 35, 70, and 250 m depth, using a Niskin Rosette.Waters from the Mediterranean were sampled onboard R/V Antédon II outside Marseille (43 18 0 N, 5 42 0 E) on 06 October 2020, using tracemetal clean Go-Flo bottles.Deep waters were collected from 84 m depth, and shallow waters were collected from the depths of 5.6 and 11.5 m and mixed.All freshwater samples were collected in Sweden from different aquatic systems.Water from the pond Laduviken (59 36 0 N, 18 08 0 E) was sampled on 29 June 2020.Samples were taken from the shore of Lake Grossjön (61 48 0 N, 16 60 0 E) and the river Voxnan (61 35 0 N, 16 06 0 E) on the 17-18 October 2020.Finally, stream water from a forested area in Täby (59 27 0 N, 18 46 0 E) was sampled on 04 July 2021.

Experiments
Experiments testing the linear range and response factor of DMHg and MMHg, and the solid sorbent used to preconcentrate the analytes were conducted at Tekran Instruments Corporation.All other experiments described below were conducted at the Department of Environmental Science, Stockholm University.

Evaluation of the analytical procedures
All interference tests were executed in a similar manner, where a fixed amount of DMHg was added to treated samples and controls.DMHg was added within a short time frame to avoid degassing of the DMHg stock solution during sample preparation.Samples were thereafter analyzed within the same day on the Tekran 2700.
The effect of NaTEB and acetate buffer on DMHg recovery was tested in purified water (Milli-Q ® ) and Baltic water from the depths of 35 and 250 m, analyzed with or without the addition of 10 mg L À1 of NaTEB and 15 mmol L À1 acetate buffer.In this test, measured concentrations of DMHg varied between 8.41 and 9.54 pmol L À1 DMHg.
To test the linear range and response factor of DMHg and MMHg, respectively, calibration standards containing both DMHg and MMHg and only DMHg were prepared.For the calibration standards containing both DMHg and MMHg, DMHg was injected using a gas-tight syringe into a closed vial already containing the MMHg standard and derivatization agents (acetate buffer and NaTEB).The test was done both for smaller (40 mL vials, 30 mL solution) and larger volumes (250 mL flasks, 180 mL solution).The calibration standards ranged from 0.02 to 4.00 ng L À1 (corresponding to 0.6-120 pg Hg) and from 0.01 to 0.40 ng L À1 (corresponding to 1.8-72 pg Hg) for the smaller and larger volume, respectively.The gas chromatograph thermal program was adjusted to allow a better separation of Hg 0 , DMHg, and MMHg, and for the larger volume, the purging time was increased to 300 s.Purging flow was set to 80 mL min À1 for the 40 mL vials.For the 250 mL flasks, flows up to 250 mL min À1 were evaluated, with 120 mL min À1 selected as a compromise for optimizing the response for the 180 mL liquid volume while not overburdening the hardware setup, originally built for smaller volumes.
To evaluate the solid sorbent used to pre-concentrate the analytes, we compared traps with Tenax TA (Buchem B.V.) and Bond Elut ENV (Agilent Technologies), hereafter referred to as Tenax and Bond Elut.In the first test, the internal trap in the Tekran 2700 (containing Tenax) was emptied and packed with Bond Elut.After adjustment of the desorbing temperature, calibration curves of MMHg and DMHg were prepared and compared to those obtained using the standard sorbent (Tenax).In the second test, large volume calibration standards (180 mL in 250 mL flasks) were used and purged for 5 and 10 min, respectively.The analytes were preconcentrated using external traps (borosilicate glass cartridges with an inner diameter of 6 mm packed with 0.15 g Bond Elut or Tenax).The traps were kept at a temperature of 30 C during the loading using a Tekran 2030 cartridge heater.A second trap was installed (maintained at a temperature of 30 C during collection) after the first trap to collect any potential breakthrough.The borosilicate glass cartridges were then desorbed in a custom-built external desorber and transferred to the Tekran 2700 internal Tenax trap.Some loaded Bond Elut and Tenax cartridges (sealed and double-bagged) were left at room temperature (and some Bond Elut traps were kept at 4 C) over the weekend and desorbed on the following Monday.
