US20210301411A1

US20210301411A1 - Metallic mesh-based gas diffusion electrodes for utilization of sparingly soluble gases in electrochemical reactions with nonaqueous electrolytes

Info

Publication number: US20210301411A1
Application number: US17/211,235
Authority: US
Inventors: Karthish Manthiram; Nikifar Lazouski
Original assignee: Massachusetts Institute of Technology
Current assignee: Massachusetts Institute of Technology
Priority date: 2020-03-26
Filing date: 2021-03-24
Publication date: 2021-09-30
Also published as: WO2021195229A1

Abstract

A system and method for supplying a gas to an electrochemical system is described.

Description

CLAIM OF PRIORITY

This application claims priority to U.S. Provisional Application No. 63/000,458, filed Mar. 26, 2020, which is incorporated by reference in its entirety.

FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

This invention was made with Government support under Grant No. CBET1944007 awarded by the National Science Foundation. The Government has certain rights in the invention.

FIELD OF THE INVENTION

The invention relates to gas diffusion electrodes.

BACKGROUND

Electrochemical transformations in nonaqueous solvents are important for synthetic and energy storage applications. Use of nonpolar gaseous reactants such as nitrogen and hydrogen in nonaqueous solvents can be limited by their low solubility and slow transport. Conventional gas diffusion electrodes can improve transport of gaseous species in aqueous electrolytes by facilitating efficient gas-liquid contacting in the vicinity of the electrode. Conventional gas diffusion electrodes cannot improve the transport in many nonaqueous electrolytes, however, as the hydrophobic interactions necessary for creating gas-liquid contacting are not present in nonaqueous electrolytes. This can lead to flooding of the electrode and low rates for gas utilization when using nonaqueous electrolytes.

SUMMARY

In one aspect, an electrochemical system can include a housing including a chamber, an electrode within the housing, and a gas permeable metal on a surface of the electrode in contact with the chamber.
In another aspect, a method of supplying a gas to an electrochemical system can include contacting a gas with a gas permeable metal on a surface of an electrode in a chamber of a housing. The gas can be a precursor that is converted to a reactive gas by the electrode.
In another aspect, a method of oxidizing or reducing a gas can include contacting a gas with a gas permeable metal on a surface of an electrode. The gas can be a sparingly soluble gas. The sparingly soluble gas can be nitrogen or hydrogen. In certain circumstances, the ammonia can be produced at a Faradaic yield of at least 30% or at least 40%. In certain circumstances, the method can include supplying a pressure of the gas in the chamber to create a pressure differential at the electrode. The method can allow for the use of a sparingly soluble gas as a reagent in chemical reactions.
In another aspect, an electrochemical system can include a first electrode including a housing including a chamber, an electrode within the housing, and a gas permeable metal on a surface of the electrode in contact with the chamber, and a second electrode including a gas inlet to a housing including a gas permeable metal on a surface of an electrode and a first outlet to release a product from the system.
In certain circumstances, the system can include a gas inlet to the housing.
In certain circumstances, the system can include a first outlet of the housing to release a product from the housing.
In certain circumstances, the gas permeable metal can include a porous metal or a metal mesh system. In certain circumstances, each gas permeable metal can include a metal mesh system. In certain circumstances, the metal mesh can include 100, 200, 300, 400 or 500 fibers per inch. In other circumstances, the metal mesh can be asymmetric and include 100, 200, 300, 400 or 500 fibers per inch in one direction and 500, 1000, 1500, or 2000 fibers per inch in a second direction.
In certain circumstances, the gas permeable metal can include openings of between 1 and 200 micrometers, preferably between 2 and 100 micrometers.
In certain circumstances, each gas permeable metal can include openings of between 1 and 200 micrometers, preferably between 2 and 100 micrometers.
In certain circumstances, the gas permeable metal can include metal fibers or a porous metal.
In certain circumstances, at least one gas permeable metal can include metal fibers or a porous metal.
In certain circumstances, the gas permeable metal can include stainless steel, steel, nickel, iron, copper, silver, gold, or platinum.
In certain circumstances, the gas permeable metal can include a catalyst on a surface of the gas permeable metal. For example, the catalyst can include a surface treated with catalytic nanoparticles or catalytic nanoparticles deposited on the surface. In certain circumstances, the gas permeable metal can include a catalyst, for example, a catalytic metal, metal oxide, metal sulfide, or metal phosphide.
In certain circumstances, the gas permeable metal can be exposed to a pressure gradient. In certain circumstances, at least one gas permeable metal can be exposed to a pressure gradient. In certain circumstances, the method can include supplying a pressure of the gas in the chamber to create a pressure differential in the housing.
In certain circumstances, the method can include applying a voltage to the electrode.
Other aspects, embodiments, and features will be apparent from the following description, the drawings, and the claims.

BRIEF DESCRIPTION OF DRAWINGS

FIGS. 1A-1D depict a comparison of gas-liquid interfaces in various electrode setups. FIG. 1A shows gas diffusion through bulk aqueous electrolyte to a flooded electrode. Note that a nonaqueous electrolyte performs analogously in this system. FIG. 1B shows a traditional hydrophobic carbon fiber-based gas diffusion electrode with an aqueous electrolyte. FIG. 1C shows a traditional carbon fiber-based gas diffusion electrode in the absence of hydrophobic repulsion when a nonaqueous electrolyte is used. FIG. 1D shows a metallic support-based gas diffusion electrode for use in nonaqueous electrolytes.

FIG. 2 depicts a graph showing hydrogen oxidation at high rates on platinum-coated stainless steel cloths occurs in THF-based electrolyte at the anode, while nitrogen reduction on lithium-coated stainless steel cloths occurs at the cathode.

FIG. 3 depicts a graph comparing performance enabled by the architectural advances presented here (this work; stars) relative to work done in the literature (all other points). The system described herein achieves record-high rates at relatively high Faradaic efficiencies.

FIG. 4 depicts a schematic showing a box half filled with liquid, including the relative energies of the liquid and gas phases in the box.

FIG. 5 depicts a schematic showing a box half filled with liquid having a wall, including the relative energies of the liquid and gas phases in the box.

FIG. 6 depicts a schematic showing a box half filled with liquid having a wall and a metastable configuration, including the relative energies of the liquid and gas phases in the box.

FIG. 7 depicts a schematic showing a box half filled with liquid having horizontal electrodes at a gas-liquid interface, including the relative energies of the liquid and gas phases in the box.

FIG. 8 depicts a schematic showing a box half filled with liquid having a hydrophobic GDE or a non-hydrophobic GDE, including the relative energies of the liquid and gas phases in the box.

FIG. 9 depicts a schematic showing a box half filled with liquid having a gas pressure gradient, including the relative energies of the liquid and gas phases in the box.

FIGS. 10A-10C depict kinetic and transport considerations for lithium-mediated nitrogen reduction. FIG. 10A shows reactions present in a lithium-mediated catalytic cycle for nitrogen reduction. (FIG. 10B, FIG. 10C) Diffusion limitations observed in electrochemical reactions involving sparingly soluble gases (FIG. 10B) hydrogen and (FIG. 10C) nitrogen in a 1 M LiBF₄, 0.11 M ethanol in tetrahydrofuran electrolyte at flooded platinum and steel electrodes, respectively. The data in FIG. 10B is collected by performing a linear sweep voltammogram at 5 mV s⁻¹. The dashed line in FIG. 10B helps to guide the eye. The solid line in FIG. 10C is a fit of the data to a kinetic-transport model for ammonia production. Error bars in FIG. 10C are one standard deviation of multiple replicates (n≥2).

FIGS. 11A-11D depict structure of a gas diffusion electrode (GDE). FIG. 11A shows a hydrophobic GDE with an aqueous electrolyte, where well-defined gas-liquid contacting exists. FIG. 11B shows a hydrophobic GDE with a nonaqueous electrolyte, where considerable wetting of the carbon fibers occurs, effectively flooding the catalyst. FIG. 11C show a catalyst-coated steel cloth. A lack of significant capillary action and the presence of a non-zero pressure gradient across the cloth prevent complete catalyst flooding. FIG. 11D shows proton donor cycling in a cell with a proton-producing anode.

FIGS. 12A-12D depict efficiency of the steel cloth-based GDEs for the hydrogen oxidation reaction (HOR) and the nitrogen reduction reaction (NRR). FIG. 12A shows a comparison of HOR Faradaic efficiency (FE) of Pt-coated steel cloths (Pt/SSC) and Pt-loaded carbon papers (Pt/C) at different pressure gradients across the GDEs. FIG. 12B shows the effect of pressure gradient across a Pt/SCC on HOR FE at 25 mA cm⁻²applied current density. FIG. 12C shows the production rate of ammonia as a function of applied current density on steel cloth cathodes at a pressure gradient of 1 kPa across the steel cloth. Solution phase ammonia is found in the electrolyte while gas phase ammonia in the acid trap after the cell. FIG. 12D shows the effect of pressure gradient across a steel cloth cathode on FE toward NH₃at 15 mA cm⁻²applied current density. Vertical error bars in FIGS. 12A and 12B represent a combination of uncertainty in HOR quantification and standard deviation between experiments (n≥2), while in FIGS. 12C and 12D they represent one standard deviation between multiple replicates of the same experiments (n≥2). Horizontal error bars in FIGS. 12B and 12D represent the range of pressure gradient values required for gas flow through the SSCs. Raw data can be found in FIGS. 37A-37D and Tables 2-5. The dashed lines in FIGS. 12B and 12D represent the onset of gas breakthrough in the SSC, which is the Laplace pressure. In all experiments, 7.2 C of charge were passed to measure either HOR or NRR FE.

FIGS. 13A-13D depict coupling of electrodes for a sustainable overall reaction. FIG. 13A shows a comparison of continuous ammonia production metrics at ambient conditions between this work and reported highest rates in nonaqueous electrolytes in the literature. See, for example, Lazouski, N., Schiffer, Z. J., Williams, K. & Manthiram, K. Understanding Continuous Lithium-Mediated Electrochemical Nitrogen Reduction. Joule 3, 1127-1139 (2019); Andersen, S. Z. et al. A rigorous electrochemical ammonia synthesis protocol with quantitative isotope measurements. Nature 570, 504-508 (2019); and Tsuneto, A., Kudo, A. & Sakata, T. Lithium-mediated electrochemical reduction of high pressure N₂to NH₃ . J. Electroanal. Chem. 367, 183-188 (1994), each of which is incorporated by reference in its entirety. FIG. 13B shows changes in Faradaic efficiency toward ammonia with different anode chemistries in experiments where 7.2 C of charge were passed at an applied current density of 20 mA cm⁻², with 10 standard cubic centimeters per minute of gas flowing past the electrode, across which the pressure gradient is 1 kPa. Error bars represent the standard deviation of multiple replicates of the same experiment (n≥2). The insets show the anolyte after longer-term continuous operation at 20 mA cm⁻²for one hour at respective anodes. The dark solution (Pt foil) contains poorly-defined tetrahydrofuran (THF) oxidation products, while the clear solution (Pt/SCC) shows few signs of THF oxidation. FIG. 13C shows a schematic of an electrochemical Haber-Bosch (eHB) reactor coupled to a water-splitting reactor. FIG. 13D shows a photograph depicting a model of an eHB reactor coupled to a water electrolyzer, with the reactors highlighted.

FIGS. 14A-14D depict scanning electron microscopy (SEM) images stainless steel cloth electrodes. A Zeiss-Merlin HR-SEM with an HE-SE2 detector was used to collect images. FIG. 14A shows low magnification image of bare stainless steel cloths (SSCs) (FIGS. 19A-19F). FIG. 14B shows medium magnification images of nickel-coated SSCs (FIGS. 20A-20F). FIG. 14C shows high magnification images of smoothly platinum-coated SSCs (FIGS. 21A-21F). FIG. 14D shows high magnification images of roughly platinum-coated SSCs (FIGS. 22A-22F). Note that the HOR FE does not differ between smoothly and roughly platinum-coated SSCs.

FIGS. 15A-15C depict control experiments confirming nitrogen reduction to ammonia. FIG. 15A shows a comparison between the Faradaic efficiency toward ammonia when various gases are fed to the cell. When using N₂with different isotopic compositions, the ammonia yields are practically identical, which is a sign that N₂reduction is responsible for ammonia formation. See, for example, Andersen, S. Z. et al. A rigorous electrochemical ammonia synthesis protocol with quantitative isotope measurements. Nature 570, 504-508 (2019), which is incorporated by reference in its entirety. There is little to no ammonia formed when Ar is used as the feed gas and in the absence of current. Vertical error bars represent the uncertainty in Faradaic efficiency quantification of a single experiment. FIG. 15B shows the amount of ammonia quantified in the base and acid traps used to clean the inlet gas, and the concentration of ammonia in a post-cell acid trap for comparison. FIG. 15C shows unscaled NMR spectra of electrolyte and acid trap solutions. When ¹⁴N₂is used as the feed gas, only a triplet from ¹⁴NH₄ ⁺is detected in both the trap and solution, while both ¹⁵NH₄ ⁺ and ¹⁴NH₄ ⁺ are detected when ¹⁵N₂is fed. ˜92% of the NH₄ ⁺is ¹⁵NH₄, which suggests some ¹⁴N₂contamination in the experiment, as the nominal isotopic content of the ¹⁵N₂is 98%. The peaks shift slightly due to differences in solvent composition (THF-water mixtures). The peak at ˜6.87 is from butylated hydroxytoluene (BHT) found in the THF. The 25 mA experiments were performed by using a 3-compartment cell with a platinum foil anode, while the 20 mA experiments used a cell with no separator between electrolyte compartments and a Pt/SSC anode.

