The determination of dimethyl sulfoxide in natural waters using electrochemical reduction

A highly specific electrochemical reduction method has been developed that enables the trace level measurement of dimethyl sulfoxide (DMSO) concentration in natural waters. Following the sparging of native dimethyl sulfide (DMS) from the sample, DMSO is reduced to DMS using a novel electrochemical workflow that relies upon CuSO4 as a redox mediator. The DMS produced through DMSO reduction is collected, concentrated, and detected using a previously described Purge & Trap‐Atmospheric Pressure Chemical Ionization‐Tandem Mass Spectrometry (P&T‐APCI‐MS/MS) analytical workflow. The method provides a 0.5 pM detection limit for the analysis of DMSO in 10 mL sample volumes, with a demonstrated method precision of 5.4% for the analysis of consecutive 10 nM aqueous standards. The method selectivity for DMSO was evaluated using a range of commonly observed marine organosulfur compounds, none of which were found to interfere with the analysis at a reduction potential of 4 V. Method intercomparison confirmed that the electrochemical reduction provides results that are equivalent (at the 95% confidence level) to an established TiCl3 reduction protocol for the analysis of both freshwater and seawater samples. Relative to established methods of DMSO reduction, the electrochemical method provides excellent selectivity and reproducibility, and offers the potential for automated, high‐throughput analysis. In addition, the new electrochemical method does not require expensive, difficult to procure enzymes or hazardous, corrosive chemical reagents. Depth profile measurements of DMSO, DMS, and dimethylsulfoniopropionate (DMSP) for unfiltered seawater samples collected in Saanich Inlet, a coastal fjord in British Columbia, demonstrate the effectiveness of the DMSO reduction method in an oceanographic context.

Dimethyl sulfoxide (DMSO) is widely distributed in both marine and freshwater environments (Andreae 1980;Gibson et al. 1990) and can also be found in appreciable concentrations in rainwater (Harvey and Lang 1986;Kiene and Gerard 1994). This compound is ubiquitous within surface seawater (Sim o and Vila-Costa 2006) and may also be detected throughout the water column at trace or ultra-trace levels. It is produced and consumed through biological (Hatton et al. 2004) and abiotic (Brimblecombe and Shooter 1986) processes and is thought to play an active role in the cycles of sulfur, carbon, and energy within aquatic systems. Interest in the marine distribution of DMSO and its biogeochemical cycling is predominantly driven by its relationship to dimethyl sulfide (DMS), a volatile, climate-active gas. Emission of marine DMS is the main biogenic source of nonanthropogenic organic sulfur to the atmosphere (Bates et al. 1992;Gondwe et al. 2003), where it oxidizes rapidly to form sulfate aerosols, which backscatter incident solar radiation and act as cloud condensation nuclei. For this reason, oceanic DMS emission has been suggested as potential feedback control on global climate (Charlson et al. 1987).
Metabolic reduction of DMSO may contribute significantly to the marine DMS pool (Spiese et al. 2009;Herr et al. 2021), while DMS, in turn, can be oxidized to DMSO through either biotic processes (Andreae 1980) or abiotic photochemical oxidation (Toole et al. 2003). The rates and biological function of DMSO/DMS redox cycling remain poorly understood, though it has been suggested that interconversion between DMS and DMSO may be utilized to scavenge reactive oxygen species, in response to oxidative stresses (Sunda et al. 2002). DMSO may also be produced through the intracellular oxidation and degradation of the osmolyte dimethylsulfoniopropionate (DMSP; Thume et al. 2018) rather than through the direct oxidation of DMS. In order to disentangle the underlying processes within the DMSO cycle, robust methods are needed to detect and study the distribution of this compound at low levels in natural environments.
The precise and accurate measurement of DMSO concentration within aqueous samples represents a significant analytical challenge. DMSO has a relatively high boiling point (189 C) and is decidedly polar, making it highly miscible in water and difficult to isolate from an aqueous matrix using traditional extraction techniques. Although DMSO may be sparingly extracted from an aqueous matrix using a variety of organic solvents (Ogata and Fujii 1979;Pearson et al. 1981), the reported recovery efficiency is typically poor, leading to low accuracy and under-estimation (de Mora et al. 1993). The high salt content of seawater, along with other dissolved and particulate constituents, also introduces significant matrix effects that complicate DMSO analysis. Although some direct inject gas chromatographic methods have been reported for the determination of DMSO in various sample types (Paulin et al. 1966;Watts et al. 1987), these methods typically lack the sensitivity required for the trace level determination of DMSO within a complex matrix such as seawater.
