Mimicking the Microbial Oxidation of Elemental Sulfur with a Biphasic Electrochemical Cell

The lack of an artificial system that mimics elemental sulfur (S 8 ) oxidation by microorganisms inhibits a deep mechanistic understanding of the sulfur cycle in the biosphere and the metabolism of sulfur-oxidizing microorganisms. In this article, we present a biphasic system that mimics biochemical sulfur oxidation under ambient conditions using a liquid|liquid (L|L) electrochemical cell and gold nanoparticles (AuNPs) as an interfacial catalyst. The interface between two solvents of very different polarity is an ideal environment to oxidise S 8 , overcoming the incompatible solubilities of the hydrophobic reactants (O 2 and S 8 ) and hydrophilic products (H + , SO 32– , SO 42– , etc. ). The interfacial AuNPs provide a catalytic surface onto which O 2 and S 8 can adsorb. Control over the driving force for the reaction is provided by polarizing the L|L interface externally and tuning the Fermi level of the interfacial AuNPs by the adsorption of aqueous anions.


Introduction
Sulfur and its species are important geochemical agents, with the sulfur cycle vital to maintaining the natural equilibrium of the biosphere. Most sulfur redox reactions in the biosphere are thermodynamically allowed but kinetically sluggish, requiring catalysis by microorganisms capable of using inorganic sulfur compounds in their metabolism, either as electron donors or acceptors. [1] Indeed, the earliest organisms on Earth gained energy from the metabolism of sulfur compounds, including elemental sulfur. [2] Technologically, microbiological sulfur oxidation unpins many industrial processes such as mineral biotechnology, [1] microbial corrosion, [3] decreasing the pH of alkaline soils [4] and the development of fertilizers. [5] Plants assimilate mainly sulfate ions, but there are many commercial fertilizers that contain sulfur in elemental form. Their effectiveness depends on the rate at which elemental sulfur is oxidized to sulfate ions by microorganisms in the soil. [6] The most stable allotrope of elemental sulfur is -S8. This highly hydrophobic molecule is neither wetted nor dissolved by water. To oxidize -S8, it has been proposed that microorganisms first dissolve it in the hydrophobic regions of their outer biomembranes.
Subsequently, -S8 can be transported inside the cell by surfactants or react at the biomembrane|water interface. [7] In this context, many fundamental questions remain concerning the sulfur oxidation reaction and associated kinetics. Why is this reaction almost exclusively driven by microorganisms in the biosphere? Can a chemical pathway be designed for this reaction with a low activation energy and without the need of a biomolecule?
Biochemical reactions at a biomembrane interface are typically coupled to ion transfer and electron transfer chain reactions that either increase the electric field across the membrane or drive thermodynamically uphill reactions, like ATP synthesis. [8] A powerful approach to mimic and model such processes, involving simultaneous ion and electron transfer reactions, is to electrochemically polarize an interface between two immiscible electrolyte solutions (ITIES). [9][10][11] Electrochemical reactions at such polarized liquid|liquid (L|L) interfaces may be complex due to the simultaneous coupling of not only interfacial ion and electron transfer reactions but also homogeneous chemical reactions, interfacial adsorption of ions or neutral species and the transport of solvent molecules across the interface by diffusion or facilitated by ion transfer. [12] In the absence of a microorganism, the activation energy to form sulfate from S8, molecular oxygen (O2) and water is prohibitively high at ambient conditions. Thus, a catalyst is required to generate a new chemical pathway to oxidize S8 with a low activation energy.
Gold nanoparticles (AuNPs) [13] and nanoporous gold [14] are good catalysts for many oxidation reactions, despite the adsorption of O2 on macroscopic gold surfaces being unfavoured. [15] In particular, supported AuNPs show unusually high catalytic activity for the oxidation of carbon monoxide and alcohols by O2. [16] On the other hand, Tada et al. [2] reported that S8 can be adsorbed on AuNP surfaces with cleavage of the S-S bond at a surface coverage () < 1/3, while molecular adsorption takes place concurrently with atomic adsorption at  > 1/3.