The recovery of DMHg in natural waters was tested in three experiments where the area of the DMHg peak measured in various water types was compared to the area of the DMHg peak measured in purified water.The measured concentration of DMHg in purified water samples varied between 3.82-9.39pmol L À1 among the tests.Experiments evaluating the effect of salinity were performed with NaCl and aquarium salt (Instant Ocean ® ), which were dissolved in purified water to salinities between 8.75 and 35‰ and compared to samples only containing purified water.DMHg concentrations in purified water were constrained between 2.9-3.1 pmol L À1 .
Multiple tests were conducted investigating any interferences by varying amounts of sulfide (dissolved as H 2 S/HS À , and solid as FeS(s) and CdS(s)) on DMHg recovery.In total, the effect of Na 2 S was investigated in 12 experiments, FeS(s) in 9 experiments, and CdS(s) in 1 experiment.Experiments were conducted at three different temperature settings (room temperature [ca.20 C], 40 C, and 60 C) and three controlled pH ranges using citrate, potassium phosphate, and carbonate buffers (pH 5, 7.5, and pH 9), as well as under non-buffered conditions.DMHg sample concentrations varied between 0.0039 and 25.0 nmol L À1 .The effect of masking reagents used for MMHg recovery at high dissolved sulfide concentrations (Yang et al. 2009) was tested in one experiment where DMHg was added to samples with HS À of increasing concentrations (pH 9, controlled by 26 mmol L À1 carbonate buffer).Samples were thereafter subsampled into vials spiked with varying amounts of CuSO 4 , sodium oxalate, and NaTEB.

Evaluation of the experimental procedures
Tests of reproducibility of DMHg transfer using gas-tight syringes (Hamilton ® ) or mechanical pipettes were repeated, where DMHg of the same stock solution was transferred into several subsamples (n = 3-5).These tests were performed with different sizes and volumes of syringes (10-250 μL) and pipettes (20-200 and 1000-10,000 μL).All transfers were done with the needle or pipette tip end below the surface of the receiving solution.Syringes were washed in methanol between transfers.All stock solutions were prepared in purified water and analyzed with 15 mmol L À1 acetate buffer and 10 mg L À1 NaTEB.Analyzed DMHg concentrations ranged between 0.44-35.9pmol L À1 .
Two tests were carried out, where the suitability of filters for filtration of DMHg containing aqueous solutions was investigated.In the first test, syringes were filled with purified water, Arctic Ocean water, and stream water spiked with DMHg to a concentration of about 600 pmol L À1 .The spiked waters were thereafter pushed through filters with polyester sulfone (PES) membranes (Filtropur S; Sarstedt) and added with a syringe needle under the water level to 40 mL autosampler vials.The recovery was compared to controls where no filters were used.The second test tested Whatman™ glass fiber filters (0.7 μm, 47 mm) and polycarbonate Nuclepore™ filters (0.2 μm; Whatman™).Before use, the glass filters had been pre-combusted, and the Nuclepore™ filters were washed overnight in 10.5% vol/vol each of HNO 3 and HCl.DMHg was added to a 2 L bottle to create a low-concentration stock where the large volume would inhibit significant degassing during the experiment.Degassing during the filtration step was further avoided using a three-way valve with Teflon tubing connected to (1) the inlet tube, submerged in the 2 L container, (2) a 50 mL syringe used to estimate sample volumes and pump DMHg stock solution through the filter, and (3) tubing attached to a filter cassette.Short pieces of silicon tubing were used to connect the Teflon tubing to the valve.All tubing and glassware were washed with HCl overnight before use.The DMHg concentration of the stock solution (around 2 pmol L À1 ) was monitored throughout the experiment.Samples of both experiments were spiked with 15 mmol L À1 acetate buffer, and with 10 mg L À1 NaTEB added before analysis.The amount of sample added to each vial was determined by weight.Sample concentrations were compared to stock solutions.The setup was also tested without a filter in the cassette to assure that no losses of DMHg occurred by, for example, absorption or diffusion of DMHg through the tubing or the filter cassette (with no filter) used.