The current gold standard method for the determination of trace DMSO concentrations in aqueous samples requires the chemical reduction of DMSO to DMS prior to analysis. DMS produced through DMSO reduction is purged from solution under an inert gas flow, concentrated in a cryo-trap or upon an adsorbent substrate, then desorbed rapidly for downstream separation via gas chromatography (GC) and detection using a range of optical or mass spectrometry (MS) based methods. The chemical reduction of DMSO to DMS has been demonstrated using a range of inorganic reducing agents including: chromium (II) chloride (CrCl 2 ; Andreae 1980;Richards et al. 1994), tin (II) chloride (SnCl 2 ; Anness 1981), titanium (III) chloride (TiCl 3 ; Kiene and Gerard 1994), and sodium borohydride (NaBH 4 ;Andreae 1980;Sim o et al. 1996). Such reagent-based methods often require cumbersome sample preparation procedures, including reagent and glassware preconditioning, elevated reaction temperatures and the capture of evolved acidic vapors. In addition, strong chemical reducing agents are typically quite toxic, and thus require careful handling, storage, transport, and disposal strategies. Methodologically speaking, a significant reaction time (in excess of an hour) is often required to attain appreciable DMSO to DMS conversion using reagent-based reduction methods. These limitations may lead to poor method accuracy and reproducibility, although this may be partially mitigated using isotopically labeled DMSO internal standards, provided that a mass spectrometer is employed for detection. In addition to operational concerns, endogenous interferences are common for reagentbased reduction methods, reducing method selectivity. For example, DMSP is a common interferent for the NaBH 4 reduction method, making it necessary to correct for the contribution of DMSP to the observed DMS signal during DMSO analysis. Finally, the requirement for strict temperature control and prolonged reaction times pose challenges to method automation, throttling sample throughput and potentially impacting method reproducibility. Sim o (1998) presents a comprehensive review of the various chemical reagent-based reduction methods which are commonly utilized to convert DMSO to DMS within marine seawater samples prior to analysis. The performance attributes and limitations associate with each technique are summarized for reference.
In order to address the limitations of reagent-based reduction methods, Hatton et al. (1994) developed a highly specific enzyme-linked DMSO reduction method, in which the molybdenum-containing enzyme, DMSOreductase (DMSOR), is used to catalyze the reduction of DMSO to DMS. DMSOR may be isolated and purified from a live culture of the bacterium Rhodobacter Capsulatus, with an aliquot added directly to each sample alongside ethylenediaminetetraacetic acid and flavin mononucleotide to facilitate catalytic electron donation and transport. The sample is then briefly sparged with N 2 to create a semi-anaerobic environment, and illuminated for approximately 30 min, promoting the efficient and selective reduction of DMSO. This enzymatic method has been shown to be relatively fast, accurate, and selective, with no endogenous interferences reported for the analysis of trace DMSO in natural seawater samples. Asher et al. (2015) were able to successfully automate the DMSOR reduction protocol, developing an analysis workflow for the sequential determination of DMS, DSMO, and DMSP within seawater samples drawn from a ship-board underway supply. Although the enzymatic method is robust and reliable, the high commercial cost and significant production delays of the purified enzyme pose a significant challenge for routine measurements. Consequently, researchers must typically synthesize DMSOR inhouse using a live bacterial culture, in order to obtain sufficient quantity and purity of the enzyme (Hatton et al. 1994;Asher et al. 2015). Once synthesized, the enzyme is highly labile and must be stored under temperature-controlled conditions, placing additional constraints on the method. These factors have limited the practicality of the enzymatic reduction method for routine, high-throughput, field-based DMSO measurements.
In this paper, an alternative electrochemical DMSO reduction method is presented that supports the quantitative detection of trace level DMSO in discrete seawater samples. The new method is rapid, accurate, selective, and reproducible and does not require expensive, hazardous, or labile reagents that may be difficult to procure. If a mass spectrometer is utilized for detection, internal standards may be used to further improve method accuracy and reproducibility, while also shortening analysis times by correcting for incomplete analyte reduction. Although the method was developed explicitly for the analysis of seawater samples, the workflow described herein may be applied to the determination of trace DMSO in other matrices such as freshwater, food and beverage, industrial, pharmaceutical, and clinical samples.