Complications arise when using supported AuNPs, as the solid support is going to affect the electronic state of the NPs. [17] Therefore, determining the effect of solely the NPs size and electronic structure on the kinetics of a reaction, for example the O2 reduction reaction (ORR), becomes difficult. In this sense, a polarized L|L interface functionalised with a film of AuNPs has some advantages when studying catalytic reactions. For example, there are two ways to change the Fermi level of NPs at a polarized L|L interface and tune their electronic state: by the adsorption of ions or in the presence of a redox pair in solution. [18,19] The past decade has seen a major increase in activity to functionalise the polarized L|L interface with various catalytic nanomaterials to study energy related reactions, [20] especially the ORR in the presence of AuNP films. [21,22] Herein, we describe a novel methodology to functionalise a polarized L|L interface with a film of AuNPs. This article provides the first experimental evidence of the physisorption of Cland OHanions on such interfacial AuNPs. The oxidation reaction of elemental sulfur by O2 at a polarized L|L interface catalysed by AuNPs is studied under ambient conditions and highlights the role of OHanions to enhance this oxidation reaction.

Results and Discussion
AuNP film growth at the polarized liquid|liquid interface. The preparation of homogeneous and stable AuNP films at immiscible L|L interfaces is a major challenge. [21,23,24] Usually the surface is not homogeneously covered, significant aggregation of particles takes place at the interface and potential cycling can induce movement of the particles generating an interfacial convection (interfacial stirring). [25] Citrate ions are typically used to stabilize AuNP suspensions in bulk aqueous solutions. [26] However, such suspensions are very stable and, due to inter-particle electrostatic repulsion, the AuNPs rarely deposit in significant numbers at the L|L interface on contacting the aqueous phase with an immiscible organic solvent. [27] Various approaches have been demonstrated to weaken the electrostatic repulsion between AuNPs and induce interfacial deposition while retaining the constituent properties of the individual AuNPs to the greatest extent possible. These include functionalizing the surfaces of charged colloidal NPs with charge-neutral organic "modifiers" [28,29] or adding amphiphilic salts to the biphasic system. [30][31][32] The latter act as "promoters" if the salts contain organic soluble ions of opposite charge to the NPs that screen inter-particle electrostatic repulsion. [23] Here, we replace the citrate buffer solution with a phosphate buffered saline (PBS) solution, pH 7.4, see electrochemical cell 1 in Scheme 1. The PBS-capped AuNP suspensions remain stable in the bulk aqueous phase. However, if these AuNPs come into proximity with a L|L interface formed by water and -trifluorotoluene (TFT) they lose their ionic solvation shells, induced by the electric field at the interface and/or by changes of the local solvent polarity. In other words, PBS gives the dynamic freedom necessary to avoid agglomeration in the bulk aqueous phase but, at the same time, facilitates a change of the AuNPs ionic and solvation cloud at the L|L interface.

Scheme 1.
Four-electrode electrochemical cell configurations used to prepare the AuNP film at the polarized liquid|liquid (L|L) interface (electrochemical cell 1) and then use the formed interfacial AuNP film to oxidise sulfur biphasically (electrochemical cell 2). PBS is phosphate buffered saline solution at pH 7.4, BACl is bis(triphenylphosphoranylidene)ammonium chloride, BATB is the organic electrolyte salt bis(triphenylphosphoranylidene)ammonium tetrakis(pentafluorophenyl)borate and the organic solvent is α,α,α-trifluorotoluene (TFT). The experiments were carried out either under aerobic or anaerobic conditions, as detailed in the text.
Electrochemistry at polarized L|L interfaces both facilitates AuNP film formation and provides a novel method to monitor and control the AuNP films growth using cyclic voltammetry (CV), see Figure 1A. The adsorption of aqueous anions on the AuNPs gives a well-defined reversible signal between -0.2 V and +0.2 V. With potential cycling, the amount of AuNPs at the interface increases and so too does the adsorption of anions. This electrochemically driven experimental approach yields a homogeneous film of AuNPs and provides an external control of the film thickness ( Figure 1B). The experiment was performed under an argon (Ar) atmosphere as potential cycling in the presence of O2 can slowly oxidise the AuNPs irreversibly (discussed vide infra) and change the active surface where anion adsorption takes place. Cyclic voltammetry (CV) as AuNPs deposit at the L|L interface over 100 potential cycles (solid lines). Every 10 th CV is shown, starting with the first CV cycle, and the arrow shows the evolution with time. The dotted line shows a numerical simulation of the CV using a Frumkin isotherm (repulsive interaction factor of 13.5 kJ mol -1 , total surface charge 17.5 μC cm -2 ). The scan rate used was 25 mV·s -1 . The conditions in the glovebox were Ar (1 bar) and O2 (1.5 ppm). The electrochemical cell used to prepare the AuNP film was as described by electrochemical cell 1 in Scheme 1 with x = 0. (B) Optical image of the AuNP film at the L|L interface after undeposited AuNPs in the bulk aqueous phase were removed by careful sequential washing with 10 mM LiCl. This image represents electrochemical cell 2 in Scheme 1 with x = 0 and X = Cl -.