Long-term stability experiments were performed with Arctic surface waters, Mediterranean waters from two depths, Baltic waters of three depths, and freshwater from Lake Grossjön and the Voxnan River.In these experiments, DMHg was spiked into water samples in 40 mL amber vials to a concentration of approximately 10 pmol L À1 .The total sample volume equaled 39 mL.Samples were incubated for up to 3 weeks at 4 C and room temperature.At the time of analysis, 9 mL was removed from the sample, which was thereafter prepared with 15 mmol L À1 acetate buffer and 10 mg L À1 NaTEB (purified water, Baltic water at 35 and 250 m depth, Mediterranean waters, freshwaters) or analyzed without the addition of reagents for direct ethylation (purified water, Arctic surface water, and Baltic water at 72 m depth).
Adsorption of DMHg on glass surfaces was tested by adding 0.5 mm disruptor beads (lime glass; Scientific Industries) to 40 mL amber glass vials (borosilicate, Supelco ® ).Beads were assumed to be perfectly spherical, and the volume was calculated after the weight of water displaced per weight unit of beads.One gram of beads represented an estimated surface area of 45 cm 2 .Beads were added to vials in quantities of 0, 2, 4, 8, or 16 g, thereby increasing the total surface area from ca 80 cm 2 to around 800 cm 2 .All treatments were prepared in triplicates.DMHg was added to vials and incubated for 5.5 h.Upon sampling, 5 mL was taken from each vial and added under the water surface to 25 mL purified water with 15 mmol L À1 acetate buffer.NaTEB (10 mg L À1 ) was added before analysis.Measured DMHg concentrations were corrected to adjust for diluting DMHg to varying degrees in treatments with varying volumes of beads.The experiment was duplicated with acid-washed beads (0-8 g) after it was discovered that non-washed beads introduced low concentrations of Hg II .In this experiment, the concentrations of DMHg were adjusted to be the same for all treatments.Sample DMHg concentrations were in the range of 10-100 pmol L À1 .

Analytical methods
Methylated Hg species (DMHg and MMHg) were analyzed through direct ethylation (of MMHg) on the Tekran 2700 and calibrated with external calibration prepared from certified MMHg standards (1000 mg L À1 ; Alfa Aesar ® ).Total Hg was analyzed on a Tekran 2600 Total Hg analyzer following standard protocols (US EPA 2002).pH was measured with a pH meter (Orion Star™ A214, Orion™ 8102SC ROSS ® electrode) and occasionally monitored with pH sticks.Salinity was measured with a handheld salinity meter (HCO 304, VWR ® ).

Statistical methods
All statistical tests were conducted using JMP Pro (version 15.0.0).The data distribution was assessed using the Shapiro-Wilk normality test and visually inspecting density plots and Q-Q plots.All tested parameters were deemed to be normally distributed.Statistically significant differences were determined using a one-way analysis of variance following Tukey's pairwise post hoc test (statistical differences indicated by p-value < 0.05).Bivariate relationships were tested using simple linear regression analysis, and the null hypothesis (that the slope coefficient is 0) was rejected for p < 0.05.

Analytical considerations
Analytical approaches commonly used for the quantification of MMHg and DMHg in natural waters typically start with the extraction and pre-concentration of MMHg and DMHg from the collected water.Such methods were also utilized in this study.As DMHg is a dissolved gas, it can be purged out from the sample (using either N 2 (g) or Ar(g)) and then preconcentrated on a solid sorbent (e.g., Carbotrap(R) B or Tenax).In contrast, MMHg is an ionic form of Hg that typically is ethylated (derivatized, e.g., using NaTEB) to then be purged and trapped.For some matrices, direct ethylation is feasible.In other cases, including saline waters and waters with high concentrations of reduced sulfur, interferences may affect MMHg recovery.In such cases, sample pretreatments are required (Yang et al. 2009;Bloom 2011;Munson et al. 2014;Mansfield and Black 2015).After both forms have been preconcentrated, they are thermally desorbed and separated using gas chromatography.Hg can then be detected with analytical tools such as CVAFS after conversion of the methylated Hg species to Hg 0 by pyrolysis or with inductively coupled plasma mass spectrometry.