Standards and reagents
DMS, DMSO, sodium hydroxide, and methanol (each with a purity >99.9%) were all purchased from Sigma Aldrich and used without further purification. Concentrated sulfuric acid (95-99%) and optima grade hydrogen peroxide (30%) were both purchased from Fisher Scientific, while copper (II) sulfate solution (20% w/w) was purchased from VWR Canada. DMS and DMSO were each separately diluted in methanol to produce individual stock solutions with concentrations of 0.034 and 0.035 M, respectively. The stock solutions were sequentially diluted in deionized water (produced using an in-house Millipore MilliQ generator) to create low-concentration standards on a daily basis. Both natural and isotopically labeled D 6 -DMSP standards were previously synthesized using the method of Challenger and Simpson (1948) and used without further purification. Stock DMSP standard solutions were prepared in methanol (acidified with 1% H 2 SO 4 ) and then diluted in deionized water as required. In order to assess potential method interferences, dimethyl sulfone, glutathione, methionine, thiamin, cystine, cysteine, homocysteine, biotin, methionine sulfoxide, and s-methylmethionine were each purchased as dry solids from Sigma Aldrich at the highest available purity (≥ 98%). Each solid was analytically weighed and dissolved in methanol to produce individual single-component standard solutions with concentrations of $ 10 mM. Each solution was then further diluted in deionized water as required. Highpurity D 6 -DMSO (> 99.9%) was purchased from Sigma Aldrich and utilized as an internal standard for analyses in which the electrochemical reduction of DMSO was not allowed to reach completion.

Apparatus
The custom-built purge and trap sample handling system and commercial mass spectrometer used for the detection of DMS have been described previously by McCulloch et al. (2020). This detection system was coupled with a custom-built electrochemical cell, based on a membrane separated H-cell design (Fig. 1). Undivided cell configurations were also explored, but these yielded highly variable DMSO reduction efficiency and poor reproducibility.
Glass components of the H-cell were custom fabricated using glass blowing facilities at the University of British Columbia. We note, however, that similar electrochemical cell designs are widely available commercially. To assemble the electrochemical cell, a 4 cm Â 4 cm preconditioned piece of Nafion (see membrane conditioning procedure below) is sandwiched between two circular ground glass flange joints, defining the anode and cathode half-cells. An S/J style joint clamp is used to securely hold the membrane between the ground glass flanges. Each half-cell is designed to have an internal volume of approximately 20 mL, limiting the liquid sample volume to around 10 mL when sufficient headspace is provided for effective sample sparging. Conical ground glass joints (female 19/26 joint) located at the top of each half-cell are included to provide access to both the anode and cathode cells, while allowing each electrode assembly to be easily removed and reproducibly reinserted into its respective halfcell. Two polytetrafluoroethylene (PTFE) thermometer centering stoppers (1/4 00 ID) are used to securely hold each of the electrode assemblies described below. Since the cathode cell is pressurized during the sample sparging process, a custom polyvinyl chloride (PVC) retaining clip is installed to maintain a robust seal between the joints and stoppers.
High-purity platinum foil (> 99.9%, 0.025 00 thick) was purchased from Alfa Aesar. Rectangular-shaped platinum electrodes are produced by cutting the foil into two identical 1 Â 2 cm strips. The foil strips are then securely connected to pieces of nickel-clad copper wire ($ 1/16 00 OD) using a rudimentary, but effective compression method as illustrated in Fig. 1b. A short length of PTFE tubing (1/8 00 OD Â 1/16 00 ID) is placed around the electrode wire, in order to insulate it from the stainless steel body of the electrode assembly. A 1/4 00 Swagelok 316 stainless steel cross union is used to support the cathode electrode assembly. The electrode wire is first roughly straightened by hand and inserted vertically through the union. The wire is then secured at the top of the union using a 1/4 00 to 1/16 00 PTFE reducing ferrule and a 1/4 00 Swagelok nut. A 1/16 00 OD (0.03 00 ID) length of PEEK tubing is inserted through one side of the union and carefully bent down (90 ) through the bottom of the union in order to introduce a flow of nitrogen sparge gas to the cell. The sparge tube is secured to the side arm of the union using a 1/4 00 to 1/16 00 PTFE reducing ferrule and 1/4 00 nut. The remaining arm of the union is connected to the trapping system using a 1/8 00 OD (1/16 00 ID) length of PTFE tubing, which is secured to the electrode assembly union using a 1/4 00 to 1/8 00 PTFE reducing ferrule. Finally, the cross union is connected to the top of the PTFE stopper using a short length of 1/4 OD (3/16 00 ID) PTFE tubing and standard 1/4 00 stainless steel Swagelok ferrules. The anode assembly is constructed using the same basic procedure, replacing the cross union with a Swagelok T-union, and eliminating the sparging tube from the anode cell configuration. The side arm of the T-union is left open to the atmosphere, allowing volatile oxidation products generated within the anode half-cell to ventilate.
A variable benchtop 15-V, 2-amp DC power supply is used to apply an electrical potential to the electrodes using breakout alligator test clips. Based on the results of various tests (see below), a constant potential of 4 V is applied to the cell for all reductive analyses unless otherwise specified. As a diagnostic measure of method performance, a basic digital multimeter is used to monitor the current delivered to the sample throughout the electrolysis process. No reference electrode is included in the current electrochemical cell design.