Physisorption of aqueous anions on the interfacial AuNPs.
The influence of the aqueous anion on the CV response of the interfacial AuNP film was probed by using either 10 mM LiCl or 10 mM LiOH as the aqueous phase under aerobic conditions ( Figure 2A). For both aqueous electrolytes, the near symmetric shape of the CVs and their behaviour with scan rate, showing a linear increase in peak current with scan rate ( Figure S1), indicate an interfacial adsorption/desorption process. Negatively charged AuNPs may attach and detach from the L|L interface with applied potential, but this process is very slow and does not yield symmetric voltammetry signals. [24] Therefore, the CV responses in Figure 2A are attributed to the adsorption of ions on the interfacial AuNPs, specifically OHand Clanions. CV responses for the adsorption of these anions at the bare L|L interface do not give well defined peaks within the polarizable potential window. Figures 2A and 2B show that the adsorption strength of OHis much higher than that for Cl -, mirroring a trend observed for the electroadsorption of these ions on Au(111). [33]  (D) Electrochemical impedance spectroscopy (EIS) spectra at +0.006 V (black circles) and +0.311 V (red squares) for the electrochemical cell described in (C). The solid lines show the fittings using the equivalent circuits depicted. Figure 2A show the results of computed numerical simulations of the CV profiles corresponding to an adsorption process following a Frumkin adsorption isotherm. The simulations solved the differential equation arising from the kinetics of the surface coverage in time upon potential cycling, the Python code is available in the Supporting Information (SI). The numerical simulations show that the maximum charge density associated with OHand Cladsorption during the negative scan are 24 and 12 C·cm -2 , respectively. The simulations also show that the adsorption of anions can be reasonably described by a Frumkin model, where the Gibbs adsorption energy increases with surface coverage, with a repulsive interaction factor of 14.5 kJ·mol -1 . The total maximum charge associated with Cladsorption is higher for AuNP films grown in an Ar atmosphere (17.5 C·cm -2 , Figure 1A) than in air (12 Figure 2A), and the CVs generated in an Ar atmosphere are more symmetric ( Figure   1A, black lines) than those in air ( Figure 2A, black line). The former suggests that the presence of O2 reduces the capacity of AuNPs to adsorb anions. On the other hand, the high value for the interaction factor, and the fact that it is almost the same for OHand Cl -, suggests that after adsorption the anions keep a significant negative charge. Thus, the anion-Au bond must be highly polar and the CVs in Figures 1A and 2A are due to the physisorption of anions, and not to their electro-adsorption. In conventional three-electrode electrochemical cells, it is electroadsorption and not physisorption of anions that is observed.

The dotted lines in
Oxide formation on Au(111) electrodes has been reported to occur when OHadsorption reaches charge densities >35 C cm -2 , meaning that this is the maximum OHcoverage that can be reached on Au. [33] Herein, the maximum coverage of OHreached on the interfacial AuNPs is ca. 24 C·cm -2 . OHis adsorbed more efficiently on a polarized Au electrode as the Fermi level of Au decreases when the OHcoverage increases with applied potential. In other words, the negative charge on adsorbed OHis screened by a net charge transfer from the adsorbed anion to the Au electrode. On the other hand, OHadsorption on the interfacial AuNPs increases their Fermi level. Thus, the AuNPs become more negatively charged, inhibiting the further adsorption of other OHanions.