FIG. 16 depicts a linear sweep voltammograms of hydrogen oxidation on Platinum-coated carbon fiber gas diffusion electrodes in 0.5 M H₂SO₄in water. The sweep rate used is 5 mV s⁻¹. Hydrogen oxidation occurs at high rates on hydrophobic platinum-loaded carbon paper (Pt/C) both when gas flows past the GDE (no pressure gradient applied) and when it flows through the GDE in an aqueous solution. Resistive losses account for most of the potential applied. When using software resistance compensation (compensated 85% of 1.7 Ohm resistance that was determined by the software using impedance spectroscopy), the current-potential curve shifts, but does not change shape significantly. The high rates of hydrogen oxidation on GDEs in aqueous solvents demonstrate that it is in fact a change of solvent from aqueous to nonaqueous (FIG. 12A) that leads to poor hydrogen oxidation rates and FEs on carbon fiber GDEs.

FIGS. 17A-17G depict establishment of a gas-liquid boundary in the cell. Panel a shows 20 μL of water on platinum-loaded carbon paper (left, Pt/C) and stainless steel cloth (right, SSC). The water droplet beads up on both materials. Panel b shows 20 μL of 1 M LiBF₄, 0.11 M EtOH, THF (the electrolyte), on Pt/C and SSC. The electrolyte penetrated into the Pt/C and spread through it, while the spreading was finite on steel cloth, demonstrating the lack of capillary action on the SSC. FIG. 17C shows side and front photos of modified cell used to image the gas and electrolyte compartments. A glass slide is used in place of a PEEK back plate and membrane for the gas and electrolyte compartments, respectively. O-rings are used to evenly distribute pressure on the glass slides, and polycarbonate back plates are used to hold the structure together. The compartments are imaged without electrolyte (“no elec”), with half the usual volume of electrolyte (“880 μL”), with the usual volume of electrolyte of 1.75 mL and gas flowing through the electrode (“through”), gas flowing past the electrode at 1 kPa (“flow past”), and under lack of pressure gradient (“no ΔP”). FIG. 17D shows the electrolyte compartment when using an SSC. FIG. 17E shows the gas compartment when using an SSC. FIG. 17F shows the electrolyte compartment when using Δt/C. FIG. 17G shows the gas compartment when using Pt/C.

FIGS. 18A-18 b depict images of various stages of plating platinum onto steel cloths. Note that the cloths have been cut into smaller pieces after metal plating for ease of presentation. FIG. 18A shows steel cloths under glass slides. FIG. 18B shows freely moving steel cloths. Note the brown tint that appears after striking the cloths with nickel, and the darker, black tint that appears after plating platinum onto the nickel.

FIGS. 19A-19F depict scanning electron microscopy (SEM) images of bare stainless steel cloths. A Zeiss-Merlin HR-SEM was used to collect images. FIGS. 19A-19B show low magnification. FIGS. 19C-19D show medium magnification. FIGS. 19E-19F show high magnification. Panels a, c and e used an Inlens detector, while (b, d, f) used an HE-SE₂detector.

FIGS. 20A-20F depict scanning electron microscopy (SEM) images of nickel-coated stainless steel cloths. A Zeiss-Merlin HR-SEM was used to collect images. FIGS. 20A-20B show low magnification. FIGS. 20C-20D show medium magnification. FIGS. 20E-20F show high magnification. FIGS. 20A, 20C and 20E used an Inlens detector, while FIGS. 20B, 20D and 20F used an HE-SE₂detector. Note the uncoated sections at the top of the threads. They are likely formed when one side of the cloth contacting the glass wall in the beaker cell used for nickel striking (i.e. from poor transport of Ni²⁺ to the surface).

FIGS. 21A-21F depict scanning electron microscopy (SEM) images of smoothly Pt-coated stainless steel cloths. A Zeiss-Merlin HR-SEM was used to collect images. FIGS. 21A and 21B show low magnification. FIGS. 21C and 21D show medium magnification. FIGS. 21E and 21F show high magnification. FIGS. 21A, 21C and 21E used an Inlens detector, while FIGS. 21B, 21D and 21F used an HE-SE2 detector. Note the poorly coated regions at the top of the threads. As platinum poorly adheres to stainless steel, they are likely caused by the Ni-less regions observed in Ni plated cloths.

FIGS. 22A-22F depicts scanning electron microscopy (SEM) images of roughly Pt-coated stainless steel cloths. A Zeiss-Merlin HR-SEM was used to collect images. FIGS. 22A and 22B show low magnification. FIGS. 22C and 22D show medium magnification. FIGS. 22E and 22F show high magnification. FIGS. 22A, 22C and 22E use an Inlens detector, while FIGS. 22B, 22D and 22F used an HE-SE2 detector. Regions of poor platinum coating are visible as in FIGS. 21A-21F. The platinum coating here does not adhere to the fibers as well, leading to rougher surfaces. The measured HOR FE is unaffected up to 25 mA cm⁻²of applied current.

FIGS. 23A-23E depict electrochemical characterization of SSC and Pt/SSC electrodes. A 1 M LiBF₄/0.11 M EtOH/THF electrolyte was used in a 3-compartment cell with an Ag/AgCl pseudoreference for all experiments with 100% software resistance compensation. The resistance was measured by PEIS. FIG. 23A shows linear sweep voltammograms of Pt/SSC when H₂or N₂are flowed past the electrode at 10 sccm and a 1 kPa gradient across the electrode. The high oxidation currents visible when H₂is fed to the electrode support the notion that HOR is occurring. FIG. 23B shows impedance spectra of the system in FIG. 23A. FIG. 23C shows linear sweep voltammograms of SSC when N₂or Ar are flowed past the electrode at 10 sccm and a 1 kPa gradient across the electrode. FIG. 23D shows impedance spectra of the system in FIG. 23C. FIG. 23E shows constant potential holds of the system in FIG. 23C. Note that potential overloads in PEIS prevented measurement of additional points at lower frequencies. Linear sweep voltammograms were collected at a sweep rate of 5 mV·s⁻¹.

FIGS. 24A-24B depict hydrogen mass balance-based quantification of HOR FE. A 1 M LiBF₄/0.11 M EtOH/THF electrolyte was used in a 3-compartment cell with a Pt/SSC anode for hydrogen oxidation quantification. The time required for successive bubbles of gas to leave the gas compartment (FIGS. 32A-32D) was recorded for (a) H₂and (b) N₂feed gases before and after application of 25 mA of current. When H₂was used as the feed, the time required for bubbles to evolve decreased when current was applied because H₂was being consumed at the electrode, while no change was observed when N₂was used as the feed. The FE towards HOR was found to be 105±2%, i.e. close to unity.

FIGS. 25A-25D depict the effect of flowrate of feed gas past SSC-based GDEs in electrochemical experiments at an applied current of 25 mA cm⁻². FIG. 25A shows the effect of H₂flowrate on HOR FE. Note that no significant trend in HOR FE is observed when changing the flowrate; the FE is >99% in all cases. FIG. 25B shows the nominal single-pass conversion of H₂as a function of flowrate. At low flowrates (˜0.2 sccm), the conversion of H₂is high; when hydrogen is fed at a rate corresponding to 100% conversion (0.18 sccm), depletion of hydrogen and a lowering of the pressure gradient across the Pt/SSC is observed, so a higher flowrate is required to maintain stable operation. FIG. 25C shows the effect of flowrate on NH₃FE. Note that at lower flowrates, less ammonia is found in the gas phase, while the total amount of ammonia is unchanged with flowrate. FIG. 25D shows the nominal single-pass conversion of N₂as a function of flowrate. The residence time for the gases in the gas compartment range from 8.2 minutes at 0.2 sccm to 10 seconds at 10 sccm. Vertical error bars in FIG. 25A represent the uncertainty in the FE measurement for a single run. Vertical error bars in FIGS. 25B and 25D represent the error in conversion, computed from the uncertainties in flowrate and FE. Vertical error bars in FIG. 25C represent the standard deviation of multiple replicates of the same experiment (n>2). Horizontal error bars represent the uncertainty in gas flowrate.

FIGS. 26A-26D depict use of Pt/SSC for hydrogen oxidation in propylene carbonate-based electrolyte. A 1 M LiBF₄in 9:1 propylene carbonate/dimethyl carbonate was used to demonstrate the efficacy if Pt/SSC for hydrogen oxidation. FIG. 26A shows linear sweep voltammograms (LSVs) collected a sweep rate of 5 mV s⁻¹for flooded and GDE configurations. The respective gases (N₂and H₂) were fed through the electrolyte at 10 sccm for the flooded (Pt foil) case, or past the Pt/SSC at 10 sccm for the GDE (Pt/SSC) case. FIG. 26B shows the same data as depicted in FIG. 26A, but with a larger range of current values shown to demonstrate the high H₂oxidation current obtained when using a Pt/SSC. Higher potentials (and currents) were not possible to apply using the VMP3 potentiostat, as the total cell voltage exceeded its operating range (˜10 V). FIGS. 26C-26D show time intervals between gas bubbles leaving the gas compartment of the 3-compartment cell used in Pt/SSC experiments with a propylene carbonate based-electrolyte when (FIG. 26C) H₂or (FIG. 26D) N₂was used as the feed gas. The computed Faradaic efficiency towards H₂oxidation is 112±19%, while it is 5±5% when N₂is fed to the anode.

FIGS. 27A and 27B depict potentials and Faradaic efficiencies toward NH₃in long duration experiments when using an eHB reactor. FIG. 27A shows the total cell voltage required for a 20 mA constant current in experiments utilizing a SCC cathode and a Pt/SSC anode with various separators. FIG. 27B shows the Faradaic efficiency toward NH₃in the aforementioned experiments. Vertical error bars for the Daramic 6 minute experiment in FIG. 27B represent the standard deviation of multiple replicates of the same experiment (n=3). The Faradaic efficiency toward NH₃decreases from ˜38% in short duration experiments to ˜20% in longer duration experiments, but is fairly independent of the separator used. When a separator (Celgard or Daramic) between the anode and cathode compartments (FIGS. 31A-31H) is used, the cell voltage increases with time for unknown reasons. Possible reasons include selective depletion of Li⁺ ions in the catholyte and separator fouling. In an undivided cell, no separator is used, which means that the electrolyte is free to convect between the electrodes.

FIG. 28 depicts accumulation of ammonia over time in the electrolyte when using an SSC. The amount of ammonia increases monotonically with time when 20 mA cm⁻²is applied to a cell with an SSC cathode after an induction period of approximately 2 minutes. The short induction period suggests that the lithium-mediated catalytic cycle reaches steady state rapidly. The amount of ammonia produced is estimated by diluting small fractions of the catholyte during operation, as described in Nitrogen reduction experiments—time evolution of ammonia. The total amount of ammonia produced in this experiment was measured after the experiment by utilizing the entire catholyte and was found to be 14.5±1 μmol. Vertical error bars represent a combination of estimates of the error associated with ammonia quantification and electrolyte extraction from the cell. In this experiment, the production rate and Faradaic efficiency for ammonia are computed from the slope of the linear fit of data.

FIGS. 29A-29B depict electrical energy losses in the ammonia production system. FIG. 29A shows electrical energy losses at high FE conditions (15 mA cm²applied current, 0.5 kPa pressure gradient across steel cloth). FIG. 29B shows electrical energy losses at high rate conditions (25 mA cm⁻²applied current, 1 kPa pressure gradient across steel cloth). Note that the anode reaction is assumed to be THF oxidation, as was used in the experiments, hence it is somewhat inefficient. Most of the electrical losses are due to high solution resistance. The procedure to calculate these parameters was described below.

FIGS. 30A-30B show depictions of the 3-compartment cell used for GDE experiments. FIG. 30A depicts a 3D model of the 3-compartment cell used in GDE experiments. Note: O-rings are not shown. FIG. 30B shows a photograph of the parts used to assemble a 3-compartment cell. The cell is made of polyether ether ketone (PEEK) plastic. The cell is inspired by a design used in the CO₂reduction literature. The cells were machined in-house; CAD and CAM files are available upon request.

FIGS. 31A-31H depict an assembly of a 3-compartment cell in order. The procedure is described above in Assembly of a 3-compartment cell. FIG. 31A shows the counter electrode and current collector, here a piece of platinum foil and aluminum foil, respectively. FIG. 31B shows the counter electrolyte compartment. FIG. 31C shows the Daramic polyporous separator. FIG. 31D shows the working electrolyte compartment. FIG. 31E shows the gas diffusion electrode, here a stainless steel cloth. FIG. 31F shows the aluminum current collector for gas flow. FIG. 31G shows the gas compartment. FIG. 31H shows the sealed cell.

FIGS. 32A-32D depict a pressure control setup used in gas diffusion electrode experiments. FIG. 32A shows prior to addition of electrolyte to cell, no pressure gradient across the GDE/SSC is observed as gas is passed through the GDE/SSC. FIG. 32B shows following addition of electrolyte, the pressure in the gas compartment increases until the pressure gradient across the GDE/SSC reaches the Laplace pressure, after which gas continues to pass through the GDE/SSC. FIG. 32C shows the pressure in the gas compartment is lowered, and gas begins to flow past the GDE, while the electrolyte stays in the working compartment. FIG. 32D shows in NRR experiments, a boric acid trap between the gas compartment and burette is added to capture gas phase ammonia.

FIG. 33 depicts absorbance spectra of diluted samples assayed by the salicylate method. Calibration absorbance spectra for solutions containing 0 and 60 μM of ammonia are shown. The absorbance spectra of catholyte diluted to 100 mL and further diluted 2- and 4-fold, as described in the methods, are shown. The measured ammonia concentrations of the diluted solutions are given explicitly, while the average concentration assumes that the concentrations were multiplied by their respective dilutions. The experiment was chosen at random; in this experiment, 15 mA of current were applied for a total of 7.2 C of charge, N₂was flowed past an SSC electrode in a 3-compartment cell at 10 sccm, and the pressure gradient across the SSC was 1.5 kPa. The results of this experiment can be found on data line 9 of Table 5.