Nafion membrane activation
Nafion is a polymer with a polyfluorocarbon backbone and a high concentration of sulfonic acid ligands. The membrane material is often purchased as a thin dry film, containing sodium ions bound to the sulfonic ligands. Such membranes may require conditioning or activation prior to use, replacing the sodium ions with protons. In this study, a sheet of Nafion 117 cation exchange membrane material was purchased from fuelcellstores.com. The membrane was conditioned according to a procedure described by Pujiastuti and Onggo (2016). Briefly, a 4 cm Â 4 cm piece of Nafion 117 is first placed in a 3% w/w hydrogen peroxide solution for 1 h at 80 C. The membrane is then transferred to a deionized water bath at 80 C for an additional hour in order to remove any traces of H 2 O 2 . The membrane is then transferred to a 1 M sulfuric acid bath maintained at 80 C for 1 h before a final soak in deionized water at 80 C for 1 h. The conditioned membrane is then submerged in deionized water for storage prior to use. Throughout the study, the electrochemical cell was filled with deionized water when not in use, submerging the active surfaces of the membrane.

Analytical workflows
Electrochemical reduction of DMSO Using a pipette, a fresh aliquot of 0.1 M H 2 SO 4 is added to the anode (positive) half-cell prior to the analysis of each new sample. For the apparatus described, a 10 mL volume of H 2 SO 4 is sufficient to completely cover the surface of the membrane, while only partially submerging the platinum foil electrode. The anode electrode assembly is inserted into the half-cell with the foil surface oriented roughly perpendicular to the membrane. Aqueous sample (10 mL) is then deposited into the cathode (negative) half-cell, containing a clean Teflon-coated stir bar, and 10 μL of 20% CuSO 4 solution is delivered to the sample volume via micropipette. In the present method, copper is thought to serve as a catalyst or redox mediator for the two-electron reduction of DMSO, enhancing both the efficiency and reproducibility of the method. When a potential is applied to the cell, a substantial reddish-brown precipitate can be seen forming on the surface of the platinum foil cathode. Copper itself has a relatively high standard reduction potential (E = +0.34 V) and is easily reduced via electrolysis, depositing upon the cathode during the reduction process. All or part of the precipitate could also be comprised of a copper-based inorganic complex such as CuH which itself has been utilized as a reductive catalyst in organic synthesis reactions including the reduction of carboxylic acids, aldehydes (Zhou et al. 2018) and ketones (Moser et al. 2010). When reduction is performed in the absence of added CuSO 4 , the conversion of DMSO to DMS may still be observed, albeit with poor conversion efficiency and high variability between replicates. It is assumed that the reduction of DMSO in the absence of added copper sulfate may be supported by some combination of carryover contaminants, solvent impurities, or co-introduced sample matrix components. Further study will be needed to elucidate the mechanistic role of copper sulfate within the reduction process and explore alternate reagents that could improve the rate and reproducibility of DMSO conversion.
Sodium chloride solution was added to the sample volume in order to increase the electrical conductivity of the cell and the corresponding rate of DMSO reduction. Adjustment of the sample conductivity is particularly critical for freshwater samples. The addition of 1 mL of a 30% NaCl aqueous solution to a 10 mL sample volume should produce a salt concentration roughly on par with natural seawater. This level of conductivity is sufficient to support the complete reduction of DMSO to DMS in a reasonable time frame. With further study, alternate electrolytes could be identified which serve to further improve method performance.
An Internal Standard (IS) may be employed for analyses where DMSO reduction is not able to reach completion. Where applicable, 10 μL of a 10 μM D 6 -DMSO aqueous standard was added to the sample, producing an IS concentration of 10 nM. The Purge & Trap-Atmospheric Pressure Chemical Ionization-Tandem Mass Spectrometry (P&T-APCI-MS/MS) detection method used in this study provides sub-picomolar detection limits for the analysis of DMS within discrete seawater samples, with the added capability of highly selective mass filtering. As a result, isotopically labeled standards may be employed to substantially reduce the reduction time and improve both method accuracy and reproducibility, provided that the native DMSO concentration is sufficient to allow detection. In the absence of internal standards, it is necessary to allow the electrochemical reduction to proceed to completion to obtain accurate, quantitative measurements of DMSO concentration.