The influence of the aqueous anion on the measured differential capacitance curves in the presence of the AuNP film was also investigated ( Figure 2B). The capacitances in the presence of LiCl (black dots) and LiOH (red dots) were almost the same at potentials close to the potential of zero charge (PZC) around +0.3 V, confirming that the surface coverage of interfacial AuNPs was the same for both experiments and the AuNPs were stably attached to the interface. Furthermore, the well-defined peaks at potentials lower than the PZC confirm the adsorption of negatively charged anions. To our knowledge, this is first experimental evidence that Cland OHcan be physisorbed, in their anionic form, on AuNPs at a polarized L|L interface.
The transparency of the interfacial AuNP film to ion transfer was investigated by monitoring the tetramethylammonium cation (TMA + ) transfer reaction using CV and electrochemical impedance spectroscopy (EIS) at a AuNP film functionalised L|L interface ( Figure 2C). Considering that TMA + transfers close to the PZC (compare Figures 2B and C), a low influence of the electrical double layer was expected as the impact of migration should be negligible. [9,34] The interfacial AuNP film was found to have no significant effect on the TMA + ion transfer kinetics as demonstrated by (i) the peak separation of the reversible TMA + ion transfer response being 58 mV ( Figure 2C), (ii) the ion transfer resistance being undetectable by EIS experiments (Figure 2D), and (iii) the Randles circuit accurately describing the EIS spectra at the TMA + ion transfer potential of +0.311 V ( Figure 2D).
The influence of scan rate on the signals associated with Cland OHadsorption or TMA + transfer is markedly different ( Figure 2C and Figure S1). The peak current for the Cland OHadsorption is proportional to the scan rate ( Figure S1D) and increases much faster with scan rate than the TMA + transfer signal, which is proportional to the square root of the scan rate (data not shown). Additionally, the EIS spectrum at +0.006 V ( Figure 2D) shows that the AuNP film does not behave like an ideal capacitor and a constant phase element (CPE) with an n value of 0.91 is needed to fit the experimental data. Usually a CPE with an n value lower than 1 is associated with a capacitor comprising a very rough surface. [35] Thus, this indicates that interfacial AuNP accumulation and agglomeration increases the roughness factor of the L|L interface. Previously, Younan et al. showed that the capacitance of the L|L interface increases in the presence of an adsorbed monolayer of citrate coated AuNPs and attributed this to an increase of the interfacial charge density or by an increase of the interfacial corrugation. [36] Adsorption of hydrophobic elemental sulfur on the interfacial AuNPs. The presence of O2 changes the active surface area of the interfacial AuNPs for the adsorption of anions (discussed vide supra when comparing Figures 1A and 2A). To further investigate this effect, the evolution of the CVs using electrochemical cell 2 with LiOH as the aqueous phase, see Scheme 1, under anaerobic conditions was probed ( Figure 3A). Extensive potential cycling clearly decreases the peaks related to OHadsorption, generates a broad new signal between +0.1 V and +0.5 V and leads to the cathodic peak shifting to more positive potentials. Each of these changes are consistent with the CV obtained in air (red line in Figure 3A), suggesting their origin as chemical reactions between O2 and AuNPs upon potential cycling. It seems that O2, even at very low concentrations (1.5 ppm), slowly passivates the interfacial AuNPs surface, possibly by forming a layer of oxide upon extensive potential cycling. The introduction of dissolved elemental sulfur to the organic phase stabilises the CVs when the cell is inside the glovebox, with some signals increasing slightly upon potential cycling ( Figure 3B). Interestingly, the CVs shown in Figure 3B are similar to those reported for aqueous solutions of Na2S in alkaline media on Au electrodes [37][38][39] and, in our opinion, can be rationalized in a similar way. Firstly, previous studies have indicated that both atomic and molecular sulfur can be adsorbed on gold. [38][39][40] Secondly, the shape and behaviour of the CVs with scan rate suggests that the peaks observed inside the potential window arise mainly from surface processes. Thirdly, the asymmetry of the CVs in Figure 3B, where the negative peaks are higher than the positive peaks, suggest that an irreversible chemical reaction is taking place during the potential scans. In this sense, we suggest that the following reactions are taking place in Figure 3B: 2OH − (ad) + S(ad) ⇄ 2OH(ad) + S 2− (ad) 2OH − (ad) + S (ad) ⇄ 2OH(ad) + S 2− (ad) where (ad) means adsorbed on the interfacial AuNPs' surface and (aq) means on the aqueous side of the interface.
