A simple and reliable method reducing sulfate to sulfide for multiple sulfur isotope analysis

Rationale Precise analysis of four sulfur isotopes of sulfate in geological and environmental samples provides the means to extract unique information in wide geological contexts. Reduction of sulfate to sulfide is the first step to access such information. The conventional reduction method suffers from a cumbersome distillation system, long reaction time and large volume of the reducing solution. We present a new and simple method enabling the process of multiple samples at one time with a much reduced volume of reducing solution. Methods One mL of reducing solution made of HI and NaH2PO2 was added to a septum glass tube with dry sulfate. The tube was heated at 124°C and the produced H2S was purged with inert gas (He or N2) through gas‐washing tubes and then collected by NaOH solution. The collected H2S was converted into Ag2S by adding AgNO3 solution and the co‐precipitated Ag2O was removed by adding a few drops of concentrated HNO3. Results Within 2–3 h, a 100% yield was observed for samples with 0.2–2.5 μmol Na2SO4. The reduction rate was much slower for BaSO4 and a complete reduction was not observed. International sulfur reference materials, NBS‐127, SO‐5 and SO‐6, were processed with this method, and the measured against accepted δ34S values yielded a linear regression line which had a slope of 0.99 ± 0.01 and a R 2 value of 0.998. Conclusions The new methodology is easy to handle and allows us to process multiple samples at a time. It has also demonstrated good reproducibility in terms of H2S yield and for further isotope analysis. It is thus a good alternative to the conventional manual method, especially when processing samples with limited amount of sulfate available.


| INTRODUCTION
Stable sulfur isotopes have been widely used to trace a range of biogeochemical processes. 1 The discovery in 2000 of the mass-independent isotopic fractionations of sulfur isotopes (S-MIF) in sulfate and sulfide in Archean rocks 2 showed the potential of the S-MIF signals for tracking the oxygenation of the atmosphere 2.4 Gy ago, 3 and the geochemical evolution of Mars. 4 The S-MIF signals in ice-core sulfate have also been observed and demonstrated to be useful for tracking the sulfur cycle in today's stratosphere and they serve as a unique proxy of large volcanic eruptions that inject sulfur into the stratosphere and thus have global climate impacts. [5][6][7][8] Multiple sulfur isotope compositions can also help to constrain the oceanic sulfur cycle (e.g., 9,10 ).
To access the S-MIF signals, precise analysis of the four sulfur isotopes ( 32 S, 33 S, 34 S and 36 S) is necessary. The isotopic results are expressed as δ 3x S = 3x R sample / 3x R CDT − 1, where x = 3, 4, and 6, and the δ values are calculated using the CDT standard. The S-MIF values are then defined by: The isotopic analysis is conventionally performed by reducing sulfate (SO 4 2− ) to hydrogen sulfide (H 2 S), converting H 2 S into silver sulfide (Ag 2 S), and fluorinating Ag 2 S to sulfur hexafluoride (SF 6 ) for isotopic composition analysis by isotope ratio mass spectrometry (IRMS). 2,6,11,12 The reduction from SO 4 2− to H 2 S is mainly achieved by two different reducing agents: tin(II) (Sn 2+ ) solutions and hydroiodic acid (HI)/hypophosphorous acid (H 3 PO 2 ) mixtures. [13][14][15] The Sn 2+ solution is mainly applied to solid samples (e.g., minerals) with an optimum reaction temperature between 280 and 300°C, and the HI reducing solution can be applied to aqueous samples at 100-125°C. 14 Currently, the most widely used reducing method in sulfur isotope geochemistry follows the reducing agent recipe (500 mL concentrated HI, 816 mL concentrated HCl, and 245 mL 50% H 3 PO 2 ) of Thode et al., 16 and uses a distillation apparatus similar to that described in Forrest and Newman. 17 In the reducing solution of Thode et al, 16 high concentrations of HI seem to be the most important component of the reducing agent for complete sulfate reduction, and the presence of H 3 PO 2 or NaH 2 PO 2 increases the reduction speed by maintaining a high hydroiodic acid to iodine ratio which is one of the factors favoring the reduction. 14,18 HCl is only of secondary importance and its presence is suggested to increase the acidity and volume, and reduce the use of relatively expensive HI. 13,19 However, Gustafsson 20 found the presence of water to be detrimental for the reduction because water tends to dilute and thus lower the concentration of HI, and at lower HI concentration, side products (viz, SO 2 and elemental S) will be formed. 18  reducing solution containing only HI (57%) and NaH 2 PO 2 salt, and found a good reduction yield. In particular, Davis and Lindstrom 18 found that the optimum composition of the reducing solution for complete and fast sulfate reduction is 0.13 g NaH 2 PO 2 in 1 mL HI (57%).
In these studies, aqueous sulfate samples were processed and a cumbersome distillation apparatus was used.
In summary, it seems that the best composition of the reducing solution would be a mixture of 0.13 g NaH 2 PO 2 in 1 mL HI (57%), and the amount of water in the reduction experiment should be limited.