Once each reagent has been delivered to the sample, the electrochemical cell is sealed by inserting the electrode assemblies into their respective half-cells. The PVC retaining clip is then connected to the top of the cathode cell, preventing leakage and the loss of volatile analytes during sparging. For the analysis of both fresh and seawater samples, a pre-sparge may be necessary to remove native DMS prior to the reduction of DMSO. Native DMS, along with other volatile compounds present within the matrix, may either be vented to the atmosphere, or trapped and desorbed for analysis using the PT-APCI-MS/MS workflow. Gentle stirring is initiated and the electrical leads are connected to their respective electrode assemblies. Although stirring is not inherently required for electrochemical reduction, the reduction rate in the absence of stirring may be limited by the diffusive transport of analyte and/or critical reagents to the electrode surface. Continuous sparging during reduction also creates turbulent mixing, which accelerates the reaction rate compared to samples that are not stirred or sparged. For this study, the sample was continuously sparged with high-purity (99.999%) N 2 at a flow rate of 100 mL min À1 throughout the DMSO reduction process in order to remove and trap DMS as it is generated. For all analyses in this study, the sample was sparged for an additional 2 min beyond the electrochemical reduction period, ensuring that all DMS generated was fully stripped and collected prior to analysis.

Sequential analysis of DMS, DMSO, and DMSP
A timing diagram is presented in Fig. 2, illustrating the steps involved in the determination of DMS, DMSO, and DMSP within a single discrete aqueous sample. The analysis sequence involves a number of sequential steps. First, a fresh sample is added to the electrochemical cell, along with the required reagents and an IS. To increase reproducibility between replicates, a fresh aliquot of H 2 SO 4 is added to the Fig. 2. Timing diagram for a workflow that enables the sequential analysis of DMS, DMSO, and DMSP within a single 10 mL aqueous sample. Note that an internal standard is employed to compensate for the incomplete reduction of DMSO to DMS within a restricted 5-min reaction time.
anode half-cell prior to the analysis of each new sample. Native DMS is first sparged from the solution and measured using P&T-APCI-MS/MS (McCulloch et al. 2020). For a 10 mL sample volume, 7-min of sparging is sufficient to entirely remove all native DMS from the sample.
Following the removal of background DMS, DMSO is then electrochemically reduced to DMS, collected, and analyzed as described above. The required duration of the reduction phase is dependent upon the concentration of DMSO and the performance characteristics of the detector utilized. When using more advanced detectors that can detect isotopically-labeled compounds, only a small amount of DMSO reduction needs to be performed. For the P&T-APCI-MS/MS method utilized here, 5 min of reduction was typically sufficient for most surface seawater samples when an IS was utilized. For complete conversion of the native DMSO to DMS up to a 30-min reduction period may be required.
As noted above, a rust colored precipitate may be seen forming upon the submerged portion of the platinum electrode. This deposit can be easily removed from the electrode surface by briefly reversing the polarity of the applied potential for 10-20 s between analyses. The precipitate is rapidly reoxidized, dissolving back into solution, leaving behind a clean platinum foil surface. A third sacrificial copper wire electrode may also be used in place of the platinum anode counter electrode assembly in order to collect any copper that may travel across the membrane during the electrode cleaning process.
After DMSO reduction is complete, some or all of the sample may be transferred via syringe or pipette into an auxiliary sparging vessel, where native DMSP is hydrolyzed to cleave DMS prior to analysis. The hydrolysis of DMSP requires the addition of a highly concentrated base (typically $ 10 M NaOH), which may degrade the performance of the Nafion membrane. For this reason, DMSP hydrolysis and analysis are performed within a dedicated all-glass sparging chamber, which does not contact the Nafion exchange membrane. Once the analysis of all compounds is complete, the sample is removed from the apparatus via syringe, and both the electrochemical cells and the auxiliary sparging chamber are thoroughly rinsed with deionized water prior to the introduction of the next sample.
Field study, sampling, and ancillary data collection Seawater sampling was performed at two locations within the coastal waters of Vancouver, British Columbia, Canada. The first set of samples was obtained during a single-day expedition on 07 October 2021 aboard the RV John Strickland. Seawater was collected at a single sampling location (SI03) in Saanich Inlet (48 35.30 N, 123 30.22 W), a well-characterized, seasonally anoxic fjord on the southeast coast of Vancouver Island. A depth profile of seawater samples was collected using a combination of Niskin and Go-Flow bottles from depths ranging from the surface down to 200 m-approximately 28 m from the sea floor. On 25 October 2021, a large volume seawater sample was collected from the mouth of False Creek Inlet in Vancouver, British Columbia (49 16 0 37.4 00 N 123 08 0 23.4 00 W). This water was used to perform a series of basic method development and characterization experiments. On 20 November 2021, a freshwater sample was collected from Burnaby Lake, Burnaby, British Columbia (49 14 0 32.4 00 N 122 56 0 07.2 00 W) using a single Niskin bottle closed just below the surface (depth 0 m). For all experiments, water collected for DMS/DMSO/DMSP analysis was distributed into 20-mL serum vials using a short length of thick-walled silicon tubing. Each vial was over filled three volumes before being closed using a butyl rubber stopper and crimp sealed. Samples for DMS analysis were sealed without entrained bubbles or headspace. These samples were analyzed at the University of British Columbia as soon as possible following collection (a maximum delay of 8 h from the time of collection). DMSO samples were half-filled (to avoid vial cracking during preservation/storage), crimp sealed, and frozen at À80 C prior to storage in the dark for up to 1 week prior to analysis. These samples were thawed and equilibrated in a room temperature water bath prior to analysis. DMSP samples were filled without headspace and preserved via acidification with 100 μL of concentrated sulfuric acid (50% v/v) prior to sealing.