The reduction of atomic (S) or molecular (Sx) sulfur adsorbed on the interfacial AuNPs can be promoted by the adsorption of OH -(Equations (1) to (3)), leading to an additional negative peak between -0.15 and +0.2 V in the presence of sulfur ( Figure 3D). The potential windows were reduced in Figures 3C and 3D to highlight (i) a reversible signal due to OHphysisorption is seen in both CVs (red lines) and (ii) an irreversible signal is seen in the presence of either O2 or sulfur (blue lines). The blue signals in Figure 3D can be attributed to a redox reaction between sulfur and adsorbed OH -.
Differential capacitance curves in the absence of sulfur in the organic phase further support the fact that O2 modified the AuNP's surface, with the capacitance under aerobic conditions being higher ( Figure 4A). The AuNP's surface is usually oxide-free under ambient conditions. However, in alkaline media, such as the LiOH aqueous phase used in Figure 4A, gold oxidation proceeds at lower potentials and therefore the formation of gold oxides upon potential cycling leads to higher capacitances. [41] Another possibility for the origin of the gold passivation process can be an irreversible reaction between O2 and unidentified contaminants present in the organic phase. Therefore, we attribute the lower capacitance close to the PZC during the negative scan to the adsorption of atomic and molecular sulfur on the interfacial AuNPs (blue circles, Figure 4B).
At negative or positive potentials, the AuNP's surface is hydrophilic and the adsorption of S8 is unlikely. Furthermore, S8 does not adsorb on the interfacial AuNPs scanning positively as their surface has already been coated in adsorbed OHspecies and is too hydrophilic.
The adsorption of S8 scanning negatively agrees with the observation that the peak between -0.2 V and +0.2 V is higher for the negative scan direction curve (blue circles, Figure   4B). This behaviour may come from pseudo-capacitance currents due to the reduction of  To study the oxidation of sulfur on a well-defined interfacial AuNP film, after the experiment shown in Figure 5A was complete, all suspended AuNPs were removed from the aqueous phase and replaced with pure 10 mM LiCl by sequential washing steps. The CV of this electrochemical cell is shown in Figure 5C (red line) and is identical to the CV shown in Figure 5B (black line). However, a remarkable increase of current over the whole potential window occurs when the LiCl aqueous phase was replaced by LiOH through a further sequential washing step ( Figure 5C, black line). The OHanions significantly enhance the redox reactions that are taking place, which we attribute to low temperature sulfur oxidation at a polarized L|L interface in the presence of interfacial AuNPs. It is also noted that the CV in Figure 5C with LiOH (black line) under aerobic conditions is markedly different to the CVs in Figure 3B obtained under anaerobic conditions, using otherwise identical experimental conditions.
The signals R and O in Figure 5D are related to irreversible processes and grow upon potential cycling. The latter indicates that a redox reaction is taking place and the concentration of reactive intermediaries increases with time. This is the behaviour expected for biphasic S8 oxidation, where intermediaries such as thiosulfate, sulfite and sulfate anions are formed and transferred irreversibly to the aqueous phase. The complexity of the processes taking place at the interface is highlighted by the many signals observed across the full potential window (see insert, Figure 5D), likely due to ion transfer of sulfur oxidation products to the aqueous phase such as polythionates, dithionate, dithionite, thiosulfate, etc. These ions transfer within the available potential window at the polarized L|L interface due to their low charge densities and high solubilities in both the aqueous and organic phases. However, their analytical detection directly at the interface is extremely challenging due to their instabilities' in low temperature aqueous systems with respect to sulfide, sulfate and elemental sulfur species. [42] The role of OHto enhance the catalytic activity of AuNPs towards oxidation processes, such as the enhanced currents attributed to S8 oxidation in Figures 5C and D when LiOH is the aqueous electrolyte, has precedent in the literature. For example, supported AuNPs have a higher catalytic activity for the oxidation of carbon monoxide and alcohols at alkaline rather than acidic pHs. [16] Adsorbed OHplays key roles in three processes critical to the oxidation process: the adsorption of O2 on the AuNPs, as an intermediary and as a source of oxygen atoms. [16] Furthermore, Huang et al. reported that oxygen anions (O2 -) are preferentially adsorbed on AuNPs compared with O2. [43] Thus, the adsorption of OH -, or Cl -, can promote the adsorption of O2 by charging the AuNP's surface negatively. The latter increases the negative charge of O2 upon adsorption. This explanation is the same as that used to understand the adsorption of O2 on ionic solids, where charge transfer from the support to the AuNPs is required to enhance the ORR on gold. [44] Experimental evidence strongly suggests that the adsorption and reduction of O2 on AuNPs is more likely to occur thought proton-coupled electron transfer (PCET) reactions. For example, using surface-enhanced infrared absorption spectroscopy (SEIRAS), Ohta et al.