| Apparatus
The reduction train is sketched in Figure 1. ) was added to a pre-cleaned reaction tube. The reaction tube was allowed to completely dry in a 100°C oven, and the sample was then stored for later use.
In order to prepare the BaSO 4 samples, the desired volume (e.g.,  However, the purpose of processing these samples is to test potential sulfur isotope fractionation during the reduction, rather than to assess the reduction yield (which can be assessed from the samples made from drying Na 2 SO 4 solution with accurate measurement of sulfur content, or precipitating BaSO 4 from the same Na 2 SO 4 solution).

| Quantification
The yield of the reduction from sulfate (SO 4 2− ) to sulfide (S 2− ) can be directly assessed by determining the quantity of H 2 S collected in the NaOH trapping solution. Hydrogen sulfide (H 2 S) solution is known to absorb UV light with a peak absorbance at 230 nm. 22

| Procedure
Prior to the reduction, all glassware, caps, septum and PEEK tubes were cleaned with Milli-Q water. The PEEK tubes have to be flushed to ensure that there is no water left inside them; otherwise the water will block the flow of the carrier gas in the reduction line.
In a fume hood, 1 mL of reducing solution was added to a pre-prepared reaction tube to a known amount of dry sulfate. In the reaction tube, the reducing solution was purged with He for 20 min at room temperature to remove any I 2 and O 2 . The gas washing tubes ('b1' and 'b2' in Figure 1) and the collection tube ('c' in Figure 1) were prepared by adding 12 mL Milli-Q water and 12 mL 0.1 M NaOH, respectively. After the reducing solution had been purged for 20 min, the reduction train was assembled ( Figure 1) and the reaction tube was placed in the block heater and heated at 124°C. At lower temperatures the reduction speed will be slow, while if the temperature is too high, an excessive amount of phosphine (PH 3 ) will be produced from the decomposition of NaH 2 PO 2 . 14 For the alternative setup, the drying agent was in-line with the cryogenic system, and the latter was set at −200°C to trap the reaction products. When the reaction was over, the temperature of the cryogenic trap was raised to −120°C when the produced H 2 S was released and trapped in the collection tube.
The collection tube was removed from the reduction train after the reaction was complete.

| Isotope analysis
To explore potential sulfur isotope fractionation during the reduction, Because of the small amount of samples (<0.5 mg Ag 2 S), a microvolume cold finger of an isotope ratio mass spectrometer (MAT 253; Thermo Scientific, Bremen, Germany) working in dual-inlet mode was used to concentrate the sample gas for isotope analysis. 24 The analytical uncertainty (1σ) for the instrument was 0.25‰ for δ 34 S values, 0.010‰ for Δ 33 S and 0.062‰ for Δ 36 S obtained by replicate analysis (N = 4) of IAEA-S-1 over a period of 4 weeks (once a week) when the processed sulfate standards were also measured for sulfur isotopic composition.