FIGS. 34A-34D depict typical calibration curves for quantifying ammonia using the indophenol method. Note that the difference between the absorbance at 650 nm and 475 nm, called the absorbance signal, is used to make the calibration curves. FIG. 34A shows absorbance spectra for various concentrations of NH₃in pure water. FIG. 34B shows the resulting calibration curve for NH₃in pure water. FIG. 34C shows absorbance spectra for various concentrations of NH₃in water containing 5% of 1 M LiBF₄/THF electrolyte by volume. FIG. 34D shows the resulting calibration curve for NH₃in water containing 5% v/v 1 M LiBF₄/THF electrolyte. Vertical error bars in FIGS. 34B and 34D denote the standard deviation in absorbance measured between two solutions of equal concentration. Horizontal error bars in FIGS. 34B and 34D denote the uncertainty in ammonia concentration when preparing solutions. Note that the effective extinction coefficient is lower in the presence of electrolyte than in pure water; the ammonia concentration is typically underestimated because the pure water calibration curve is used while samples contain some electrolyte.

FIGS. 35A-35C depict comparison of NMR and the salicylate assay for ammonia quantification. FIG. 35A shows NMR spectra of 3 ammonia calibration solutions with varied NH₄ ⁺ concentrations in simulated acidified electrolyte (1.75 mL of 1 M LiBF₄/0.11 M EtOH/THF electrolyte, diluted to 4 mL with 0.05 M H₂SO₄) and a sample of unknown concentration with 1 mM maleic acid as an internal standard. The spectra were referenced to the maleic acid peak, which was chosen to have a chemical shift of 6.37 ppm. The solution of unknown concentration was obtained by running applying 15 mA to a 3-compartment cell with an SSC cathode through which 10 sccm of N₂was flowed. FIG. 35B shows the relative peak areas of the NH₄ ⁺peaks and the maleic acid peak. FIG. 35C shows measured ammonia concentrations via the salicylate assay and the relative intensities of NMR peaks. The measured concentrations from the two methods are practically identical if maleic acid is used as an internal standard. Using BHT in the THF as the internal standard predicts slightly lower concentrations with large error, likely due to the smaller concentration of BHT protons in the sample solutions.

FIGS. 36A-36D depict ferrocenium calibration curves for quantifying HOR Faradaic efficiency. FIG. 36A shows absorbance spectra for solutions of ferrocenium in water at a range of concentrations. FIG. 36B shows absorbance spectra obtained for solutions involved in HOR control experiments. One control experiment involved using N₂instead of H₂as the feed gas, while the others involved applying no current for the duration of the experiment. FIG. 36C shows calibration curves for the 255 nm ferrocenium signal using differently prepared solutions. FIG. 36D shows calibration curves for the 619 nm ferrocenium signal using differently prepared solutions.

FIGS. 37A-37D depict efficiency of the steel cloth-based GDEs for HOR and NRR with raw data shown. FIG. 37A shows comparison of HOR Faradaic efficiency (FE) of Pt-coated steel cloths (Pt/SSC) and Pt-loaded carbon papers (Pt/C) in various configurations. FIG. 37B shows the effect of pressure gradient across a Pt/SCC on HOR FE at 25 mA cm⁻²applied current density. FIG. 37C shows the production rate of ammonia as a function of applied current density on steel cloth cathodes at pressure gradient of 1 kPa across the steel cloth. Solution phase ammonia is found in the electrolyte while gas phase ammonia in the acid trap after the cell. FIG. 37D shows the effect of pressure gradient across a steel cloth cathode on FE toward NH₃at 15 mA cm⁻²applied current density. Horizontal error bars in FIGS. 37B and 37D represent the range of pressure gradient values required for gas flow through the SSCs. The dashed lines in FIGS. 37B and 37D represent the onset of gas breakthrough in the SSC, which is the Laplace pressure. In all experiments, 7.2 C of charge were passed to both measure HOR or NRR FE.

FIGS. 38A-38D depict methods used to obtain accurate current and potential measurements when using a Tekpower DC power source. FIG. 38A shows a schematic of a circuit to measure current and charge passed. This scheme was used to quantify the charge passed in all experiments. FIG. 38B shows a schematic of a circuit to safely measure and record the total potential applied to the cell. This scheme was never actually used to record the potential, as the DC power source has a digital voltage readout (FIG. 13D). FIG. 38C shows a typical calibration of a resistor resistance using a VMP3 potentiostat. FIG. 38D shows an example measurement of the voltage drop across the resistor in a 15 mA NRR experiment and the resulting charge passed.

FIGS. 39A-39D depict evidence of strong Li⁺—NH₃interactions. FIG. 39A shows a comparison of ammonia concentrations in ammonia-saturated THF and ammonia-saturated 1 M LiBF₄in THF. FIG. 39B shows visual differences between ammonia-saturated THF with and without LiBF₄. When LiBF₄is present, the solution separates into two phases. It is speculated that the two phases may be less dense ammonia-saturated THF on top and more dense [Li(NH₃)x] [BF₄] on the bottom. FIG. 39C shows results of stripping ˜20 mM ammonia in THF with N₂at 10 sccm for 10 minutes. Almost all ammonia left the solution, and was caught in the 0.1 M H₃BO₃trap. FIG. 39D shows results of stripping ˜20 mM ammonia in 1 M LiBF₄in THF with N₂at 10 sccm for 10 minutes. Little ammonia (˜2%) was stripped into the trap, demonstrating the ability of LiBF₄to prevent ammonia stripping. Vertical error bars in FIGS. 39A, 39B and 39D represent the standard deviation of computed concentrations of sample solutions; the concentrations were computed by using various dilutions of the concentrated sample solutions (n≥2).

FIG. 40 depicts the demonstration of facile ferrocene oxidation in the electrolyte. A linear sweep voltammogram (LSV) measured at a 5 mV s⁻¹sweep rate for an electrolyte containing 10 mM ferrocene, 0.11 M EtOH, 1 M LiBF₄in THF while flowing 10 sccm of N₂in a 2-compartment cell at a platinum anode. While thermodynamically ferrocene oxidation is preferred to THF oxidation, it is important to demonstrate that this is accurate kinetically. Here, it can be seen that ferrocene oxidation starts to occur at potentials slightly below 0 V vs Fc⁺/Fc due to a Nernstian shift, after which it quickly reaches the transport-limited current density. THF oxidation occurs at higher potentials; the exact value of THF oxidation was found to be sensitive to platinum anode preparation and age, as well as electrolyte composition. The data were smoothed by using a 50th percentile smooth filter.

FIG. 41 depicts an electrochemical system.

FIG. 42A depicts a basic standalone electrode architecture. FIG. 42B depicts a basic standalone electrode architecture diagram with see-through edges.

FIG. 43 depicts a flow-in configuration of the standalone electrode.

FIG. 44 depicts a flow-past configuration of the standalone electrode.

FIG. 45 depicts an exemplary electrochemical reaction system utilizing a flow-in and a flow-past standalone GDE electrodes.

FIGS. 46A-46B depict linear sweep voltammograms obtained at 20 mV s-1 when using standalone electrodes for in a flow-in configuration hydrogen oxidation in (FIG. 46A) aqueous electrolyte (FIG. 46B) nonaqueous, acetonitrile electrolyte. Applied potentials are IR compensated.