For some applications, it may be beneficial to combine the direct ethylation and analysis of MMHg with the analysis of DMHg.This approach was, for example, valuable in our photodecomposition experiments (West et al. 2022), where minimizing the volumes at each subsampling was desirable (to reduce the amount of headspace created).Also, the approach significantly reduced the time required to run all the analyses.To assure that we had no artificial degradation of DMHg during the MMHg derivatization step, we quantified the recovery of DMHg in water with and without the addition of acetate buffer and NaTEB to total concentrations of 15 mmol L À1 and 10 mg L À1 , respectively (standard protocol for MMHg derivatization on the Tekran 2700 Methylmercury analyzer system).This test was executed for purified water and Baltic seawater collected at a depth of 35 and 250 m.No loss of DMHg recovery was observed in any of these tested waters (p > 0.05), demonstrating that it is possible to combine direct ethylation for MMHg analysis and DMHg analysis (Supporting Information Fig. S2).Combining analysis for MMHg and DMHg is, however, not feasible for samples where extraction of MMHg is required prior to the derivatization step by, for example, distillation or acidification (US EPA 1998; Bowman and Hammerschmidt 2011;Munson et al. 2014) nor when preservation of MMHg is needed (Black et al. 2009).For samples with high sulfide content, a masking approach has been suggested to enable direct derivatization of MMHg (Yang et al. 2009).This masking approach entails the use of copper sulfate (CuSO 4 ) to form insoluble copper-sulfide complexes, followed by the addition of sodium oxalate (Na 2 C 2 O 4 ) to chelate any Cu 2+ remaining in solution.To test if this masking approach can be combined with the analysis of DMHg, we performed recovery tests also with 5-25 μmol L À1 CuSO 4 and 10-50 μmol L À1 sodium oxalate (Supporting Information Table S1).In one sample group, reagents were added to the vial after the sample, and a DMHg recovery of 89 AE 0.8% was observed.In the remaining three sets of samples, where DMHg was added after the reagents, the recovery ranged from 101 AE 2% to 119 AE 3%, suggesting the appropriateness of combined analysis of DMHg and MMHg in samples where masking agents are added to increase the recovery of MMHg.The lower recovery for the first group was possibly due to the degassing of DMHg during the addition of reagents.
Due to the high toxicity of DMHg, MMHg calibration curves are commonly used to quantify DMHg in natural waters (Bowman and Hammerschmidt 2011;Bowman et al. 2016).We evaluated the validity of this approach by preparing and comparing calibration curves for both species (ranging from 0.02 to 4.00 ng L À1 ).This test was conducted with both smaller (40 mL flasks containing 30 mL of solution) and larger volumes of standards (250 mL flasks containing 180 mL of solutions).In both cases (Fig. 1), we observed a linear response with increasing concentration (coefficient of determination R 2 > 0.997) and that the calibration curves for both methylated species were close to identical.For the smaller volume (Fig. 1A), the calibration curve for MMHg was slightly below the one for DMHg (error of 6.6%).This discrepancy could be explained by the use of aged NaTEB for the ethylation of MMHg.For the larger volume (where freshly prepared NaTEB was used; Fig. 1B), the difference between the two calibration curves was only 3.6%.In both these tests, the MMHg and DMHg were combined and analyzed from the same vials.A similar agreement between the response of both species was also seen in the repeated analysis of the DMHg standard together with a 33553 methylmercury(II) standard solution in the years after obtaining the standards (Supporting Information Fig. S1).As linear responses were observed for both DMHg and MMHg (through the addition of buffer and ethylation agents), this further confirms our conclusion above that direct analysis of MMHg can be combined with the analysis of DMHg.