| H 2 S collection agents
The reduction product, H 2 S, has to be collected and converted into Ag 2 S before fluorination for isotope analysis. As mentioned above, Cd(CH 3 CO 2 ) 2 16,17 and AgNO 3 13 have both been shown to be able to efficiently trap H 2 S by forming CdS and Ag 2 S precipitates, respectively.
The CdS precipitate is further converted into Ag 2 S by adding AgNO 3 solution. 16,17 The conventional reducing solution commonly contains phosphorous acid (H 3 PO 3 ) or hypophosphorous acid (H 3 PO 2 ), 13 and phosphine (PH 3 ) is produced when the reducing solution is heated. 18 Once PH 3 comes in contact with AgNO 3 , it reduces Ag + to Ag 0 and this leads to excess precipitate in addition to Ag 2 S. 17 To prevent this, Thode et al 16

| H 2 S yield
In the 0.1 M NaOH trapping solution, sulfide was mainly present in the form of HS − ( Figure 3A). Figure 3B shows the typical absorbance spectra of two Na 2 S working standards (in 0.1 M NaOH matrix) and two NaOH trapping solutions after 2 h collection of H 2 S and, as expected, the absorbance spectra peak was at~230 nm, consistent with that from Guenther et al. 22 Figure 3C shows the plot of the average of the calibration curve over 3 days of analyzing working standards.
As described in section 2.3, three different sulfate samples were processed using our system, Na 2 SO 4 , BaSO 4 -EB and P-BaSO 4 , and the time-resolved H 2 S yields from these three materials are plotted in In general, Na 2 SO 4 was reduced faster than P-BaSO 4 , and much faster than BaSO 4 -EB. Regardless of the quantity of the starting sulfate, after 1 h of reduction an average H 2 S yield of 85.7 ± 10.3% was reached when Na 2 SO 4 was the starting material. In comparison, the H 2 S yield after 1 h of reduction was 63.9 ± 2.1% for BaSO 4 -EB and only 18.5 ± 0.04% for P-BaSO 4 . After 2 h, a 99.5 ± 3.7% yield was reached for Na 2 SO 4 , indicating the completion of the reduction.
However, after 2 h, it appeared that no more H 2 S was produced for

| Isotope analysis of the standard materials
Since the overall goal of reducing sulfate to sulfide is to perform the four-sulfur isotopes analysis, we processed three different barium   Figure 6. A least-squares linear regression gives a slope of (0.99 ± 0.01), suggesting good reproducibility and the conservation of sulfur isotopic composition during the reduction of sulfate to sulfide using our reducing system, despite the reduction yields of these standard materials not being 100%. This is not a surprise. In fact, if any sulfur isotope fractionation occurs during the reduction, it would be between the solid BaSO 4 and the dissolved HSO 4 − (the form of SO 4 2− in concentrated acid solution), but not in the step(s) from SO 4 2− to H 2 S because the dissolved part is

| CONCLUSIONS
We present a simple and reliable reducing method modified from the literature for the conversion of sulfate into sulfide for four-sulfur isotopes analysis. This system is simple to set up, easy to replace and cheap to acquire and is made from sealed test tubes and PEEK flow lines (metal part, e.g. needle, in contact with the hot reducing solution is not allowed). This method uses a reducing solution made of 100 mL 57% HI and 13 g NaH 2 PO 2 , and a very small amount (1 mL This new approach was demonstrated to produce H 2 S very rapidly with a 100% recovery when soluble sulfate salt was used (e.g., Na 2 SO 4 ), as opposed to BaSO 4 for which the kinetic was slow and conversion never reached 100% even after overnight reaction. However, despite the relatively low reduction yield for BaSO 4 , there was no significant isotope fractionation effect induced by the reduction. As it is the dissolved part of the sulfate salt that reacts with the reducing solution, this method is most suitable for natural samples containing soluble sulfate (e.g., aerosol, snow and ice core), which can be extracted (e.g., by the resin method) and converted into Na 2 SO 4 . The use of the barite precipitate method for sulfate extraction and purification is not recommended as the salt solubility inhibits the reduction speed and yield. If BaSO 4 is the main form of sulfate (e.g., barite), increasing the volume of the reducing solution and/or the reaction time may improve the H 2 S yield although there is no guarantee of a complete conversion.
While poor conversion and fluorination yields do not seem to introduce isotope fractionations, poor yield reduces the sensitivity of the method to sample sizes above a few micromoles of sulfate and it may also have consequence on the mass-dependent slopes between the sulfur isotope ratios as the 33 S/ 32 S ratios of the international standards have never been calibrated.