Characterization of the electrochemical reduction method
For the apparatus and workflow described above, the parameters summarized in Table 1 were determined to be suitable for the quantitative detection of trace level DMSO within a 10 mL natural water sample volume, providing an excellent combination of reduction efficiency and interference-free performance. Below, the impact of various operational parameters on the performance of the method is discussed.
The rate of DMSO conversion to DMS and the associated reduction efficiency may be influenced by the chemical Internal std (D 6 -DMSO) concentration 10 nM composition of the sample, the total cell conductivity, and the strength of the applied potential. As a result, a series of tests were performed to examine the influence of both H 2 SO 4 and CuSO 4 concentration, and the reduction potential on the observed conversion efficiency. The results presented in Fig. 3a demonstrate the effect of H 2 SO 4 concentration on the DMS signal response obtained for the analysis of a 10 nM DMSO standard. A 5-min reduction period was adopted to ensure that DMS generated through DMSO reduction was not limited by the amount of analyte remaining in the solution. Test solutions were prepared as pure aqueous standards (containing 3% sodium chloride w/v), and also with an unfiltered seawater matrix. The seawater standards contained a 10 nM isotopically labeled D 6 -DMSO IS to differentiate isotopically labeled DMS from the naturally occurring DMSO. Under both matrix conditions, maximum conversion efficiency was observed with an H 2 SO 4 concentration of 0.1 M, with decreasing efficiency above and below this concentration. Although the matrix composition appears to influence DMSO reduction efficiency, performance may be optimized toward a concentration of around 0.1 M. The results presented in Fig. 3b demonstrate the influence of CuSO 4 concentration on DMSO reduction across a range of reaction times. In the absence of added CuSO 4 , the electrochemical reduction of DMSO to DMS was found to be inefficient, and the conversion failed to reach completion within a 30-min period. Even a small addition of the CuSO 4 solution significantly increased the efficiency and reproducibility of the method. For short reduction periods (5 min or less), higher copper concentration yielded increased reduction rates and DMS yields. However, for reduction periods beyond 5 min, excessive CuSO 4 concentration did not necessarily result in improved DMS yields. Indeed, at CuSO 4 concentrations at or above 2 mM, the process failed to achieve complete DMSO conversion, even when a 30-min reduction time was provided. Best performance was obtained at a CuSO 4 concentration between 0.08 and 0.8 mM. Below this range, the method failed to attain complete conversion within a 30-min reduction period. Under the conditions described in Table 1, complete DMSO to DMS conversion was attained within approximately 25 min of reduction time. To ensure complete DMSO conversion, 30 min of reduction was employed for all further experiments unless otherwise stated.
The voltage applied to the electrochemical cell can influence both the efficiency of the reduction and selectivity of the method, as demonstrated by the data presented in Fig. 3c. Modest DMSO reduction (resulting in detectable DMS generation) was observed at a potential of 2.5 V (data not shown), with no conversion observed below this potential. Within the narrow window of 3.0 and 3.3 V, there was a substantial increase in DMSO reduction efficiency, although the conversion failed to reach completion within a 30-min reaction period. Between 4.0 and 6.0 V, there was no significant improvement in the reduction efficiency, and the method was able to attain complete DMSO reduction within about 25 min. Although there may be chemical environments and/or apparatus configurations that provide enhanced performance, these results suggest that cell operation beyond 4.0 V is not advantageous, and this value was selected for all subsequent experiments, unless otherwise specified.
The observed voltage dependence of DMSO reduction provides some insight into possible underlying mechanisms. Below 2.5 V, DMSO conversion was not observed. This result suggests that H 2 gas (which is produced at the cathode at potentials above $ 1.8 V) is not in itself a strong enough reducing agent to facilitate DMSO reduction in seawater, although hydrogen may still be coupled to the mechanisms responsible for DMSO conversion. At present, the exact mechanism underlying the two-electron electrochemical reduction of DMSO remains unclear. We note, however, that DMSO has been utilized as a nonaqueous solvent within electrochemical methods for decades, providing some validation to our observations. As an electrochemical solvent, DMSO is reported to have a cathodic limit ranging from À1.8 to À3.0 V, beyond which it is reduced on the electrode surface (Aurbach and Gofer 1999). This limit may be influenced by the solution temperature, electrode materials, and supporting electrolyte(s) selected. Beyond its cathodic limit, electrolysis of DMSO may result in the formation of various volatile products including hydrogen, methane (Giordano et al. 1966), and, as shown by our results, dimethyl sulfide.