DETAILED DESCRIPTION

A method to utilize sparingly soluble gases in electrochemical reactions at high rates in nonaqueous solvents is described. The method can be relevant for electroorganic synthesis and fuel production where control of proton activity is important. The method relies on metallic supports and a pressure gradient applied across the electrode. The method can be used in a variety of electrochemical systems, for example, as applied herein to hydrogen oxidation in two nonaqueous solvents and nitrogen reduction in one solvent; the two chemistries are coupled to produce ammonia from nitrogen and hydrogen at high rates.
A sparingly soluble gas, generally, is a non-polar gas that does not react or interact favorably with a solvents. For example, N₂, H₂, CO, and CH₄are sparingly soluble. Gasses with solubilities less than 50-100 mM at 1 atm can be considered to be a sparingly soluble gas. A gas with a Henry's constant <0.05 M/atm can be considered to be sparingly soluble gas.
An electrochemical system can include a housing including a chamber, an electrode within the housing, and a gas permeable metal on a surface of the electrode in contact with the chamber. The system can be used in a method of supplying a gas to an electrochemical system can include contacting a gas with a gas permeable metal on a surface of an electrode in a chamber of a housing. The method can include applying a voltage to the electrode. The system can be a gas diffusion electrochemical system, in which metallic supports can be used in the GDEs combined with a pressure gradient across the GDE. The gas can be at a higher pressure relative to the liquid, to obtain effective gas-liquid contacting at the electrode surface for high reaction rates in nonaqueous solvents. Metallic supports can avoid flooding of the electrodes in the absence of hydrophobic repulsion, while the pressure gradient helps maintain the gas-liquid interface in the desired location. As the method relies on a physical effect (a pressure gradient) to establish the gas-liquid boundary, it can be used with any solvent, including nonaqueous solvents.
The gas permeable metal can be a metal support, which can include metal fibers or a porous metal. The metallic support can be stainless steel, which was be woven into a fine cloths with very thin fibers. In one example, the cloth can be a 400×400 mesh, which contains 400 fibers 25 micrometers in diameter per inch of length, with a spacing of approximately 25-40 micrometers between fibers. The gas permeable metal can include openings of between 1 and 200 micrometers, preferably between 2 and 100 micrometers. The metallic supports can be made from any metal that is amenable to forming, including, but not limited to stainless steel (304 and 316), steel, nickel, iron, copper, silver, gold, or platinum. The metals can be formed into porous materials which are gas permeable, such as metal cloths and meshes, but also metal filters and sponges. The characteristic pore size of the material can be at least as large as 200 micrometers, and down to 2 micrometers; the pore size can be smaller if larger operating pressure gradients are desired. In certain circumstances, the gas permeable metal can include a metal mesh system. The metal mesh can be symmetric or asymmetric. In certain circumstances, the metal mesh can include 100, 200, 300, 400 or 500 fibers per inch. In other circumstances, the metal mesh can be asymmetric and include 100, 200, 300, 400 or 500 fibers per inch in one direction and 500, 1000, 1500, or 2000 fibers per inch in a second direction. In certain circumstances, the gas permeable metal can be exposed to a pressure gradient. In certain circumstances, the method can include supplying a pressure of the gas in the chamber to create a pressure differential in the housing. The pressure gradient applied across the cloth can be 0.5 to 10 kilopascals, for example, 1, 2, 3, 4 or 5 kilopascals. The pressure gradient can depend on the electrolyte used and the pore size of the support. At high pressure gradients, above the Laplace pressure of the material, gas may cross the support and enter the electrolyte; the invention still works even under these operating conditions.
The gas permeable metal can include additional catalysts grown, placed, or deposited on a surface of the metal. Potential catalysts can include: metals, such as silver, gold, platinum, nickel, lithium, zinc, or titanium; metal oxides, such as iridium oxide, cobalt oxide, iron oxide, copper oxide, titanium oxide, or silver oxide; metal sulfides, such as molybdenum sulfide, or cadmium sulfide; metal nitrides, such as lithium nitride, cobalt nitride, nickel nitride, and mixtures thereof; metal phosphides, such as cobalt phosphide, nickel phosphide, or mixtures thereof; molecular catalysts, such as metal phthalocyanines, such as cobalt phthalocyanine or metal porphyrins.
Catalysts can be deposited onto the metal substrate or synthesized on its surface. Methods for deposition include electroplating of metals, electroless plating of metals, electrophoretic deposition, sputtering, pulse laser deposition, chemical vapor deposition, spin-coating, or application of catalyst inks. Methods for in-situ manufacture include oxidation (for making oxides), treatment with nitrogen and ammonia (for making nitrides), heating with sulfur (for making sulfides), heating with phosphorus (for making phosphides), or thermal decomposition of complex materials.
Referring to FIG. 41, gas diffusion electrochemical system 10 includes a housing 12, a first electrode 16, and a second electrode 20. A voltage can be applied to the first electrode and the electrode. One or more of the first electrode or the second electrode, or both, can include a catalyst composition. Substrate inlet 14 can be used to introduce a gas species that will be oxidized or reduced, such as nitrogen, oxygen, hydrogen, carbon monoxide or carbon dioxide gas into the housing 12. The gas species in housing 12 can be pressurized relative to the electrolyte fluid 25. The first electrode 16 can include a gas permeable metal configured to contact with the gas species. The half-reaction taking place in housing 12 involves the gas species. The second electrode 20 in housing 18 is opposite first electrode 16. The second electrode 20 can, optionally, include a second catalyst composition. A separator 30 can be present between first electrode 16 and second electrode 20. An electrolyte fluid 25 can be positioned between the first electrode 16 and the second electrode 20. The electrolyte fluid can include an aqueous solvent or a non-aqueous solvent or a water-containing non-aqueous solvent. Depending on the circumstances, the non-aqueous solvent can include 0%, 0.2%, 0.5%, 1.0%, 5%, or 10% water. The solution can flow through housing 12, including the electrochemical product, can be carried out of outlet 40. A half-reaction taking place in housing 18 to generate a second electrochemical product can involve a second gas species that enters housing 18 at port 15 and the product can exit the housing 18 through port 30. Housing 12 and housing 18 can be separate structures or can form a single structure. A pressure gradient can exist between the gas-containing housing, housing 12, and the electrolyte fluid 25. A pressure gradient can exist between the gas-containing housing, housing 18, and the electrolyte fluid 25.
The gas species can be a gas species that can be oxidized or reduced, for example, N₂, O₂, H₂, CO or CO₂.
The solvent can be an inert organic solvent that in which the electrolyte salt, substrate, and proton carrier can be dissolved. In certain circumstances, a carbonylation reaction or reductive amination reaction can involve a substrate dissolved in the solvent. The concentration of the proton carrier can be 5 mM, 10 mM, 15 mM, 20 mM, 25 mM, 30 mM, 35 mM, 40 mM, 50 mM, 60 mM, 70 mM, 80 mM, 90 mM, 100 mM, 200 mM, 250 mM, 300 mM, 350 mM, 400 mM, 450 mM, or 500 mM. For example, for nitrogen reduction, the concentration of the proton carrier can be 50 mM or higher. The concentration of the substrate can be 5 mM, 10 mM, 15 mM, 20 mM, 25 mM, 30 mM, 35 mM, 40 mM, 50 mM, 60 mM, 70 mM, 80 mM, 90 mM, 100 mM, 200 mM, 250 mM, 300 mM, 350 mM, 400 mM, 450 mM, or 500 mM. For example, for nitrogen reduction, the concentration of the substrate can be 50 mM or higher. The concentration of the electrolyte can be 0.1 M, 0.25 M, 0.5 M, 1 M, 2 M, 4 M, 5 M, 6 M, 7 M, 8 M, 9 M, or 10 M.
The temperature and pressure can be ambient temperature and pressure. There can be a pressure gradient between housing 12 and electrolyte fluid 25 and housing 18 and electrolyte fluid 25.
The reaction product can be produced in a gas phase.
The voltage can be between about 0.2V and 40.0V, between about 0.4V and 35.0V, or between about 1V and 30.0V. For example, the voltage can be about 1.0V, 5.0V, 10.0V, 15.0V, 20.0V, 25.0V, 30.0V, 35.0V, or 40.0V.
Each of the first electrode and the second electrode can be or can include a noble metal, for example, platinum or palladium.
A variety of reactor designs can implement the method. The method described herein can be performed under various different electrochemical cell geometries and configurations, which include an anode and cathode, connected to an external power source, with an ionically conductive medium between the two electrodes. A third reference electrode may be incorporated if necessary for control of the potential at the electrodes. Resistive losses can be reduced by decreasing the distance between electrodes. The process may be conducted under batch or continuous conditions. An ionically conductive membrane, such as Nafion or Selemion, or a separator, such as Celgard or Daramic, can be used in the structure, but is not required.
The overall reaction may be tuned by choice of the cathode and the reactor conditions. For instance, if nitrogen is flowed to the cathode, then nitrogen will be reduced to generate ammonia; if hydrogen is flowed to the anode, then hydrogen will be reduced to generate protons.
The nonaqueous solvent can include acetonitrile, DMSO (dimethyl sulfoxide), DMF (dimethylformamide), THF (tetrahydrofuran), DCM (dichloromethane), and propionitrile. The electrolyte can contain a conductive salt such as TBABF₄(Tetrabutylammonium tetrafluoroborate), TBAPF₆(Tetrabutylammonium hexafluorophosphate), NaClO₄(Sodium perchlorate), LiClO₄(Lithium perchlorate), or TEAP(tetraethylammonium perchlorate), or a combination thereof. The non-aqueous solvent can assist substrate solubility.
The electrodes can include a catalyst. Catalysts that may be used in these GDEs include metals such as alkali metals such as lithium, sodium, potassium, alkali-earth metals such as magnesium, transition metals such nickel, platinum, copper, gold, silver. Metallic catalysts can be deposited onto the supports electrochemically from solution, via electroless plating, or sputtered onto the supports ex situ. Catalyst nanoparticles such as metal oxides, metal nitrides, and metal sulfides can be deposited onto the supports via drop casting, sputtering, or pulse laser deposition. For nitrogen reduction, lithium metal can be deposited in situ electrochemically. For hydrogen oxidation, nickel can be deposited electrochemically onto stainless steel cloths, onto which platinum is then electrochemically deposited.
Electrochemical transformations in nonaqueous solvents are important for synthetic and energy storage applications. Use of nonpolar gaseous reactants such as nitrogen and hydrogen in nonaqueous solvents is limited by their low solubility and slow transport. Conventional gas diffusion electrodes improve transport of gaseous species in aqueous electrolytes by facilitating efficient gas-liquid contacting in the vicinity of the electrode. Their use with nonaqueous solvents is hampered by the absence of hydrophobic repulsion between the liquid phase and carbon fiber support. Herein, a method to overcome transport limitations in tetrahydrofuran is reported using a stainless steel cloth-based support for ammonia synthesis paired with hydrogen oxidation. An ammonia partial current density of 8.7±1.5 mA cm⁻²and a Faradaic efficiency of 35±6% are obtained using a lithium-mediated approach. Hydrogen oxidation current densities up to 25 mA cm⁻²are obtained in two nonaqueous solvents with nearly unity Faradaic efficiency. The approach can be applied to produce ammonia from nitrogen and water splitting-derived hydrogen.
Electrochemical synthesis of chemicals is an attractive alternative approach to traditional thermochemical methods. In some reactions, electric potential can act as a thermodynamic driving force instead of high temperatures and pressures, which may allow for operation at milder conditions and in a modular fashion. See, for example, Schiffer, Z. J. & Manthiram, K. Electrification and Decarbonization of the Chemical Industry. Joule 1, 10-14 (2017); and Yan, M., Kawamata, Y. & Baran, P. S. Synthetic Organic Electrochemistry: Calling All Engineers. Angew. Chemie Int. Ed. 57, 4149-4155 (2018), each of which is incorporated by reference in its entirety. Ammonia (NH₃) production is an example of a reaction that may benefit from being operated electrochemically. See, for example, Chen, J. G. et al. Beyond fossil fuel-driven nitrogen transformations. Science (80). 360, eaar6611 (2018); and Soloveichik, G. Electrochemical synthesis of ammonia as a potential alternative to the Haber-Bosch process. Nat. Catal. 2, 377-380 (2019), each of which is incorporated by reference in its entirety. NH₃is currently produced predominantly via the Haber-Bosch process, which operates at high temperatures (300-500° C.) and pressures (200-300 bar) and requires a coupled steam reforming plant for hydrogen (H₂) production. See, for example, Shipman, M. A. & Symes, M. D. Recent progress towards the electrosynthesis of ammonia from sustainable resources. Catal. Today 286, 57-68 (2017), which is incorporated by reference in its entirety. This leads to high capital costs for the process and centralization of production, a situation that is poorly matched with the distributed nature of ammonia utilization. See, for example, Comer, B. M. et al. Prospects and Challenges for Solar Fertilizers. Joule 3, 1578-1605 (2019) which is incorporated by reference in its entirety. Alternative methods for producing hydrogen, such as water splitting, may overcome some of the issues associated with the traditional Haber-Bosch process, such as the large amount of CO₂emissions and high capital cost associated with steam reforming. See, for example, Suryanto, B. H. R. et al. Challenges and prospects in the catalysis of electroreduction of nitrogen to ammonia. Nat. Catal. 2, 290-296 (2019), which is incorporated by reference in its entirety. However, these methods do not overcome the need for large scales for the ammonia synthesis reactor itself, as it must still be run at high temperatures and pressures. An electrochemical process—even one which utilizes multiple reactors—that produces ammonia from nitrogen and water requires a thermodynamic minimum potential of 1.17 V at standard conditions. See, for example, Soloveichik, G. Electrochemical synthesis of ammonia as a potential alternative to the Haber-Bosch process. Nat. Catal. 2, 377-380 (2019), which is incorporated by reference in its entirety. Potential is a potent thermodynamic driver, providing mild conditions conducive to modular and small-scale operation of electrochemical processes. See, for example, Foster, S. L. et al. Catalysts for nitrogen reduction to ammonia. Nat. Catal. 1, 490-500 (2018), which is incorporated by reference in its entirety.