A few solid absorbents have been used to preconcentrate the Hg for speciation analysis (i.e., separation of Hg 0 , Hg II , MMHg, and DMHg).Baya et al. (2013) have previously tested the efficiency of Tenax, Carbotrap (R) B, and Bond Elut to trap DMHg and ethylated MMHg from the gaseous phase.According to their study, the best (simultaneous) recovery of both species was obtained with the Bond Elut sorbent (recovery of MMHg: 100 AE 8.1% for Tenax, 61 AE 32.5% for Carbotrap (R) B and 98 AE 9.2% for Bond Elut; recovery of DMHg; 64 AE 17.3% for Tenax, 100 AE 0.3% for Carbotrap (R) B, and 95 AE 8.1% for Bond Elut).Here, we conducted tests to compare the Bond Elut with the more commonly used Tenax adsorbent (which is also standard in the MeHg analyzer used).In the first test, calibration standards were compared for when the internal trap in the MeHg analyzer was filled with Tenax and Bond Elut, respectively.For both sorbents, a linear response was observed in the tested concentration range (up to 4.0 ng L À1 ; Supporting Information Fig. S3).The response of the Bond Elut was, however, about half that of Tenax.We attribute this lower response to the fact that the Bond Elut is a very fine powder and that the 10 cm long packing in the metallic tube where the Tenax is originally placed caused significant back pressure.Tests were also conducted with external traps, corroborating the good agreement between DMHg and MMHg calibration curves.When using the external traps, the recovery (slope of the calibration curve) was 16-20% higher.These results align with earlier experiments (data not shown), showing a higher recovery when external traps are used.We attribute this effect to differences in the size and materials of the cartridges and different flows used for species transportation during the desorption (for the internal trap, ca.20 mL min À1 ; for the external traps, 80-120 mL min À1 ).The recovery of cartridges stored at room temperature and over the weekend matched those analyzed the same day.A lower response was found for the Bond Elut stored at 4 C, probably due to slower heating of the traps during the desorption as they may not have been fully equilibrated to room temperature before desorbing.No MMHg nor DMHg was detected on cartridges placed downstream of the analytical traps during the pre-concentration, demonstrating that there was no breakthrough on the analytical traps.To conclude, MMHg and DMHg responded in a very similar way and were successfully separated despite the different experimental conditions tested (sample prepared in 40 mL vials or 250 mL bottles, at loading times ranging up to 10 min, Tenax vs. Bond Elut, internal loading vs. external loading and desorption).The fact that we, in contrast to an earlier study (Baya et al. 2013), did not observe any breakthrough of DMHg on the Tenax trap can be explained by the relatively short loading times tested in our study (in the earlier study, 100-200 L of air was loaded at a flow rate of between 1 and 2 L min À1 ).Care should be taken for applications with longer loading times (> 10 min) to ensure that no breakthrough occurs.
Recovery of DMHg among contrasting waters was tested using surface water collected from the Arctic Ocean, water collected at different depths in the Baltic Sea, and water collected from a freshwater lake.These samples were selected to represent the range of organic matter content and salinities typical along the land-ocean aquatic continuum (West et al. 2022).Recoveries were calculated by comparing the concentration of DMHg detected in purified water.The recovery of DMHg was close to or just above 100% for all tested waters (range of 102 AE 1.6% to 118 AE 3.8%; Supporting Information Fig. S4).The highest recovery (118 AE 3.8%, p < 0.05) was observed for water collected from the Arctic Ocean (and could not be explained by ambient DMHg remaining in the sample, as the total ambient Hg concentration of the Arctic waters was only 2.2% of the average DMHg concentration added for these tests).Additional tests with aquarium salt and NaCl solutions confirmed good recoveries of DMHg purged out at salinities ranging up to 35‰ (Supporting Information Fig. S5, p > 0.05 when the detected DMHg concentrations were tested as a function of the salinity).In conclusion, our recovery tests with natural waters and salt solutions suggest that neither salinity nor dissolved organic carbon (DOC) content affected the recovery of DMHg, and the cause for the slightly higher recovery in the Arctic water is unknown.
