Selective Production of Oxygen from Seawater by Oxidic Metallate Catalysts.

Although the emphasis of water splitting is typically on hydrogen generation, there is a value in the oxygen byproduct especially for life support in field operations. For such applications, the production of a pure, unadulterated oxygen stream is highly desired under environmental conditions. Here, we demonstrate that self-healing oxygen evolution catalysts composed of cobalt or nickel are capable of selectively producing oxygen from both 0.5 M NaCl solutions and seawater. Differential electrochemical mass spectrometry demonstrates the absence of halogen in the product stream, and chemical analysis shows the production of only minute amounts of hypohalous acid.

Introduction

The oxygen evolution reaction (OER), in which water is anodically converted to oxygen, has unique applications when performed under environmental conditions. Although OER is often considered as an ancillary byproduct to the cathodic hydrogen evolution reaction for the storage of renewable energy,[1−3] the reaction has great value for the generation of breathable oxygen from water, especially for extended autonomous human-based undersea operations.[4] For these applications, it is of interest to generate oxygen directly from seawater[5] and, therefore, the effects of seawater on the long-term stability of the catalyst and the selective generation of oxygen become important considerations. Oxygen production from seawater is particularly challenging because the halide salts present in seawater potentially introduce undesirable oxidation products, most notably Cl2 and Br2. Further complicating the situation is the fact that the thermodynamic standard potentials for these reactions are within a few hundred millivolts of that for water oxidation (Scheme ).[6] Thus, successful water oxidation catalysts operating in seawater must demonstrate kinetic selectivity for water oxidation and minimize ancillary oxidative side-reactions.

Scheme 1

Diagram Showing Standard Reduction Potentials for Redox Reactions Relevant to Seawater Electrolysis

We have developed catalysts that are capable of self-healing,[7] thus allowing water splitting to be performed under environmental conditions. Included among these catalysts are oxidic metallates of manganese,[8−10] cobalt,[11,12] and nickel.[13,14] For cobalt−phosphate (CoPi), we have previously shown these catalysts to be capable of operating in natural waters[15] at faradaic efficiencies near 100%. However, we have yet to assess their selectivity against halogen gas formation. To this end, we now apply the technique of differential electrochemical mass spectrometry (DEMS) to quantify the production of halogen from seawater using CoPi and nickel-borate (NiBi) water-splitting catalysts.

Results and Discussion

DEMS is an experimental technique in which an electrochemical flow cell is interfaced with a mass spectrometer via a gas-permeable membrane[16−18] to allow products generated electrochemically at the working electrode to be detected by mass spectrometry in real time. Our group has previously used DEMS to examine the production of O2 by metallate oxygen-evolving catalysts (M-OECs) as well as to perform isotopic labeling studies to identify the details of OER and hydrocarbon oxidation by M-OEC catalysts.[19,20] As we show here, this technique is particularly valuable for measuring halogen gas production as these gases are easily detected by mass spectrometry.

Chloride is present in seawater at high concentrations (∼0.5 M)[21] and, thus, constitutes the greatest potential source of a toxic byproduct. Figure shows the Pourbaix diagram[25] for aqueous Cl– solutions; the most probable Cl– oxidation products in the potential range needed for O2 generation are Cl2, HClO, and ClO–. Consequently, we undertook studies to investigate the products of water oxidation in the presence of Cl– and to quantitate the production of Cl2, HClO, and ClO–.

Figure 1

Simplified Pourbaix diagram at 25 °C for the aqueous metastable hypochlorites in the chlorine–water system (1 M). Solid black lines delineate regions of stability for chlorine-containing species. Dashed blue lines delineate regions of stability for O2, H2O, and H2, from top to bottom. The purple and green bars indicate the potential ranges over which CoPi and NiBi DEMS experiments were performed, respectively.

To emulate the concentration of NaCl in seawater, a solution of 0.5 M NaCl and 0.1 M buffer (either KPi at pH 7 or KBi at pH 9.2 for CoPi and NiBi, respectively) was prepared. This solution was flowed through the DEMS electrochemical cell at a rate of 60 mL/h with CoPi or NiBi as the catalyst. Cyclic voltammetry was employed for current–time measurements because it enabled us to easily examine potential byproducts over a wide potential range.