Assessing potential interferences
The results presented in Fig. 4 demonstrate the impact of reduction potential on DMS production for solutions containing both DMSO and the prospective marine interferent DMSP. For these experiments, the reduction period was limited to 5 min to ensure that depletion of the DMSO content did not result in diminished DMS yield. Data are provided for both the analysis of a single-component 10 nM DMSO standard (circles) and a 100 nM DMSP standard (squares). Results presented in Fig. 4a confirm that DMSO reduction is observed at potentials above approximately 2.5 V, with a continued increase in efficiency up to roughly 8.0 V. At higher potentials, the reduction yield was found to diminish slightly, potentially due to competitive side reactions at the electrode surface. Results obtained for the analysis of the DMSP standard demonstrate apparent DMS formation from DMSP cleavage at a potential greater than approximately 2.5 V. However, when the data for both DMSO and DMSP are normalized to the signal intensity obtained for the DMSO standard at 4.5 V, the traces overlap nearly perfectly up to the point of normalization, but diverge significantly beyond this point (Fig. 4b).
This observation empirically suggests that DMSO could be present as a contaminant within the synthesized DMSP stock solution utilized. Indeed, the presence of significant DMSO contamination (at approximately 3% of the DMSP concentration) was confirmed via HPLC-APCI-MS/MS analysis (data not shown). Figure 4c presents normalized DMSP data with the normalized DMSO contribution subtracted, leaving behind only the DMS production that is likely attributable to DMSP cleavage. These results suggest that DMSP is not likely to be a significant interferant up to approximately 4.5 V, although it may contribute to DMS production at higher voltages. Based on this finding, all further experiments were performed with an electrochemical cell potential of 4 V to maximize the overall DMSO reduction efficiency and minimize interference from DMSP. We note, however, that the cleavage of DMSP to DMS at higher voltages could enable a method for sequential measurement of DMS, DMSO, and DMSP within a single sample, using an entirely electrochemically driven approach.
Beyond DMSP, other co-introduced organosulfur compounds could also participate in electrochemically driven degradation pathways, leading to the formation of DMS and an apparent overestimation of the DMSO concentration. To assess the likelihood for such interferences over a range of reduction potentials from 2.5 to 10 V, we examined the electrochemical production of DMS from a variety of aqueous standards, each containing a single prospective marine organosulfur interferent. At a reduction potential of 4 V, DMS production was only observed for the analysis of the DMSO standard, and only DMSP exhibited significant reduction at 7 V, as noted in Table 2. Aside from DMSP, none of the other prospective interferents displayed evidence of electrochemically driven DMS production across the range of potentials evaluated. This result suggests that the electrochemical method is unlikely to be impacted by the presence of common native organosulfur interferents.
Competitive inhibition of the electrochemical method is not anticipated, as highly electromotive matrix components (e.g., Au, Ag, Pt, etc.) are unlikely to be present in excessive quantities. Such substances could potentially act to divert electrical current away from the desired DMSO reduction pathway, but our results suggest that they do not pose a significant problem in our method. Water electrolysis (evidenced through the ample generation of hydrogen gas bubbles at the cathode surface) is a potential competitive process, but this pathway does not appear to prevent the reduction of DMSO, though it could influence method efficiency and affect the time required for complete reduction.

Method intercomparison
The performance of the electrochemical method was evaluated against an established titanium chloride (TiCl 3 ) DMSO reduction protocol (Kiene and Gerard 1994) for the analysis of both seawater and freshwater samples. For these tests, both long (30 min) and short (5 and 10 min for seawater and   freshwater, respectively) reduction protocols were utilized. The results of this method comparison are summarized in Table 3. Statistical analysis (Student's T-test) of the results from the method intercomparison indicate that these reduction methods produced DMSO concentrations that were not different at the 95% confidence level. These results also indicate that isotopically labeled internal standards may be confidently leveraged to decrease the reduction time, provided that a mass spectrometer with sufficient sensitivity is available for detection.

Precision and linearity
Method reproducibility was examined through the replicate analysis of a 10 nM DMSO aqueous standard (10 mL sample volume with a 3% NaCl concentration). For the analysis of 10 consecutive samples, the coefficient of variation between sequential DMS signals was 5.4% by peak area and 5.5% by peak height. This level of precision is consistent with the reproducibility provided by the P&T-APCI-MS/MS method used for DMS analysis (3.9% by peak area and 4.1% by peak height; McCulloch et al. 2020), indicating that the introduction of the electrochemical reduction method does not substantially impact the reproducibility of the overall analysis.