The potential interference of reduced sulfur on DMHg quantification was assessed using data from several experiments where DMHg was quantified at the start of the incubation experiments.The recoveries were calculated based on the concentrations measured in purified water in the same experiment.Reduced sulfur species tested included dissolved sulfide (Na 2 S), the metastable iron sulfide mineral Mackinawite (FeS (s)), and cadmium sulfide (CdS(s)).In addition, the dataset includes the addition of DMHg in sample matrix preheated to 40 C and 60 C for at least 1 h and samples kept at room temperature, and solutions/slurries that either were unbuffered or buffered to a pH of about 5 or 9.At DMHg : sulfide molar ratios > 10 À6 , recoveries close to 100% were achieved for all the tested forms of reduced sulfur (Fig. 2).At DMHg : sulfide molar ratios of 10 À6 and below, recoveries were low (well below 60%) for a number of samples.The trend of decreasing recoveries at lower DMHg : S molar ratios was most apparent for FeS(s).This trend is in line with previous measurements where we demonstrated that DMHg is more reactive toward the reduced sulfur sites on the FeS(s) mineral than toward dissolved sulfide (West et al. 2020).For CdS(s), the tested range was above 10 À6 , and good recoveries were observed.In our previous work, we noted that while DMHg degraded in the presence of FeS(s) and dissolved sulfide, no degradation was observed in the presence of CdS(s), suggesting that DMHg was less reactive toward the CdS(s) mineral (West et al. 2020).Recovery tests at lower (10 À6 and below) DMHg : S molar ratios are still warranted for future studies if DMHg would be quantified from such solutions.No apparent effect of temperature (preheating of the samples prior to the addition of DMHg) nor pH was observed (Fig. 2; Supporting Information Fig. S6).

Experimental approaches
Tests evaluating the suitability of transferring techniques of DMHg-containing solutions were performed with gas-tight syringes (designed for volumes of 10, 50, and 250 μL) and automated pipettes (designed for volumes of 200 and 10,000 μL; Supporting Information Table S2).All transfers were made by injecting the stock solution under the water surface of the receiving solution to avoid losses.For the 50 and 250 μL syringes and the automated pipettes tested, the variability among replicates was low (relative standard deviation [RSD] ranging from 0.85% to 5.2%).Higher variability was observed for the 10 μL syringe (RSD of 20% and 25%, respectively).We thus conclude that even though DMHg is volatile, subsampling can be successfully made using pipettes, which may be preferred as the transfer is less time-consuming and does not require cleaning between samples (as needed for the gas-tight syringes).Aligned with the observations made at Tekran (described above), these tests also confirm that injection of the parent solution under the surface of the receiving solution (and thereafter quickly capping the container) yields good results.It should be noted that in most of these tests, we transferred quantities corresponding to the maximum volume for each pipette.Tests where 28.6 and 50 μL of solution were transferred using a 20-200 μL pipette did, however, also result in low variability (RSD of 0.87% and 1.9%, respectively).
Losses of DMHg were tested using syringe filters (PES) and membrane filters (glass fiber filters and Nuclepore™ filters).While no loss of DMHg was observed while filtering the DMHg dissolved in purified water through the glass filter (p > 0.05), the same procedure using Nuclepore™ filters resulted in almost 40% loss of DMHg ( p < 0.05; Supporting Information Fig. S7).At the same time, the PES syringe filter tested resulted in a nearly complete loss of DMHg (recoveries of DMHg were < 3%, p < 0.05).The results for the syringe filter were consistent among the water types tested (purified water, Arctic Ocean surface water, and water from a forest stream; Supporting Information Fig. S7).While the actual loss of DMHg during filtration may vary with factors including pore or mesh size, filter pretreatment, brands, and volumes filtered, these results demonstrate that many types of filters may be unsuitable for studies on DMHg.The US EPA method 1631 (appendix A) also discourages filtration when quantifying DMHg in water (US EPA 1998).