Despite the attractiveness of electrochemistry in synthetic applications, several challenges must be overcome to allow for efficient scale-up of the technology. One of the most important issues is the use of non-renewable reactants at the counter electrode. For reductive chemistries, which are important for energy storage and certain synthetic applications, the counter reaction is often oxidation of solvent or sacrificial anodes made of active metals. See, for example, Jiao, F. & Xu, B. Electrochemical Ammonia Synthesis and Ammonia Fuel Cells. Adv. Mater. 31, 1805173 (2019); Davis, S. J. et al. Net-zero emissions energy systems. Science (80). 360, eaas9793 (2018); Liu, X., Jiao, Y., Zheng, Y., Jaroniec, M. & Qiao, S.-Z. Building Up a Picture of the Electrocatalytic Nitrogen Reduction Activity of Transition Metal Single-Atom Catalysts. J. Am. Chem. Soc. 141, 9664-9672 (2019); Peters, B. K. et al. Scalable and safe synthetic organic electroreduction inspired by Li-ion battery chemistry. Science (80-.). 363, 838-845 (2019); Matthessen, R., Fransaer, J., Binnemans, K. & De Vos, D. E. Electrocarboxylation: towards sustainable and efficient synthesis of valuable carboxylic acids. Beilstein J. Org. Chem. 10, 2484-2500 (2014); and Motile, S. et al. Modern Electrochemical Aspects for the Synthesis of Value-Added Organic Products. Angew. Chemie-Int. Ed. 57, 6018-6041 (2018), each of which is incorporated by reference in its entirety. In aqueous systems, solvent oxidation is permissible and often desired. However, oxidation of organic solvents or sacrificial anodes decreases the atom economy of reactions greatly and makes processes utilizing these reactions poorly amenable to continuous operation. See, for example, Matthessen, R., Fransaer, J., Binnemans, K. & De Vos, D. E. Electrocarboxylation: towards sustainable and efficient synthesis of valuable carboxylic acids. Beilstein J. Org. Chem. 10, 2484-2500 (2014), which is incorporated by reference in its entirety.
In the context of nitrogen reduction, one method that produces ammonia at high rates and Faradaic efficiencies is the lithium-mediated approach. The approach involves reacting lithium metal with nitrogen to form lithium nitride, a reaction which is spontaneous at ambient conditions. The lithium nitride is then protonated to make ammonia and a lithium salt. The lithium salt is electrochemically reduced to lithium metal to close the catalytic cycle (FIG. 10A). The efficacy of the chemistry has been demonstrated in batch processes in which the aforementioned reactions are run with temporal separation; mostly they differ in the method used to generate lithium metal. See, for example, McEnaney, J. M. et al. Ammonia synthesis from N₂and H₂O using a lithium cycling electrification strategy at atmospheric pressure. Energy Environ. Sci. 10, 1621-1630 (2017); Kim, K., Chen, Y., Han, J.-I., Yoon, H. C. & Li, W. Lithium-mediated ammonia synthesis from water and nitrogen: a membrane-free approach enabled by an immiscible aqueous/organic hybrid electrolyte system. Green Chem. (2019) doi:10.1039/C9GC01338E; Kim, K. et al. Electrochemical Synthesis of Ammonia from Water and Nitrogen: A Lithium-Mediated Approach Using Lithium-Ion Conducting Glass Ceramics. ChemSusChem 11, 120-124 (2018); and Kim, K. et al. Lithium-Mediated Ammonia Electro-Synthesis: Effect of CsClO 4 on Lithium Plating Efficiency and Ammonia Synthesis. J. Electrochem. Soc. 165, F1027-F1031 (2018), each of which is incorporated by reference in its entirety. While these processes demonstrate high Faradaic efficiencies, they are not directly amenable to continuous ammonia production, though approaches to utilize rotating reactors for a pseudo-continuous process have been proposed. See, for example, McEnaney, J. M. US20180029895A1, which is incorporated by reference in its entirety. In this regard, continuous processes in which all three reactions happen simultaneously are attractive. Typically, continuous processes utilize a lithium salt in tetrahydrofuran (THF) electrolyte with ethanol as a proton source. See, for example, Lazouski, N., Schiffer, Z. J., Williams, K. & Manthiram, K. Understanding Continuous Lithium-Mediated Electrochemical Nitrogen Reduction. Joule 3, 1127-1139 (2019); Andersen, S. Z. et al. A rigorous electrochemical ammonia synthesis protocol with quantitative isotope measurements. Nature 570, 504-508 (2019); Tsuneto, A., Kudo, A. & Sakata, T. Lithium-mediated electrochemical reduction of high pressure N₂to NH₃ . J. Electroanal. Chem. 367, 183-188 (1994); and Schwalbe, J. A. et al. A Combined Theory-Experiment analysis of the Surface Species in Lithium Mediated NH₃Electrosynthesis. ChemElectroChem celc.201902124 (2020) doi:10.1002/celc.201902124, each of which is incorporated by reference in its entirety. While the cathode reactions in this system are well-described (FIG. 10A), the anode reaction is poorly defined and likely involves THF oxidation. Solvent decomposition prevents the method from being a practical approach to ammonia production.
Oxidizing H₂at the anode to produce protons of a controlled thermodynamic activity avoids the aforementioned issues. See, for example, Singh, A. R. et al. Strategies toward Selective Electrochemical Ammonia Synthesis. ACS Catal. 9, 8316-8324 (2019), which is incorporated by reference in its entirety. As an added benefit, hydrogen oxidation can be used as a renewable anode reaction for synthetic applications in which sacrificial anodes are used, allowing for continuous production of useful chemicals. See, for example, Matthessen, R., Fransaer, J., Binnemans, K. & De Vos, D. E. Electrocarboxylation: towards sustainable and efficient synthesis of valuable carboxylic acids. Beilstein J. Org. Chem. 10, 2484-2500 (2014), which is incorporated by reference in its entirety. However, the rate of hydrogen oxidation in nonaqueous solvents at flooded electrodes is limited by the solubility of hydrogen and its corresponding diffusion-limited oxidation rate, equal to several milliamperes per square centimeter (mA cm⁻²) (FIG. 10B), which is far too low for practical applications. See, for example, Gibanel, F., Lopez, M. C., Royo, F. M., Santafé, J. & Urieta, J. S. Solubility of nonpolar gases in tetrahydrofuran at 0 to 30° C. and 101.33 kPa partial pressure of gas. J. Solution Chem. 22, 211-217 (1993); and Bard, A. J. & Faulkner, L. R. Electrochemical Methods. Fundamentals and Applications. (John Wiley & Sons, Inc, 2001), each of which is incorporated by reference in its entirety. Similar diffusion limitations are pronounced for gaseous feedstocks such as N₂(FIG. 10C). See, for example, Lazouski, N., Schiffer, Z. J., Williams, K. & Manthiram, K. Understanding Continuous Lithium-Mediated Electrochemical Nitrogen Reduction. Joule 3, 1127-1139 (2019); and Zhou, F. et al. Electro-synthesis of ammonia from nitrogen at ambient temperature and pressure in ionic liquids. Energy Environ. Sci. 10, 2516-2520 (2017), each of which is incorporated by reference in its entirety.
One way to overcome diffusion limitations for gaseous reactants in electrochemical reactions is to use gas diffusion electrodes (GDEs), in which intimate contact between the gas, electrolyte, and catalyst is generated. See, for example, Mathur, V. & Crawford, J. Fundamentals of Gas Diffusion Layers in PEM Fuel Cells. Recent Trends Fuel Cell Sci. Technol. 400, 116-128 (2007), which is incorporated by reference in its entirety. This contacting minimizes the distance that gas molecules have to travel through the electrolyte to react at the catalyst (FIG. 11A), thus achieving much higher diffusion rates than are possible at flooded electrodes in aqueous electrolytes. GDEs have been used in hydrogen fuel cells and for CO and CO₂reduction. See, for example, Mathur, V. & Crawford, J. Fundamentals of Gas Diffusion Layers in PEM Fuel Cells. Recent Trends Fuel Cell Sci. Technol. 400, 116-128 (2007); Litster, S. & McLean, G. PEM fuel cell electrodes. J. Power Sources 130, 61-76 (2004); Ripatti, D. S., Veltman, T. R. & Kanan, M. W. Carbon Monoxide Gas Diffusion Electrolysis that Produces Concentrated C₂Products with High Single-Pass Conversion. Joule 3, 240-256 (2018); Ren, S. et al. Molecular electrocatalysts can mediate fast, selective CO₂reduction in a flow cell. Science (80-.). 365, 367-369 (2019); Higgins, D., Hahn, C., Xiang, C., Jaramillo, T. F. & Weber, A. Z. Gas-Diffusion Electrodes for Carbon Dioxide Reduction: A New Paradigm. ACS Energy Lett. 4, 317-324 (2019); and Burdyny, T. & Smith, W. A. CO₂reduction on gas-diffusion electrodes and why catalytic performance must be assessed at commercially-relevant conditions. Energy Environ. Sci. 12, 1442-1453 (2019), each of which is incorporated by reference in its entirety.
In the aforementioned applications, the electrolytes are typically aqueous solutions or water-saturated polymeric materials, while the GDE support is hydrophobized to control wetting. See, for example, Mathur, V. & Crawford, J. Fundamentals of Gas Diffusion Layers in PEM Fuel Cells. Recent Trends Fuel Cell Sci. Technol. 400, 116-128 (2007), which is incorporated by reference in its entirety. The hydrophobic interactions between the electrolyte and support, as well as the small pore size in the support prevent electrolyte penetration and flooding into the fibrous structure of the GDE. Instead, a thin layer of electrolyte is in contact with the catalyst through which reactant gas molecules must diffuse (FIG. 11A). See, for example, Weng, L. C., Bell, A. T. & Weber, A. Z. Modeling gas-diffusion electrodes for CO₂reduction. Phys. Chem. Chem. Phys. 20, 16973-16984 (2018), which is incorporated by reference in its entirety. If the primary component of the electrolyte is a nonaqueous solvent, such as tetrahydrofuran (THF), then the interactions between the support and electrolyte are no longer unfavorable, which leads to penetration of the electrolyte into the fibrous structure of the GDE (FIG. 11B), effectively flooding the catalyst. The gas must then diffuse over large distances through the solution to react (FIG. 11B), lowering the maximum obtainable current densities. See, for example, Tran, C., Yang, X.-Q. & Qu, D. Investigation of the gas-diffusion-electrode used as lithium/air cathode in non-aqueous electrolyte and the importance of carbon material porosity. J. Power Sources 195, 2057-2063 (2010), which is incorporated by reference in its entirety. Flooding prevents the use of standard carbon fiber-based GDEs with nonaqueous electrolytes for improving the rates of reactions involving sparingly soluble gases. While some approaches to overcome these issues have been reported, no previously reported GDE generated effective gas-liquid contact to greatly increase the obtained current for electrosynthetic applications in nonaqueous solvents. See, for example, Tran, C., Yang, X.-Q. & Qu, D. Investigation of the gas-diffusion-electrode used as lithium/air cathode in non-aqueous electrolyte and the importance of carbon material porosity. J. Power Sources 195, 2057-2063 (2010); Balaish, M., Kraytsberg, A. & Ein-Eli, Y. Realization of an artificial three-phase reaction zone in a Li-Air battery. ChemElectroChem 1, 90-94 (2014); and Gourdin, G., Xiao, N., McCulloch, W. & Wu, Y. Use of Polarization Curves and Impedance Analyses to Optimize the ‘triple-Phase Boundary’ in K—O₂Batteries. ACS Appl. Mater. Interfaces 11, 2925-2934 (2019), each of which is incorporated by reference in its entirety. As described herein, GDE-like behavior was obtained by controlling material wetting and electrolyte penetration into supports when using nonaqueous electrolytes. A stainless steel cloth (SSC) was used as the substrate onto which the catalyst was deposited. The electrolyte penetration was controlled by maintaining a pressure gradient across the cloth. The approach was used to efficiently oxidize H₂on Pt-coated steel cloths (Pt/SSC) at current densities up to 25 mA cm⁻²in tetrahydrofuran and propylene carbonate-based electrolytes. In addition, by using an SSC as a substrate onto which lithium metal is plated in situ, it was possible to reduce N₂to NH₃using a lithium-mediated approach. An NH₃partial current density of 8.7±1.5 mA cm⁻²and high FEs (35±6% at rate-optimized conditions, 47.5±4% at FE-optimized conditions) were obtained. The two electrodes were coupled to build an electrochemical Haber-Bosch (eHB) reactor, which can produce NH₃from N₂and H₂at ambient conditions.
An electrochemical Haber-Bosch reactor is described below, where hydrogen and nitrogen are utilized to produce ammonia at ambient conditions using electrical potential. The electrolyte contains one molar tetrafluoroborate (1 M LiBF₄) and 0.11 molar ethanol (EtOH) in tetrahydrofuran (THF); it is nonaqueous, as water is not used as the bulk solvent for the electrolyte. Nitrogen gas is reduced at the cathode on a stainless steel cloth by lithium metal which is electrochemically plated onto the mesh in situ. The steel cloth is set up to act as gas diffusion electrode, with the electrolyte and gas well-separated by the sloth, which generates efficient gas-liquid contacting at the electrode. The steel mesh acts as the support onto which the catalyst, lithium metal, is deposited in situ. Nitrogen gas was flowed past the electrode, with single pass conversions reaching ˜10%.
At the anode, the electrode was a platinum-coated stainless steel cloth that was also set up as a gas diffusion electrode. Hydrogen gas was oxidized at anode with nearly unity (>99%) Faradaic efficiency and rates at least an order of magnitude higher than possible at flooded electrodes (˜2.5 mA cm⁻²at flooded electrodes vs 25 mA cm⁻²at the GDE). High single pass conversions (>80%) of the feed gas were demonstrated.
In addition to demonstrating the ability to oxidize hydrogen and reduce nitrogen in THF-based electrolytes, the work also demonstrates that platinum coated stainless steel cloths can be used to oxidize hydrogen in other nonaqueous solvents. The solvent used to demonstrate this was 1 M LiBF₄in 9:1 propylene carbonate/ethylene carbonate. In this solvent, the transport-limited current density for hydrogen oxidation at conventional flooded platinum electrodes is ˜0.25 mA cm⁻², fartoo low for practical applications. Oxidation was demonstrated with unity selectivity at rates two orders of magnitude higher than previously achieved, at 25 mA cm⁻².
The architecture described herein can also be used to improve the selectivities and rates of fuel production from gaseous feedstocks such as N₂, CO, and CO₂by alleviating diffusion limitations and allowing for precise control of proton activity in a nonaqueous solvent. For example, the hydrogen oxidation anode described herein can also be coupled to a wide range of cathodic hydrogenations, making it broadly useful in electroorganic synthesis, an emerging area that is finding commercial interest in energy and pharmaceutical companies. The methods and systems described herein currently holds the record for the highest rates of ammonia synthesis at ambient conditions (FIG. 3). For example, the ammonia production rate can be greater than 1×10⁻⁸mol cm⁻²s⁻¹. The Faradaic efficiency can be greater than 30%, preferably greater than 40%.
As electrochemical reactions involving poorly soluble gases at flooded electrodes are limited by the diffusion rate of gas molecules from the bulk electrolyte to the electrode surface, methods to decrease the diffusion distance have been proposed. In gas diffusion electrodes, the interface between the gas and liquid phases is positioned close to the electroactive surface to significantly reduce the distance that the gas must travel, thus increasing rates. In traditional carbon-fiber gas diffusion electrodes, which are used with aqueous electrolytes, the carbon fibers are hydrophobic; the hydrophobic repulsion between the electrolyte and carbon fibers, as well as the small spacing (pore size) between the fibers prevents catalyst flooding by the electrolyte. Thus, a well-defined boundary between the gas and liquid close to the catalyst is obtained. Hydrophobic coatings of carbon fibers with nonaqueous electrolytes do not allow for development of a well-defined gas-liquid interface as the interactions between the carbon fiber and electrolyte and no longer unfavorable, and sometimes favorable, leading to flooding of the catalyst and increases in the distance that molecules need to diffuse.