Figure shows the results of cyclic voltammetry and DEMS experiments for the CoPi and NiBi catalysts. The redox feature at 0.7 V has been observed for NiBi in the presence of salts and has been assigned to a Ni(II)(OH2)2/Ni(III)O(OH) surface redox event.[23] For both catalysts, no Cl2 production is observed over the entire potential range examined. Although we did not detect any Cl2 in the reaction headspace, it is known that Cl2 will react with water in neutral or basic solutions to form HClO/ClO–.[22] Thus, we performed additional bulk electrolysis experiments to determine whether we were producing appreciable amounts of HClO/ClO–. Electrolysis was performed at an OER overpotential of ηOER = 490 mV for ∼12 h. We employed a spectroscopic detection test to quantify HClO/ClO– in our electrolyte post-electrolysis (see the Experimental Section). We determined that the faradaic efficiencies for HClO/ClO– production were 3.2 ± 1.1 and 0.4 ± 0.1% for CoPi and NiBi, respectively.

Figure 2

Cyclic voltammetry (top panel) and DEMS (middle and bottom panels) experimental data for CoPi (left) and NiBi (right) films operated in an aqueous solution containing 500 mM NaCl and either 100 mM KPi at pH 7 for CoPi experiments or 100 mM KBi at pH 9.2 for NiBi experiments. Cyclic voltammetry was performed at a scan rate of 5 mV/s.

We also examined the selectivity for oxygen generation by CoBi and NiBi catalysts in seawater. Using water collected from Boston Harbor, a solution buffered to pH 9.2 with 100 mM KBi was prepared. We note that we were unable to prepare solutions using KPi as a buffer, as phosphate is incompatible with the Ca2+ ions present in seawater and rapidly precipitates as calcium phosphate. Figure shows that Cl2 was not detected from seawater oxidation over all potentials examined. We note that the NiII(OH)2/NiIIIO(OH) surface redox feature is suppressed is seawater as has been observed previously,[15] and its disappearance is suggestive of surface passivation in seawater. An additional byproduct to consider when electrolyzing seawater is Br2, which can be produced from the oxidation of Br– anions present in natural seawater. Although Br– is present in significantly lower concentrations (∼0.8 mM)[24] than Cl– in saltwater, this concentration difference may be offset by the fact that Br– can be oxidized to Br2 at a lower potential than both Cl– oxidation and OER (Scheme ). However, as Figure reveals, no Br2 was produced upon seawater electrolysis. Analogous to Cl2, Br2 may react with water in neutral or basic solutions to form HBrO/BrO–.[24] Thus, we also tested for HClO/ClO– and HBrO/BrO– production during electrolysis of seawater. Using the same procedure as was used for the NaCl electrolysis, we determined that the faradaic efficiencies for the combined production of HClO/ClO– and HBrO/BrO– were 1.4 ± 0.5 and 4.3 ± 0.1% for CoBi and NiBi, respectively.

Figure 3

Cyclic voltammetry (top panel) and DEMS (middle and bottom panels) experimental data for CoBi (left) and NiBi (right) films operated in seawater collected from Boston Harbor buffered to pH 9.2 with 100 mM KBi. Cyclic voltammetry was performed at a scan rate of 5 mV/s.

The excellent selectivity of these catalysts against Cl2 and Br2 is likely a result of their ability to operate in neutral pH environments, which disfavor the production of Cl2 or Br2 gas.[26] Further, the minute concentrations of HClO/ClO– and HBrO/BrO– produced by these catalysts demonstrate that these catalysts are largely selective against halide oxidation. Two features of these catalysts likely contribute to this selectivity. First, the potential for Cl–/Br– oxidation to Cl2/Br2 is pH-independent, whereas the potential for water oxidation decreases with increasing pH. Thus, the ability of these catalysts to operate in neutral (as opposed to acidic) solutions lowers the thermodynamic potential for water oxidation relative to halide oxidation. Second, and more importantly, O–O bond formation occurs by the reaction of terminal oxygens coordinated to the edge of the CoPi/Bi and NiBi metallate catalysts.[22] The edges of the metallate clusters are exposed by substitution of Bi and Pi with water, as shown in the first step of the reaction cycle depicted in Scheme . Owing to the large disparity in reactant concentrations, H2O can outcompete halide ions for edge sites. Accordingly, we posit that water oxidation is favored kinetically over the oxidation of halide ions.