The long-term stability of DMHg when dissolved in contrasting natural waters (ranging from high saline and low DOC marine waters to low saline and high DOC rich freshwaters) was tested both at 4 C and 20 C (Supporting Information Fig. S8).At 4 C, no loss of DMHg was observed in the two marine water collected in the Mediterranean Sea (incubated for 20 d, p > 0.05 for surface water tested and p < 0.05 showing a positive correlation for the deep water tested) nor from the water from Lake Grossjön ( p > 0.05), while a slight loss was observed for the freshwater (incubated for 15 d) collected from the Voxnan River ( p < 0.05).For samples stored at 20 C, significant losses of DMHg (p < 0.05) were observed in six of the nine waters tested (statistical data provided in Supporting Information Fig. S8), with recoveries ranging from 50 AE 21% to 88 AE 4% at the end of the incubations.No clear trend was observed between the fresh and marine waters tested for incubations at higher temperature.Potential processes contributing to these losses include dark abiotic degradation processes, evasion of DMHg through the septa, and adsorption of DMHg to the vial walls.To test the degree of adsorption of DMHg to the walls, DMHg was incubated in vials containing different amounts of round glass beads.No loss of DMHg was, however, observed as the glass surface area was increased from ca. 80 cm 2 (surface of the vial walls alone) to 800 cm 2 (surface of the vial walls and 16 g of glass beads) (Supporting Information Fig. S9).This test suggests that the loss of DMHg through wall adsorption is limited (p > 0.05 when detected DMHg concentrations were tested as a function of glass bead surface area).
Protocols for synthesizing low levels of DMHg are important to avoid handling pure DMHg solutions.Two commonly used approaches for MMHg synthetization can also be used for DMHg production.First, DMHg may be synthesized by the Grignard reagent magnesium tetramethyl chloride.The yield of DMHg in this approach is high (up to 100%) but results in stock solutions containing organic solvents (toluene and THF).For some applications, such as photodegradation studies, trace amounts of these solvents may cause experimental artifacts.We thus developed a synthetization protocol using the alternative methylating agent methylcobalamin (MeB 12 ).To separate the DMHg from the MeB 12 and not fully methylated Hg forms (Hg II and MMHg), the reaction was carried out in a smaller reaction vial, and the DMHg was then allowed to diffuse out from the reaction solution to a second solution.This separation was achieved by placing the smaller reaction vial into a larger closed vial containing the receiving solution.The latter could thereafter be transferred with a syringe and added to purified water and buffer in another vial.The DMHg concentrations of the resulting stock solutions (ca.30 mL in volume), quantified after dilution on the Tekran 2700 methylmercury analyzer, ranged between ca.55 and 470 nmol L À1 from four separate synthetizations.With dilution during transfer estimated to 20%, this corresponded to reaction yields of ca.5-40%.Although the yields were not 100%, the new approach resulted in a solution of DMHg free from other chemical forms of Hg and organic solvents.Method development revealed two findings that should be considered if the method is to be adapted.First, even with the ratio of Hg II : MeB 12 unchanged, our tests showed that the absolute concentrations of reactant and methyl donor are of great importance for successful DMHg synthetization.We recommend using Hg II in concentrations in the same range as those initially used by Jimenez-Moreno (2013), that is, around 3.5 mg L À1 Hg (most syntheses were performed with the double concentration, 7 mg L À1 ).Second, the chemical speciation of Hg II also appears to be critical for a successful high-yield synthetization.This conclusion was based on failed attempts to reproduce the results of using the 1000 mg L À1 AAS standard (Hg II in 12% vol/vol HNO 3 , Sigma-Aldrich) with isotopically labeled 198 Hg II and 204 Hg II solutions (CortecNet) acidified with HCl.These attempts resulted in a lower yield and considerate fraction of MMHg (ca.30% of the DMHg content), as well as some Hg II in the receiving solution.While the reason for this is unclear, the speciation of Hg II to NO 3 À contra Cl À possibly affects the kinetics of ligand substitution and, thus, methylation yields.In addition, as volatility is the attribute separating DMHg from the reaction mixture, the co-occurrence of ionic Hg II and MMHg with DMHg in the receiving solution could possibly be explained by a lowered stability of the formed DMHg and, thus, subsequent breakdown to MMHg and Hg II when synthesizing from isotopically labeled Hg II acidified with HCl.Only when the synthetization using 204 Hg II was repeated with HNO 3 in the same proportions as for the AAS standard synthetization was the synthetization successful, with a high yield and < 1% MMHg and Hg II in the final solution.We observed the resulting DMHg stock solutions to be stable over the period of months without substantial formation of MMHg or Hg II .However, the stock solutions gradually weakened over time, likely due to the evasion of DMHg during storage and use.