To understand where the liquid-gas interface is relative to the solid surface, one can look at the relative energies of the gaseous and liquid phases under various conditions. First, assume that the electrode is parallel to the gravitational vector; some modification of the analysis can be made to accommodate angled electrodes. Next, find the configurations of gas and liquid that are energetically stable or meta-stable to determine the location of the gas-liquid interface. For these analyses, assume that the system is closed and that the amount of gas or liquid in it is constant. Transitions between states require the gas and liquid to move around, i.e. exchange places. The total energy in the system can be quantified as follows:
$E = \int (P + ρ gy + \frac{ρ v^{2}}{2}) d V$
Here, P is the pressure at a given location, ρ is the density of the phase at a given location, 9 is the gravitational acceleration constant, y is the vertical location, v is the velocity of the phase at a given location. The below analysis will assume stagnant phases, so v=0. One can note from the outset that the liquid phase is more affected by the gravitational force than the gas phase, while both phases are affected by the pressure. These heuristics can be used to predict qualitatively the configurations of gas-liquid boundaries and their stability. For ease of analysis, one can also plot an average energy as a function of position along the x-axis:
$E (x) = \int (P + ρ gy) dy \approx P \cdot H_{GDE} + \frac{ρ_{liquid} g H_{liquid}^{}}{2}$
In the simplest case, take a box only half-filled with liquid. The stable configuration is one where the liquid is all below the gas, as it is denser, irrespective of initial configuration, shown in FIG. 4. If an impenetrable wall, such as a solid electrode, is place in the box, the stable, equilibrium configuration does not change, shown in FIG. 5. However, for any initial configuration, there is a metastable configuration in which the liquid is level in each one of the compartments, but not equal between compartments, shown in FIG. 6. Going from the metastable to the equilibrium state is hampered by a large energy barrier for the liquid phase, as the liquid cannot penetrate through the wall. In these systems, the gas-liquid interface is horizontal, while the electrode is vertical, so the amount of electrode area with good gas-liquid-solid contacting is small. In theory, electrodes could be positioned horizontally for GDE-like behavior in a stable manner, shown in FIG. 7. Practically, this may be difficult to implement and scale because of the nature of the gas-liquid interface at the cathode and the anode.
A traditional hydrophobic gas diffusion electrode with an aqueous electrolyte. In this case, the barrier (wall) is porous, so the water and gas could, in theory, reach the equilibrium state in FIG. 5. However, the liquid phase (water) cannot move freely though the porous electrode because of the hydrophobic interactions between the substrate and electrolyte. This effect is manifested as a pressure gradient across the interface between the gas phase and the liquid inside the porous structure, roughly given by the Young-Laplace equation:
$P_{liquid} - P_{gas} = \frac{2 γ_{SL}}{R_{pore}}$
The gas similarly cannot enter the liquid phase, as forming the first bubble is also limited by the Laplace pressure:
$P_{gas} - P_{liquid} = \frac{2 γ_{GL}}{R_{bubble}} \approx \frac{2 γ_{GL}}{R_{pore}}$
Therefore there are two energy barriers, one for the liquid, one for the gas, which maintains a meta-stable boundary between the gas and liquid at the GDE, which allows for increased rates. When nonaqueous electrolyte is used, the energy barrier for the liquid phase penetration is broken, which leads to electrolyte penetration, catalyst flooding, and lower rates of gas utilization. In some cases, the capillary action can make the energy inside the electrode lower than outside of it (carbon fibers+THF, for instance), which exacerbates flooding. See, FIG. 8. To prevent electrolyte penetration, one can artificially impose an energy barrier for the liquid phase. As described herein, this is done by increasing the pressure in one of the compartments (the gas compartment). While this forces both phases to the lower energy compartment, the gas cannot enter it due to the Laplace pressure; the liquid phase is stably in the other compartment. See, for example, FIG. 9. In this case, the gas phase is metastable, while the liquid phase is in its lowest energy state, as long as
P _gas −P _liquid >ρgH
This requirement imposes a maximum height of GDE that can be used. The height can be increased by decreasing the pore size of the GDE, which increases the Laplace pressure of the GDE, which in turn increases the highest allowable pressure gradient. Through this analysis, surprisingly, both traditional carbon fiber GDEs and metal mesh GDEs with a pressure gradient can both utilize local energy minima for phase distributions to obtain gas-liquid interfaces close to the electrode surface for increased rates. In carbon fiber GDEs, stability is obtained by a “phantom” pressure gradient from hydrophobic interactions, while in metal mesh GDEs, the pressure gradient can be explicitly applied.
Results
Carbon fiber electrodes for hydrogen oxidation
First, SSCs were examined to improve the rates of the hydrogen oxidation reaction (HOR) in a THF-based electrolyte in order to utilize it as a renewable anode chemistry. Initially, commercially available platinum on carbon fiber (Pt/C) GDEs were used, which are capable of greatly increasing the rate of hydrogen oxidation in aqueous electrolytes (FIG. 16). When flowing H₂gas past the GDEs with a THF-based electrolyte it was found that very little H₂oxidation occurs at any applied current (FIG. 12A). Establishing a large (20 kPa) pressure gradient across the GDE to allow gas flow through the GDE in order to prevent complete flooding improved HOR FE somewhat (FIG. 12A). The pressure gradient at which flow through the GDE is observed is defined by the Laplace pressure; it was found to be 20±4 kPa for Pt/C GDEs. At pressure gradients at which gas flow through the electrode is observed, commercially available Pt/C GDEs were able to support HOR partial current of ˜12 mA cm⁻²(FIG. 12A), estimated by taking the product of the total applied current and measured HOR FE. See, for example, Santamaria, A. D., Das, P. K., MacDonald, J. C. & Weber, A. Z. Liquid-water interactions with gas-diffusion-layer surfaces. J. Electrochem. Soc. 161, F1184-F1193 (2014), which is incorporated by reference in its entirety.
Stainless steel cloth electrodes for hydrogen oxidation
The fibrous structure of Pt/C GDEs and favorable interactions between the electrolyte and carbon are responsible for flooding of the electrode (FIG. 11B, FIGS. 17A-17G). The flooding behavior makes Pt/C GDEs unsuitable for high rate gas utilization. Therefore, an alternative GDE support was explored to avoid these issues. Stainless steel cloths (SSC) were chosen as the GDE support (FIG. 11C, FIG. 14A), as, unlike carbon fibers, metal threads do not take up electrolyte by capillary action (FIGS. 17A-17G). At non-zero pressures gradients across the SSC, a well-defined separation between the gas and liquid is obtained (FIGS. 17A-17G). As stainless steel is a poor hydrogen oxidation catalyst, platinum was electrodeposited (FIGS. 14C-14D and FIGS. 18-22), which is an active HOR catalyst, onto the stainless steel cloths. The platinum-coated steel cloths (Pt/SSC) are able to oxidize H₂in the THF-based electrolyte with nearly unity FE (FIG. 12A) up to applied currents of 25 mA cm⁻²; higher currents were difficult to test due to large electrolyte resistance. This corresponds to approximately a one order of magnitude increase in the HOR current when compared to a flooded geometry (FIG. 10B).
Hydrogen oxidation was confirmed by cyclic voltammetry experiments (FIGS. 23A-23E) and by accounting for the mass balance over hydrogen (FIGS. 24A-24B). The increase in the maximum rate of HOR cannot be explained by a simple electroactive surface area effect, as a higher surface area electrode flooded by electrolyte would be subject to the same one dimensional transport-limited current density. Breaking the transport limit and resulting increases in current arise from the establishment of gas-liquid interfaces which promote gas transport close to the electrode surface.
The HOR FE demonstrates robustness to changes in non-zero pressure gradients across the Pt/SSC (FIG. 12B). Pressure gradients at or above the Laplace pressure of the cloth lead to gas flow into the electrolyte (FIGS. 17A-17G). See, for example, Santamaria, A. D., Das, P. K., MacDonald, J. C. & Weber, A. Z. Liquid-water interactions with gas-diffusion-layer surfaces. J. Electrochem. Soc. 161, F1184-F1193 (2014), which is incorporated by reference in its entirety. While gas flow through the electrode into the electrolyte does not affect the SSC performance, it may be undesirable for continuous operation due to the need to control multiple gas streams in the electrochemical cell. By maintaining the pressure gradient below the Laplace pressure of the cloth, gas flow exclusively past the electrode can be achieved. A majority of the experiments described in this work were performed with gas flow past the electrode. At non-zero pressure gradients across the Pt/SSC, the flowrate of H₂past the electrode can be varied with no change in HOR FE (FIGS. 25A-25D). At the highest applied currents and lowest flowrates, the nominal single pass conversion of hydrogen is approximately 80% (FIGS. 25A-25D), which is significantly higher than the single pass conversion of 25-35% used in traditional Haber-Bosch reactors. See, for example, Morgan, E. R., Manwell, J. F. & McGowan, J. G. Sustainable Ammonia Production from U.S. Offshore Wind Farms: A Techno-Economic Review. ACS Sustain. Chem. Eng. 5, 9554-9567 (2017), which is incorporated by reference in its entirety.
In order to demonstrate the generality of the approach of using SSC as a GDE support for nonaqueous solvents, oxidation of hydrogen at high rates in a 1 M LiBF₄in 9:1 propylene carbonate/dimethyl carbonate electrolyte was attempted. The electrolyte is similar to electrolytes used in some batch lithium-mediated nitrogen reduction approaches. See, for example, Kim, K., Chen, Y., Han, J.-I., Yoon, H. C. & Li, W. Lithium-mediated ammonia synthesis from water and nitrogen: a membrane-free approach enabled by an immiscible aqueous/organic hybrid electrolyte system. Green Chem. (2019) doi:10.1039/C9GC01338E; and Kim, K. et al. Electrochemical Synthesis of Ammonia from Water and Nitrogen: A Lithium-Mediated Approach Using Lithium-Ion Conducting Glass Ceramics. ChemSusChem 11, 120-124 (2018), each of which is incorporated by reference in its entirety. See, for example, each of which is incorporated by reference in its entirety. It was found that the transport-limited hydrogen oxidation current density at flooded Pt foil electrodes is ˜0.25 mA cm⁻²(FIGS. 26A-26D), which is an order of magnitude lower than for hydrogen oxidation in THF-based electrolyte (FIG. 10B), likely due to the propylene carbonate electrolyte's higher viscosity and lower hydrogen solubility. By using a Pt/SSC anode, it was possible to oxidize hydrogen at rates of 25 mA cm⁻²with near unity FE (FIGS. 26A-26D) in the propylene carbonate-based electrolyte, which is two orders of magnitude higher currents than in the flooded case, even despite an absence of good proton acceptors in the solution. This experiment demonstrates that SSCs can be used to obtain GDE-like behavior across different nonaqueous solvents.
Stainless steel cloth electrodes for nitrogen reduction
Having overcome transport limitations for hydrogen oxidation by using SSCs as an anode, SSCs for the cathodic reaction of nitrogen reduction was implemented. The lithium metal-mediated approach for N₂reduction has been reported to be diffusion limited in THF on flooded copper and steel foils (FIG. 10C). See, for example, Lazouski, N., Schiffer, Z. J., Williams, K. & Manthiram, K. Understanding Continuous Lithium-Mediated Electrochemical Nitrogen Reduction. Joule 3, 1127-1139 (2019), which is incorporated by reference in its entirety. In aqueous electrolytes, the transport limited current density for nitrogen reduction is typically even lower than in nonaqueous systems—approximately 0.5-1.3 mA cm⁻²—and depends on the hydrodynamics of the electrolyte (see below). The transport limited current density can be increased by decreasing the boundary layer thickness through improving the hydrodynamics, for instance with a rotating disk electrode (RDE) or microfluidic reactor, or by using gas diffusion electrodes. Claims of nitrogen reduction s electrolytes above these rates that do not utilize methods to improve nitrogen transport must be examined with scrutiny adding an additional criterion for evaluating the veracity of nitrogen reduction reports. See, for example, Andersen, S. Z. et al. A rigorous electrochemical ammonia synthesis protocol with quantitative isotope measurements. Nature 570, 504-508 (2019); and Greenlee, L. F., Renner, J. N. & Foster, S. L. The Use of Controls for Consistent and Accurate Measurements of Electrocatalytic Ammonia Synthesis from Dinitrogen. ACS Catal. 8, 7820-7827 (2018), each of which is incorporated by reference in its entirety.
To overcome diffusion limitations, a stainless steel cloth (SSC) was used as the GDE substrate onto which lithium metal was plated in situ. It was found that the rate of the nitrogen reduction reaction (NRR) is significantly enhanced compared to the flooded case (FIG. 10C) at high applied currents (FIG. 12C). This enhancement demonstrates that SSC cathodes can yield rates for continuous NRR that are competitive with literature reports (FIG. 13A, Table 1). At the highest production rates (FIG. 12C), a total of 8.6±1.4 μmol of ammonia are produced at a 1 cm²electrode after 290 seconds of polarization, resulting in 4.5±0.6 mM and 0.45±0.22 mM ammonia concentrations in the electrolyte and trap, respectively. At the highest FEs (FIG. 12D), a total of 11.8±1 μmol of ammonia are produced after 480 seconds of polarization. As the total amount of charge passed in most experiments is constant and equal to 7.2 coulombs, current was applied to the cell for longer periods of time than 290 seconds in most experiments. In longer duration experiments in which more charge is passed, 102 μmol of ammonia were produced over the course of 2 hours, with an FE of 18.8% (FIGS. 27A-27B). It was found that the total amount of ammonia produced increases monotonically with time after an induction period of approximately 2 minutes (FIG. 28), which suggests that the lithium-based catalyst cycle (FIG. 10A) rapidly reaches steady state operation. As the induction period is included in the time used to compute production rates, the rates reported here are underestimates of true values for a continuous system. Longer duration experiments (FIGS. 27A-27B) suggest that the production rate decreases with time, which could occur for a number of reasons, including but not limited to lithium ion depletion, solution phase ammonia affecting lithium plating and nitridation reactions, and incomplete proton donor recycling. Further experiments are necessary to conclusively determine and address the cause of the decreasing ammonia production rate at longer times.
Nitrogen reduction control experiments
NH₃was confirmed to be produced via N₂reduction by performing control experiments in which argon and isotopically labelled N₂were used as feed gases (FIGS. 15A-15C). No NH₃is produced with argon as a feed gas, and there is quantitative agreement between the amount of NH₃produced when ¹⁴N₂and ¹⁵N₂are used as feed gases. See, for example, Andersen, S. Z. et al. A rigorous electrochemical ammonia synthesis protocol with quantitative isotope measurements. Nature 570, 504-508 (2019), which is incorporated by reference in its entirety.