Scheme 2

Water Oxidation on CoPi Occurs Intramolecularly at the Edges of a Dicobalt Active Site

In conclusion, we have used DEMS to demonstrate that CoPi and NiBi catalysts are capable of generating O2 from seawater at high selectivity by suppressing the oxidation of Cl– and Br–. As the kinetics for oxidation by these catalysts is directed by edge reactivity of the substrate, the inability of halide to compete with water association at the edge sites of the cluster appears to be an important factor for the selective generation of oxygen. We anticipate that these results will be beneficial to applications that target the generation of oxygen using seawater as an oxygen source.

Experimental Section

Materials

Co(NO3)2·6H2O (99.999%) and Ni(NO3)2·6H2O (99.999%) were used as received from Strem. KOH 88% and KH2PO4 99.9% were reagent grade and used as received from Macron. Unless otherwise indicated, all electrolyte solutions were prepared with type I water (EMD Millipore, 18.2 MΩ cm resistivity). NaCl was ACS reagent grade and was purchased from VWR. Boston Harbor water samples were collected by hand from a location in the Fort Point Channel.

General Electrochemical Methods

All electrochemical experiments were conducted using a CH Instruments 760C or 760D bipotentiostat, a BASi Ag/AgCl reference electrode, and a Pt wire counter electrode for steady-state electrochemical measurements. The working electrodes were CoPi and NiBi, as described below. All experiments were performed at room temperature (23 ± 1 °C).

Preparation of Thin Films for Use in DEMS and Bulk Electrolysis Experiments

The CoPi and CoBi thin-film electrocatalysts were prepared via anodic electrodeposition, according to previously reported procedures.[25] The deposition of the CoPi film was carried out in 0.1 M KPi pH 7 electrolyte containing 0.5 mM Co2+. A constant potential was held at 0.85 V (all potentials are referenced to Ag/AgCl unless noted otherwise) for a total of 35 mC/cm2 charge passed. The deposition of the CoBi film was carried out in 0.1 M KBi pH 9.2 electrolyte containing 0.5 mM Co2+. A constant potential was held at 0.75 V for a total of 35 mC/cm2 charge passed. For electrodes used in DEMS experiments, a glassy carbon working electrode fabricated for the DEMS electrochemical cell (diameter 7.2 mm) was used as the substrate for deposition of catalyst films. For films used in bulk electrolysis experiments, fluorine-doped tin oxide-coated glass plates were used as substrates. Kapton tape was applied to these plates such that a 1 cm2 area was exposed to the deposition solution.

The NiBi thin-film electrocatalyst was prepared in a similar manner. The deposition was carried out in 0.1 M KBi, pH 9.2 electrolyte containing 0.5 mM Ni2+. A constant potential was held at 0.95 V for a total of 35 mC/cm2 charge passed. The NiBi film was then anodized by passing a constant current of 3.5 mA/cm2 for 30 min in a 1 M KBi, pH 9.2 solution.

Differential Electrochemical Mass Spectrometry (DEMS) Experimental Setup

DEMS experiments were conducted on a home-designed/assembled DEMS system that has a detection limit of ∼0.1 nmol. The DEMS setup consists of two differentially pumped chambers (ionization chamber and analysis chamber) and a quadrupole mass spectrometer (PrismaPlus QMF 110, Pfeiffer-Vacuum). The ionization and analysis chambers were pumped to high vacuum by two Pfeiffer 65 L/s turbomolecular pumps backed by a Pfeiffer dry diaphragm pump, to avoid contamination by oil vapors. The PrismaPlus quadrupole mass spectrometer was connected to the analysis chamber and equipped with electron multiplier/faraday cup dual detecting units. The time constant of the mass spectrometer was in the millisecond regime. Mass spectrometric data was collected with the Quadera software, with three selective channels recording m/z of 32, 70, and 160 in real-time during the electrochemical cyclic voltammetry (CV) experiments.