The level of reproducibility provided is also consistent with other purge and trap GC-based DMS detection techniques (Kiene 1996;Zindler et al. 2012;Zhang and Chen 2015). Figure 5 shows a calibration curve for the analysis of DMSO standards, demonstrating strong method linearity (R 2 = 0.9996) over four orders of magnitude in concentration (0.1-100 nM). This concentration range is suitable for the measurement of DMSO at levels commonly observed in both marine and freshwater environments. Method sensitivity was determined based upon the slope of the calibration curve, resulting in a derived detection limit (3σ) of 0.5 pM, equivalent to 5 femtomoles of DMSO within a 10 mL sample volume.

Field application
Practical application of the electrochemical reduction method was demonstrated using samples collected from Saanich Inlet off the southeastern coast of Vancouver Island, British Columbia. Figure 6a presents a depth profile of discrete DMS, DMSO, and DMSP measurements throughout the Saanich Inlet water column, alongside dissolved oxygen and chlorophyll fluorescence data to provide an oceanographic context. DMSO concentrations, determined using the electrochemical reduction method, ranged from a minimum of 0.38 nM in the deep subsurface region up to a maximum of 34.3 nM near the surface. DMS and DMSP depth profiles also exhibited similar maxima in surface waters, with the highest concentrations of 6.78 and 101 nM, respectively, within the euphotic zone. The observed near-surface maxima for DMS, DMSO, and DMSP correlate well with the chlorophyll fluorescence maximum, suggesting they are likely derived from phytoplankton metabolism. Distinct trace level (sub-nanomolar) DMS, DMSO, and DMSP accumulation was also detected throughout the deeper water column (Fig. 6b), with several localized minima and maxima detected well into the subeuphotic region. At a depth of 120 m, for example, DMSP levels rapidly increased from below 1 nM to more than 5 nM. This increase in DMSP coincides with the anoxic boundary of the water column. In contrast, DMS and DMSO both demonstrated detectable minima at a sampling depth of 135 m, and weak maxima at a depth of 150 m, well separated from the anoxic DMSP signature above. The sampling depth resolution was limited to 10-15 m throughout this portion of the water column. As a result, it is difficult to draw explicit connections between observed organosulfur accumulations observed within these regions. An additional maximum can be seen in all three sulfur compounds near the sea floor (200 m depth). This maximum appears to correlate well with the reintroduction of dissolved oxygen at depth, during the late summer deep-water renewal characteristic of Saanich Inlet. Further study will be necessary to fully elucidate the roles of DMS, DMO, and DMSP within this unique oxygen-depleted ecosystem and the associated microbial community. Nonetheless, the results presented in Fig. 6 highlight the capability of the method to capture interesting small-scale features in the distribution of DMS, DMSP, and DMSO, across natural ocean redox gradients.

Conclusions
This study demonstrates the performance and practicality of a new electrochemically driven reduction protocol, which facilitates the accurate, selective, and reproducible measurement of DMSO in aqueous solutions. When coupled with the P&T-APCI-MS/MS workflow described previously (McCulloch et al. 2020), the method can detect DMSO at concentrations typically observed within both fresh and seawater samples. The new electrochemical reduction method represents an alternative to existing enzyme-linked and reagent-based chemical reduction methods for DMSO analysis. Many of the components needed for the electrochemical apparatus may be sourced commercially at low cost, and there are no expensive, labile reagents or difficult to procure chemicals required, making it accessible to a broad range of users. Beyond the analysis of natural waters, the method may also be useful for the measurement DMSO concentration in food and beverage samples, pharmaceuticals, and some clinical matrices.
In the future, we aim to further improve the efficiency and throughput of the DMSO reduction method. For instance, the dimensions of the electrochemical cell, electrodes, and membrane may be refined to improve performance, and potentially reduce the time required for complete DMSO reduction. We also intend to study the long-term stability of the cell components, in particular the proton exchange membrane, and determine if alternate materials exist which may provide performance advantages. Following the analysis of hundreds of seawater samples over the course of several months, no performance loss or membrane degradation was observed, even under the highly acidic and saline conditions associated with this study. We will also explore ways to improve the robustness of the apparatus to increase method reproducibility and accuracy, while further examining the possibility of quantitative DMSP cleavage to DMS at elevated electrochemical potentials. We intend to incorporate the electrochemical DMSO reduction method into our existing automated analysis system, OSSCAR (Asher et al. 2015) which supports the online analysis of DMS, DMSO, and DMSP within seawater samples.