The isotope labeling experiments were performed at two different operating conditions and architectures, which is necessary for conclusive proof of nitrogen reduction. See, for example, Kibsgaard, J., Norskov, J. K. & Chorkendorff, I. The Difficulty of Proving Electrochemical Ammonia Synthesis. ACS Energy Lett. 4, 2986-2988 (2019), which is incorporated by reference in its entirety. The produced NH₃can be found in both the solution and the gas phases (FIGS. 12C-12D). This is additional evidence of GDE-like behavior of the SSC cathode. Obtaining ammonia in the gas phase is desirable as it may simplify separations in a practical process.
In the absence of a pressure gradient across the SSC, the system reverts to a flooded state and generally shows poor efficiency for N₂reduction (FIG. 12D). While the changes in total NH₃FE with non-zero pressure gradients (FIG. 12D) do not demonstrate a clear trend when accounting for the measured uncertainty, a maximum FE average combined solution and gas phase FE of 47.5±3.8% was obtained when the pressure gradient was 0.5 kPa using a SSC as the cathode. The total NH₃FE also does not significantly change when the flowrate of N₂past the SSC is varied (FIGS. 25A-25D); less NH₃is found in the gas phase at lower flowrates. The high conversions of N₂(˜10%) at low N₂flowrates (FIGS. 25A-25D) are desirable for a practical process; they also are indirect evidence of N₂reduction to NH₃, as the fraction of NO_xand NH_ximpurities in the stock N₂would need to be at approximately 1000-10000 parts per million to yield this much NH₃, several orders of magnitude higher than the maximum contamination levels of these impurities reported in the stock gas.
Demonstration of electrochemical Haber-Bosch
After obtaining efficient chemistries using SSCs independently at both the cathode and anode, the two reactions were coupled into an electrochemical Haber-Bosch (eHB) reactor, which produces NH₃from N₂and H₂at ambient conditions. Using SSC-based GDEs for both electrodes in a single reactor (FIG. 11D), high ammonia yields was maintained at the cathode (FIG. 13B). At longer reaction times, the advantage of using H₂oxidation at a Pt/SSC anode over THF oxidation at a platinum foil is evident, as solvent oxidation is avoided (FIB. 13B). The fate of the proton donor is also addressed by coupling the electrode chemistries. The proton donor, ethanol, is consumed at the cathode to produce NH₃and ethoxide, and could be regenerated from the ethoxide at the anode by protons produced from H₂oxidation (FIG. 11D). In long duration experiments, the number of protons found in ammonia was ˜80% of the number of labile protons originally present in ethanol in the electrolyte. As hydrogen evolution at the cathode also utilizes protons from ethanol, the high utilization of protons suggests that ethanol is indeed regenerated from ethoxide at the anode via hydrogen oxidation.
The eHB reactor operates at ambient conditions, which allows it to be operated at smaller scales than traditional Haber-Bosch. However, H₂is usually sourced from steam-methane reforming, which utilizes fossil fuels and is not readily modularized. See, for example, Inc., N. Equipment design and cost estimation for small modular biomass systems, synthesis gas cleanup and oxygen separation equipment. NREL Subcontract report http://www.nrel.gov/docs/fy06osti/39946.pdf (2006), which is incorporated by reference in its entirety. Water electrolysis is a modular alternative H₂source. By coupling an electrochemical Haber-Bosch reactor and a water electrolyzer (FIG. 13C and FIG. 13D), NH₃was obtained in an overall reaction involving only N₂, H₂O, and renewable electrons. By using a commercially available water splitting setup, NH₃was produced with an FE of 30±2% (FIG. 13B). The slight decrease in FE compared to eHB is likely from water contamination from the H₂stream. Coupling multiple unit operations in series can be advantageous for nonaqueous electrochemical ammonia production as it may increase the efficiency of each individual step as well as the overall process.
While development of GDEs capable of operating in nonaqueous solvents is imperative for practical electrochemical synthesis, many other aspects of system design require further development. Physical methods to recycle volatile organic solvents and separate the products when using gaseous feedstocks may be necessary in practical systems. Alternatively, bulk volatile solvents may be replaced with non-volatile analogs with similar properties, such as specially tailored polymers or ionic liquids. Polymeric electrolytes may open up avenues for manufacturing species-selective membranes for use in nonaqueous systems, analogous to Nafion in aqueous systems, and for manufacture of membrane-electrode assemblies (MEAs) for gas phase reactions. Electrolyte engineering can also decrease the ionic resistance in the cell, which is important for energy efficiency at high currents.
In the current work, for instance, the energy efficiency for NH₃production ranges from 1.4 to 2.8% at an applied cell potential of 20-30 V, with 70-80% of the energy losses coming from large electrolyte resistance (FIGS. 29A-29B). These values of energy efficiency correspond to energy consumptions of 730-1500 GJ/ton, significantly higher than contemporary values for Haber-Bosch, or even other lithium-mediated chemistries. See, for example, McEnaney, J. M. et al. Ammonia synthesis from N₂and H₂O using a lithium cycling electrification strategy at atmospheric pressure. Energy Environ. Sci. 10, 1621-1630 (2017), which is incorporated by reference in its entirety. However, further improvements to the electrolyte, cell architecture, Faradaic efficiency and cell lifetime can greatly improve these metrics by reducing sources of energy loss (FIGS. 29A-29B).
This work demonstrates the possibility of utilizing metal cloth-based supports for high rate electrochemical reactions of sparingly soluble gaseous reactants in a nonaqueous solvent. These SSCs were used to produce ammonia from nitrogen and water-derived hydrogen at the highest reported rates at ambient conditions, 30.4±3.5 nmol cm⁻²s⁻¹(FIG. 13A). The nonaqueous HOR Pt/SSC can find applications in reactions for which a continuous source of controlled-activity protons is needed, while an SSC at the cathode can be used in producing value-added chemicals from gaseous feedstocks such as N₂, CO₂, or CO. This approach for utilizing gaseous reactants can become a staple of organic electrosynthetic methodology.
Methods
Electrolyte preparation
Electrolyte solutions were prepared by dissolving 1 M of LiBF₄(Sigma-Aldrich, 98+%) in molecular sieve-dried tetrahydrofuran (Acros Organics, 99+%, stabilized with BHT) to which ethanol (VWR International Koptek, anhydrous, 200 proof) was added to yield an ethanol concentration of 0.11 M. The obtained solution was centrifuged at 6000 rpm for 10 minutes to precipitate insoluble impurities. The clear solution was transferred to oven-dried glass vials and used within 12 hours of preparation. In experiments in which hydrogen oxidation is quantified, ferrocene (Alfa Aesar, 99%) is added to the solutions to yield a ferrocene concentration of ˜0.25 M.
Preparation of platinum-coated steel cloths
Stainless steel cloths (McMaster-Carr, 304 stainless steel, 400×400 mesh) were electroplated with nickel followed by platinum (FIGS. 18A-18B). A Wood's nickel strike solution, which consists of 1 M NiCl₂(Sigma-Aldrich) and 1 M HCl (Sigma-Aldrich) in water, was used to plate nickel onto the cloths (FIGS. 19A-19F). The cloth was used as the working electrode while a piece of nickel foil (Alfa Aesar, 99+%) was used as the soluble counter electrode in an undivided beaker cell. The cloth was pretreated by applying an oxidative current of 15 mA cm_geom ⁻²for 30 seconds, immediately after which a reductive current of 30 mA cm_geom ⁻²was applied for 5 minutes obtain a nickel-plated stainless steel cloth (FIGS. 20A-20F).
After striking the cloth with nickel, platinum can be deposited. The platinum plating solution used was a citrate-ammonium bath, containing 35 mM (NH₄)₂PtCl₆(Alfa Aesar), 400 mM trisodium citrate (anhydrous, Beantown Chemical), and 75 mM of NH₄Cl (Alfa Aesar). The nickel-stricken cloth was used as the working electrode; a piece of platinum foil was used as a soluble counter electrode in a beaker cell, which was kept over a water bath at 90° C. A constant reductive current of 10 mA (˜5 mA cm_geom ⁻²) was applied to the cloth for 5 minutes. The platinum-coated cloths (FIGS. 21-22) were thoroughly rinsed with DI water to remove any ammonium containing compounds from the surface and dried at 80° C. in air prior to use.
Gas diffusion electrode experiments
Experiments were performed in 3-compartment cells (FIGS. 30-31), in which working and counter electrode compartments were separated by a Daramic 175 separator; all cell parts were oven-dried at 80° C. before use. The working electrode was a piece of stainless steel cloth in nitrogen reduction experiments or a piece of platinum-coated stainless steel cloth in hydrogen oxidation experiments. The working electrode was fixed between the working electrode compartment and a gas compartment. The working gas (e.g. N₂or H₂) was flowed first through a vial containing THF to saturate the gas with THF, followed by the gas compartment of the electrochemical cell; the gas was slightly pressurized by a water column at the outlet of the gas compartment (FIGS. 32A-32D). In experiments where propylene carbonate-based electrolyte was used, the vial contained propylene carbonate instead of THF.
1.75 mL of electrolyte was added to each electrode compartment, for a total electrolyte volume of 3.5 mL. Note that this is the volume of electrolyte added to each compartment and may not be the final electrolyte volume due to electrolyte evaporation. At this point, a pressure gradient was established across the working electrode due to the fact that the electrolyte compartment is at atmospheric pressure while the gas compartment is at positive pressure due to the water column at the compartment outlet; this pressure gradient prevents electrolyte flow into the gas compartment and establishes a robust gas-electrolyte interface. Initially, the height of the water column was chosen to redirect gas flow into the electrolyte compartment. Gas was flowed into the electrolyte for 10 minutes at 10 standard cubic centimeters per minute (sccm) to saturate the solution with the gas. The height of the water column was then lowered to obtain the desired pressure gradient across the SSC while maintaining flow past the SSC. In certain experiments, the flowrate of the gas past the electrode was also decreased at this stage (FIGS. 25A-25D). A constant current was then applied to the cell; in most experiments, 7.2 C of charge were applied irrespective of current density.
For nitrogen reduction experiments, an additional vial containing 2 mL of 0.1 M H₃BO₃(Alfa Aesar, 99.99%) was added between the gas compartment and the water column to capture any gas phase ammonia (FIGS. 32A-32D). To quantify ammonia, the catholyte was diluted to 100 mL in a volumetric flask in water, after which the obtained solutions were either used as-is or diluted 2- or 4-fold further for ammonia quantification via a colorimetric assay (FIG. 33). The boric acid trap was quenched with 500 μL of 0.4 M NaOH before being diluted to 25 mL in a volumetric flask. The ammonia content in the trap-derived solution was quantified via a colorimetric assay without further dilution.
For some hydrogen oxidation experiments, the electrolyte contained ferrocene at a concentration of ˜0.25 M, the oxidation of which was used to estimate the FE toward hydrogen oxidation (see Supplementary Discussion). In mass-balance closure and eHB cell experiments, the electrolyte was unchanged from the one used in nitrogen reduction experiments.
Nitrogen reduction control experiments
Nitrogen reduction to ammonia at SSC cathodes was confirmed by varying the feed gas in NRR experiments (FIGS. 15A-15C). All feed gases passed through three solutions before entering the cell to purify and prepare the gas. First, the gases were passed through 0.1 M NaOH to capture any NO_xin the gases, then through 0.05 M H₂SO₄to capture any NH₃, followed by THF with sieves to capture water in the gaseous stream and to saturate the gas with THF. Initially, 10 sccm of argon gas were passed through all three solutions to remove air, which contains ¹⁴N₂and O₂. Then, 5 sccm of the desired gas —Ar (Airgas), ¹⁴N₂(Airgas), or ¹⁵N₂(Cambridge Isotope Laboratories)—was passed through the solutions and fed to the cell for nitrogen reduction experiments.
Two architectures were used to confirm nitrogen reduction. In one set of experiments, a 3-compartment cell (FIGS. 30A-30B) with a SSC cathode, Pt foil anode, and a Daramic separator was used; 25 mA of current was applied to the cell for 290 seconds in these experiments. In the second set of experiments, a 4-compartment cell (FIG. 12D) with a SSC cathode, Pt/SSC anode, and no separator was used; in this configuration, 5 sccm of H₂(Airgas) was fed to the anode. 20 mA of current was applied for 360 seconds in these experiments.
The catholyte from the 3-compartment cell experiments, all the electrolyte from the 4-compartment cell experiments, and the acid trapped were acidified with 0.05 M H₂SO₄in water to convert all NH₃to NH₄₊. NMR spectra of the obtained solutions were taken with solvent suppression on a Bruker Avance Neo 500.18 MHz spectrometer (FIG. 15C). The ammonia content of the solutions was quantified via the colorimetric assay; the measured concentrations of ammonia were found to be consistent with those measured by NMR (FIGS. 35A-35C).
Hydrogen oxidation quantification
To quantify the HOR FE, an excess of ferrocene (˜0.25 M) was added to the electrolyte prior to electrolysis. As ferrocene is thermodynamically more difficult to oxidize than H₂, but easier than THF (FIG. 10B), any applied current would first oxidize H₂, assuming the oxidation is kinetically facile, followed by ferrocene, once HOR is diffusion limited. After application of current in HOR experiments using Pt/SSC anodes, the anolyte was diluted to 10 mL with N₂-purged water. The produced cloudy orange mixture was extracted with N₂-purged hexanes three times. The obtained solution was centrifuged for improved phase separation. The ferrocenium content of the produced clear aqueous solutions was quantified via UV-vis spectroscopy by using a combination of the visible (619 nm) and UV (255 nm) ferrocenium absorption peaks (FIGS. 36A-36D). The concentration of ferrocenium in solution was then used to estimate the maximum value of HOR FE. For a detailed discussion of the method, see below.
For experiments at low flowrates and high currents (i.e. at high conversions), the HOR FE was also computed by estimating the hydrogen flowrate out of the gas compartment and by using a hydrogen mass balance over the gas compartment (FIGS. 24 and 26). The results were consistent with those obtained via the ferrocene-based method. For a detailed discussion of the method, see below.
Other examples follow. The procedures and materials for nitrogen reduction outlined below have been optimized for reliable ammonia production. However, certain deviations from the procedure and material vendors, which are specifically called out in-text and in prior work, may lead to poorer ammonia production. See, for example, Lazouski, N., Schiffer, Z. J., Williams, K. & Manthiram, K. Understanding Continuous Lithium-Mediated Electrochemical Nitrogen Reduction. Joule 3, 1127-1139 (2019), which is incorporated by reference in its entirety. Follow the procedure closely for reproducibility and high yields.
Electrolyte solution preparation