In this sense, the change of free energy of reaction (4) is: where ∆ is the total change of the chemical potential. Equation (7) shows that interfacial polarization with a negative  o w  makes reaction (4) more likely. The latter leads to an enhanced adsorption of OHon the AuNPs' surface and, thus, an increase in the Fermi level of the interfacial AuNPs, EF(AuNPs). Furthermore, a negative  o w  increases the molar fraction of OH -(aq), OH − (aq) , thereby increasing the chemical potential of OHon the aqueous side of the interface and also increasing the driving force of reaction (4).
Under the approximation of Butler-Volmer equation, the electron transfer kinetic constant, k, for the ORR: is: Taking into account that  o w  affects EF(AuNPs), as discussed due to the adsorption of OH -, it is going to also affect the kinetics of the electron transfer reactions according to Equation (9).
Considering that EF(AuNPs) increases at negative  o w  to favour reduction reactions in the organic phase, and the likely favoured PCET pathway for the ORR on the AuNPs surface, we propose that the irreversible signal R at a negative  o w  in Figure 5D is related to some or all the following reactions: These reactions can be seen in general as the electroadsorption of OHcoupled to O2 reduction.
At positive  o w , EF(AuNPs) lowers sufficiently due to the desorption of OHto favour oxidation reactions in the organic phase. Thus, we propose that the irreversible signal O in Figure 5D is related to sulfur oxidation. The first step of S8 oxidation can be the reaction: However, a more likely scenario is that adsorbed sulfur atoms can be oxidised as follows: interfacial AuNPs in the presence of an organic electron donor species. [22] However, it should be noted that there is no truly definitive experimental evidence in the literature as to whether the ORR mechanism is homogeneous or heterogeneous, or if the ORR takes place in the organic or aqueous phase in the presence of catalysts adsorbed at the L|L interface. [45] The standard reduction potentials and high O2 solubility in organic solvents favour the hydrogen evolution reaction (HER) and ORR, via either the 2-or 4-electron pathway, taking place in the organic phase. [45] However, the concentration of protons in aprotic organic solvents is extremely low. Furthermore, while thermodynamically favoured, the kinetics of the HER and ORR in the organic phase may be slow as electrochemical reactions occur with higher rates in solvents with higher permittivity constants, i.e., aqueous solutions.

Conclusions
In this article, a novel methodology to form an interfacial film of AuNPs at a polarized L|L interface is developed. By employing PBS modified AuNPs, potential cycling yields a homogeneous film of interfacial AuNPs. Electrochemistry at a polarized L|L interface also provides a means of monitoring the adsorption of electrolyte species (herein Cland OHanions) on the interfacial AuNPs and driving catalytic redox reactions at the AuNP's surface (herein S8 oxidation by O2). Modelling of the CV data shows that the adsorption of anions can be reasonably described by a Frumkin model, the anion-Au bond must be highly polar and the anions are physisorbed on the interfacial AuNPs (a new insight). The homogeneous oxidation of S8 at low temperatures is unlikely in a homogeneous phase as the reactants, intermediates and products have different solubility properties. Our biphasic system overcomes these solubility limitations, while simultaneously providing an electrochemically controlled catalytic interface. Importantly, differential capacitance measurements provide evidence that S8 adsorbs on the interfacial AuNPs. The electrochemical signals attributed to S8 oxidation by O2 were enhanced when LiOH was used as the aqueous electrolyte. The OHanions have multiple roles in the mechanism, raising the Fermi level of AuNPs upon adsorption, enhancing O2 adsorption on the AuNPs, and acting as an intermediary and source of oxygen atoms. To our knowledge, this is the first report of an artificial system that can mimic the microbial oxidation of elemental sulfur at ambient conditions.