Discussion
Further experimental studies are needed to address the unknown role of DMHg in the biogeochemical cycle of Hg.Laboratory experiments entailing the use of DMHg are, however, challenging to conduct (for several reasons, including its high toxicity and volatility).Here, we present an evaluation of different analytical and experimental approaches together with a new protocol for preparing DMHg solutions free of organic solvents and where handling DMHg in a pure form is prevented.
When studying transformation reactions involving DMHg, it is likely also desirable to quantify the levels of MMHg.Our results that DMHg can be analyzed from samples where MMHg is derivatized using NaTEB, and the matrix effects of dissolved sulfide are masked using copper sulfate, means that both the cost and time of analysis can be reduced.This, in combination with the use of isotopically enriched Hg tracers, has previously allowed us to follow the loss of DMHg and the formation of MMHg (from the degradation of DMHg in the presence of reduced sulfur or light) from the same set of subsamples (West et al. 2020(West et al. , 2022)).
By demonstrating similar responses (slope of the calibration curves) of MMHg and DMHg, our study supports the use of MMHg for the quantification of DMHg under the conditions applied in this work.This means that the risks of handling DMHg-containing solutions in some cases can be avoided.We also demonstrated that this was true both when using Tenax and Bond Elut as the solid adsorbent in the preconcentration step (for loading times of up to 10 min).It remains to be investigated whether MMHg standards also are suitable when other analytical protocols are used (e.g., with longer loading times in the pre-concentration step).For example, some protocols (including some for shipboard analysis of DMHg in waters) call for larger volumes of samples and longer loading (purging) times (US EPA 1998).
Our evaluation of DMHg recoveries added to natural waters and laboratory solutions demonstrates that good recoveries can be expected from a wide range of different water matrices, including at salinities up to 35‰, in the presence of DOM, and in the presence of dissolved or particulate sulfide at DMHg : sulfide molar ratio exceeding 10 À6 .Therefore, matrixspecific calibration standards are not necessarily needed, and recovery differences among different types of water can be assumed to be neglectable.However, filtration may be needed for solutions containing a lot of particles.Online filtration is also included in some protocols for analyzing DMHg in seawater (Cutter et al. 2010).In the US EPA method 1630, it is stated that (for DMHg analysis) "Samples must not be filtered prior to analysis, or (CH 3 ) 2 Hg will be lost to the air."Although the risk of DMHg losses to the air during online filtration likely could be prevented by avoiding the formation of air pockets in the system, our data shows that adsorption of DMHg to the filter material itself may be an issue and thus needs to be considered.This is particularly the case for filters with PES membranes (e.g., found in AcroPak™ filters [Pall Corporation] commonly applied in online filtration [Cutter et al. 2010;Lamborg et al. 2012]).
Finally, we present and discuss a new synthetization protocol for preparing aqueous solutions containing DMHg free of organic solvents and where handling DMHg in a pure form is prevented.This protocol has already been successfully used to study the photodecomposition of DMHg (West et al. 2022).As the DMHg is dissolved in water, the solution is recommended to be stored cold (but not frozen).It is also important to quantify the DMHg content in the standard before use, as aqueous DMHg solutions are less stable than standards prepared using an organic solvent.

Comments and recommendations
Due to the extreme toxicity of DMHg, utmost care must be taken when handling solutions with DMHg exceeding natural levels in the laboratory.This paper is not to be used as a protocol for handling and safely conducting experiments with DMHg.However, the presented data and synthetization procedure should be considered when designing future studies to reduce the risk of experimental errors and DMHg exposure.