A dual thin-layer flow electrochemical cell made of Kel-F was connected to the ionization chamber via an angle valve. The upper chamber, which houses the electrochemical reaction, is connected to the lower chamber, which is under high vacuum and in line with the mass spectrometer, by six capillaries. In the upper compartment, the working electrode is pressed against a ∼100 mm thick Teflon gasket with an inner diameter of 6 mm, leaving an exposed working electrode area of 0.28 cm2 and resulting in an electrolyte volume of ∼3 μL. In the lower compartment, a porous Teflon membrane (Gore-Tex) supported on a stainless-steel frit serves as the interface between the electrolyte and the vacuum. It is pressed against a ∼100 mm thick Teflon gasket with an inner diameter of 6 mm. The Gore-Tex Teflon membrane has a thickness of ∼75 μm, a mean pore size of 0.02 μm, and a porosity of 50%. A leak-free Ag/AgCl reference electrode is connected at the inlet side of the flow cell by insertion into a capillary, and a Pt wire counter electrode is connected in a similar fashion at the outlet of the cell. The electrolyte was purged with argon for 1 h before being transferred to a syringe pump (Harvard Apparatus PHD 2000 Infusion). The flow of the electrolyte flow into the electrochemical cell was then controlled by the syringe pump, at 60 mL/h, which ensured fast transport of the species formed at the electrode to the mass spectrometric compartment where the volatile products were evaporated into the vacuum system of the DEMS via the porous Teflon membrane.

CV experiments were performed using a CH Instruments 760C potentiostat. All CVs were performed using a 5 mV/s scan rate. For experiments in NaCl solutions, five full cycles were performed, whereas for seawater solutions, two full cycles were performed. Good reproducibility between cycles was observed for all experimental conditions. Data from a single representative cycle is given for each experimental condition in Figures and 3.

Quantification of Hypochlorite and Hypobromite

To determine the faradaic efficiency for HClO/ClO– and HBrO/BrO– production, bulk electrolysis was performed for each catalyst/electrolyte combination. For all electrolysis experiments, the reaction was driven at ηOER = 490 mV (1.1 and 0.98 V vs Ag/AgCl for pH 7 and pH 9.2, respectively) for ∼12 h before stopping the electrolysis and analyzing the products.

Following electrolysis, HClO/ClO– and HBrO/BrO– were quantified using the N,N-diethyl-p-phenylenediamine (DPD) method.[26,27] Briefly, for each sample to be analyzed, a solution of 25 mM DPD was prepared by adding 50 μL of a 1 M DPD stock solution to 2 mL of the sample in a quartz cuvette with a 1 cm pathlength. The UV–vis absorption spectrum of the resulting solution was then recorded using a Varian Cary 5000 spectrometer. DPD is oxidized by any HClO/ClO– or HBrO/BrO– present in the solution to form a colored cation radical dye species (shown below). Here, we used absorption at 531 nm as an indicator of the concentration of the dye. Because HClO, ClO–, HBrO, and BrO– all oxidize DPD to the same compound, this assay does not distinguish between these four species. Thus, we report the combined faradaic efficiency for all of these species.Calibration curves were constructed for 100 mM KPi, pH 7.0 solutions (Figure S2) and 100 mM KBi, pH 9.2 solutions (Figure S4), using absorption at 531 nm as an indicator of the combined HClO/ClO– and HBrO/BrO– concentration. Calibration curves were constructed using ClO– but are applicable for the quantification of combined HClO, ClO–, HBrO, and BrO– for the reasons described above.

Faradaic efficiencies were calculated using the following equationwhere Vtot is the total volume of the electrolysis cell, CClO is the combined concentration of HClO/ClO– and HBrO/BrO– as determined by the DPD test, Q is the total charge passed during the bulk electrolysis, and F is Faraday’s constant.