Dry THF was used as the solvent in THF-based experiments described below. It was obtained by drying as-purchased THF over 20% v/v of freshly dried molecular sieves for at least 48 hours in a round-bottom flask sealed with a rubber septum stopper. The sieves were prepared by washing with acetone and heating at 300° C. for 5 hours in a muffle furnace. The water content of dry THF was found to be 7.1±0.3 ppm (n=3) via Karl-Fischer titration. As-purchased LiBF₄, stored in an Ar glovebox, was dissolved in dry THF to obtain electrolyte solutions containing 1 M LiBF₄. As discussed in Lazouski et al., it is imperative for the LiBF₄to be pure; LiBF₄purchased from Sigma-Aldrich was found to be sufficiently pure for these experiments, while other vendors' may require purification. Ethanol was added to the solution to obtain a concentration of EtOH of 0.11 M. Insoluble residue was removed from the solutions by centrifugation at 6000 rpm (4430 rcf) for 10 minutes. Clear electrolyte solutions were stored in oven-dried glass vials in a desiccator and used within 12 hours of preparation. While the solutions are somewhat water sensitive, performing centrifugation and solution transfer operations in atmospheric air is permissible, as long as the operations do not expose the solutions to air for long periods of time (hours, i.e. during storage). Oxygen from the atmosphere is typically purged from the electrolyte during saturation by gas flow (see below).
In propylene carbonate-based experiments, solvents and the electrolyte salt were used as received. LiBF₄was dissolved in a 9:1 by volume mixture of propylene carbonate and dimethyl carbonate to produce a 1 M LiBF₄in PC/DMC electrolyte. Dimethyl carbonate was added to the electrolyte in order to reduce the viscosity of the solution, as opposed to using pure propylene carbonate as the solvent. The resulting solution was centrifuged at 6000 rpm (4430 rcf) for 20 minutes to remove insoluble impurities. Clear electrolyte was transferred to oven-dried vials and used within 12 hours of preparation.
Nitrogen reduction experiments—steel foils
Flooded steel foil experiments (FIG. 10C) were performed analogously to copper foil experiments described in Lazouski et al. Briefly, two-compartment cells were assembled with a steel foil cathode, platinum foil anode, and a Daramic separator. Steel foils were prepared by polishing with 400 grit sandpaper, followed by 1500 grit sandpaper, and finally by rinsing with DI water and drying in air at 80° C. Daramic separators were prepared by successively soaking them in THF and water several times, as follows: as-received separators were immersed in as-purchased THF for ˜10 minutes with agitation, after which the solution turned yellow. The separators were then immersed in water for ˜10 minutes to remove THF and water-soluble impurities. The process of immersing in THF and water was repeated two more times, after which the separators were dried in air at 80° C. On rare occasions, fresh Daramic separators may initially lead to poor ammonia yields, likely due to contaminants left in the porous structure. Ammonia yields reach steady values after reusing the separator twice or more. All cell parts, separators, and electrodes were washed with DI water and dried in an oven at 80° C. prior to use. Daramic separators are also soaked in DI water for at least 10 minutes to remove electrolyte and residual ammonia from the structure prior to drying. It is recommended to rinse the cell parts with THF, acetone, and water if cell parts are new or have been previously used for other chemistries to remove impurities that may negatively affect yields. Cell parts and separators, and platinum foils are reused multiple times, while the steel foils are used once per experiment. 1.75 mL of 1 M LiBF₄, 0.11 M EtOH in THF electrolyte was added to each compartment of the cell. Note that this is the volume of electrolyte added to each compartment and may not be the final electrolyte volume due to electrolyte evaporation. THF-saturated N₂gas was bubbled through the cathode compartment at 10 standard cubic centimeters per minute (sccm) for 10 minutes to saturate the solution with N₂. A chosen constant current was then supplied to the cell using a Biologic VMP3 potentiostat, Tekpower TP3005T or Tekpower TP5003T DC power supply. When using a power supply, the current was monitored by measuring the potential drop across a calibrated 33-Ohm resistor in series with the electrochemical cell (FIGS. 38A-38D). Approximately 7.2 coulombs of charge were supplied in each experiment. The catholyte was removed from the cell and diluted with Milli-Q water in a 50 mL volumetric flask. The cathode compartment was rinsed twice with Milli-Q water and added to the flask, after which the solution was diluted to the mark. Three samples for ammonia quantification were made: one containing only 2000 μL of the diluted catholyte, one containing 1000 μL of the diluted catholyte and 1000 μL of Milli-Q water, and one containing 500 μL of the diluted catholyte and 1500 μL of Milli-Q water.
Assembly of gas diffusion cell
Gas diffusion experiments were performed using 3- and 4-compartment cells (FIGS. 30A-30B, FIG. 13D). In a 3-compartment cell, there are gas, working, and counter compartments, the latter two separated by a piece of Daramic. Daramic separators were prepared by washing in THF and water, as described above in Nitrogen reduction experiments—steel foils.
All cell parts, separators, and electrodes were dried in an oven at 80° C. prior to use. Aluminum current collectors for the GDE were made from 0.016″ thick 6061 aluminum; the hole was made with a 7/16″ wood drill bit and sanded to smooth the edges. O-rings between the compartments (FIGS. 30 and 31) are important for making a good gas seal. Fluoropolymer-coated silicon O-rings are recommended, as they are compatible with THF. The procedure to assemble a cell can be seen in FIGS. 31A-31H. In words, a “stack” is made by placing the pieces in the following order:
Once the cell is assembled, the inlet of the gas compartment is attached to a bubbler containing THF, through which the gas is bubbled. The outlet is attached to a tube that enters a tall (50 cm) burette containing a water column to control the gauge pressure inside the gas compartment and the pressure gradient across the GDE (FIGS. 32A-32D); in NRR experiments, a 0.1 M H₃BO₃in Milli-Q acid trap is inserted between the outlet of the cell and the burette (FIGS. 32A-32D) to capture gas phase ammonia.
After the cell is set up and gas is flowing through the bubbler at 10 sccm, 1.75 mL of LiBF₄/EtOH/THF electrolyte is added to the counter compartment, followed by 1.75 mL of electrolyte to the working compartment. Note that this is the volume of electrolyte added to each compartment and may not be the final electrolyte volume due to electrolyte evaporation. Initially, some of the electrolyte in the working compartment may enter the gas compartment; however, after the entire GDE is contacted by the electrolyte, the pressure in the gas compartment increases, as evidenced by motion of the gas level in the burette, and the electrolyte returns to the working compartment, after which gas begins to flow through the GDE (FIGS. 17A-17G). When using bare steel cloths, the pressure in the gas compartment is typically 1.6-1.8 kPag, corresponding to 16-18 centimeters of water column above the gas. The pressure in a flow through configuration is 1.8-2.0 kPag when platinum-coated steel cloths are used. The working compartment electrolyte is saturated with the fed gas by flowing it through the GDE for 10 minutes. Following saturation, the water level in the burette is decreased to lower the pressure in the gas compartment to the desired value and favor gas flow past the GDE and out of the gas compartment (FIGS. 32A-32D). Some gas bubbles begin to flow out into the burette almost immediately upon lowering of the pressure in the gas compartment below the flow-through pressure. The addition of an acid trap in NRR experiments (FIGS. 32A-32D) adds approximately 2 cm of water of additional gauge pressure in the gas compartment, which is accounted for when computing pressure gradients across the GDE. At this point, the flowrate of the gas can be reduced if desired, such as in experiments where the effect of flowrate past the electrode is quantified, after which current is applied. The electrolyte in the working compartment was removed and analyzed after application of current.
In a 4-compartment cell used in combined electrochemical Haber-Bosch (eHB) experiments, the setup is similar to the aforementioned 3-compartment cell. The main change is that the counter electrode is replaced by a second gas compartment and gas diffusion electrode. In summary, the stack becomes:
Nitrogen reduction experiments—steel cloth experiments
Nitrogen reduction reaction (NRR) experiments focused on studying the NRR on steel cloths were performed in 3-compartment cells, the setup of which is detailed in Assembly of a gas diffusion cell. A platinum foil anode and a circular stainless steel cloth (SSC) cathode (diameter=14 mm) were used as the electrodes. The steel cloth electrode was rinsed with DI water and dried at 80° C. prior to use; it was used only for one experiment before discarding. The platinum foil was reused indefinitely. The method of applying current and preparing quantification samples was analogous to the method described in Nitrogen reduction experiments—steel foil experiments, with one change: the catholyte was diluted in a 100 mL volumetric flask due to higher amounts of produced NH₃.
One modification that was made in NRR experiments was the addition of a 2 mL 0.1 M boric acid trap between the gas compartment and the pressure-controlling burette to capture any gas phase ammonia. In experiments where gas was fed through the SSC, the trap was inserted at the gas outlet in the catholyte chamber, so all fed gas still went through it. When no pressure gradient across the SSC was applied (FIG. 12D), no boric acid trap was present, hence no quantification of gas phase products could be performed. While this may have underestimated the amount of ammonia produced, it is likely that little ammonia was found in the gas phase in this experiment, as ammonia stripping from bulk solution does not occur (FIGS. 17A-17G and FIGS. 39A-39D). After each experiment, the solution in the trap was quenched with 500 μL of 0.4 M NaOH and diluted in a 25 mL volumetric flask with Milli-Q water, after which the ammonia concentration in the solution was quantified.
Nitrogen reduction experiments—time evolution of ammonia
Experiments were performed in which the concentration of ammonia in the electrolyte was measured as a function of operating time (FIG. 28). In these experiments, the setup was analogous to the experiments described in Nitrogen reduction experiments—steel cloth experiments, except that 1.9 mL of electrolyte was added to the cathode compartment instead of the usual 1.75 mL, providing extra electrolyte for periodically removing aliquots. A 20 mA cm⁻²current was applied for 10 minutes in these experiments to maintain a high production rate while allowing for good temporal resolution of the data. 10 sccm of THF-saturated N₂was flowed through the SSC to allow for bulk mixing in the electrolyte. Every 60 seconds, nominally 20 μL of catholyte was extracted from the cathode compartment via pipette. Because the electrolyte forms a thin film on the inside of the pipette tips that cannot readily be utilized, precise quantification of the amount withdrawn is difficult. By measuring the mass of electrolyte that can be recovered from the tips (not during the experiment), it was found that approximately 16 μL of electrolyte are used to make quantification samples. The extracted catholyte was diluted in 2 mL of Milli-Q water. The ammonia content of the two samples was quantified as described below. The amount of ammonia produced at each time point (FIG. 28) was computed according to Equation 1.
n _NH ₃(t)=(V ₀−0.02·t)·C(t) #(1)
After the experiment, the cathode was diluted in a 100 mL volumetric flask to quantify the remaining ammonia.
Ammonia quantification—calibration solution preparation A fresh calibration curve was made during each batch of quantifications. The calibration solutions contained a known amount of NH₄Cl in Milli-Q water (FIGS. 34A-34D). While ammonia samples from NRR experiments had some amount of LiBF₄, EtOH, and THF in them, they were sufficiently dilute as to not affect the quantification significantly (FIGS. 34A-34D). The absorbance of solutions containing THF is diminished, potentially leading to an underestimate in ammonia concentration, as pure water calibrations curves (FIGS. 34A-34D) were used for all analyses.
¹⁵N₂isotope labeling experiments
Isotope labeling experiments were used as a control to confirm N₂reduction to NH₃. In order to remove any NH₃and NO_xpotentially found in ¹⁵N₂stock (and house ¹⁴N₂), the gases were successively passed through solutions of 0.1 M NaOH in water (to capture NO_x), 0.05 M H₂SO₄in water (to capture NH₃), and THF containing activated molecular sieves (to capture water and to saturate the gas with THF) before being fed to the cell. Two isotope labeling experiments were performed: one utilized a typical 3-compartment cells with a steel cloth cathode, Daramic separator, and platinum foil anode with pressure control, while the other utilized an eHB reactor with a steel cloth cathode, Pt/SSC anode, and no separator.
Prior to any experiments, 10 sccm of Ar were bubbled through the entire setup (traps, bubbler, and cell) for 15 minutes to remove trace amounts of ¹⁴N₂and other impurities. The desired gas (Ar, ¹⁴N₂, or ¹⁵N₂) was fed to the cell and through the SSC at 5 sccm for 10 minutes. In the eHB experiment, H₂was fed through the Pt/SSC at 5 sccm for 10 minutes as well.
In the 3-compartment cell experiment, the pressure gradient across the SSC was decreased to 1 kPa, after which 25 mA of current was applied for 4.8 minutes. The catholyte was removed from the cell into a glass vial. The cathode compartment was rinsed with 0.05 M H₂SO₄twice and the resulting fractions were added to the catholyte. The entire mixture was diluted to ˜4 mL with 0.05 M H₂SO₄. Similarly, the boric acid trap was diluted to a total volume of ˜4 mL with 0.05 M H₂SO₄. The solution was acidified to convert all the ammonia to ammonium (NH₄ ⁺) for NMR analysis. See, for example, Nielander, A. C. et al. A Versatile Method for Ammonia Detection in a Range of Relevant Electrolytes via Direct Nuclear Magnetic Resonance Techniques. ACS Catal. 9, 5797-5802 (2019), which is incorporated by reference in its entirety.
In the eHB experiment, the pressure gradient across both electrodes was decreased to 1 kPa, after which the gases were flowed past the electrodes for an additional 2 minutes. 20 mA of current were passed for 6 minutes. The electrolyte was removed from the cell into a 10 mL volumetric flask. The cell was rinsed with 0.05 M H₂SO₄and the resulting solutions were added to the volumetric flask and diluted to the mark (10 mL). The boric acid trap was diluted to a total volume of ˜4 mL.
The ammonia content in the resulting solutions was quantified using the salicylate method described above. The electrolyte solutions for quantification were neutralized and diluted 50-fold (40 μL of sample, 10 μL of 0.4 M NaOH, 1950 μL of water) in 3 compartment case or 20-fold (100 μL of sample, 25 μL of 0.4 M NaOH, 1875 μL of water) in the eHB case, while the trap solutions were diluted 10-fold (200 μL of sample, 50 μL of 0.4 M NaOH, 1750 μL of water). NMR spectra of the undiluted solutions were measured on a three-channel Bruker Avance Neo 500.18 MHz spectrometer. Solvent suppression of the largest one (H₂O) or three (THF+H₂O) peaks was used to increase the signal-to-noise ratio for ammonium peaks. Locking and shimming was done on ¹H from water in the solution; no additional compounds were added to the solution prior to NMR. 64 scans were measured for all spectra. The N—H coupling in the NMR spectra confirms ammonia formation from feed N₂(FIG. 15C).
Preparation of platinum-coated steel cloths
In order to prepare an SSC for hydrogen oxidation, platinum metal, an effective HOR catalyst, was electrodeposited onto the steel cloths. It was found that platinum metal has poor adhesion to stainless steel, which has also been observed in the literature. See, for example, Stoychev, D., Papoutsis, A., Kelaidopoulou, A., Kokkinidis, G. & Milchev, A. Electrodeposition of platinum on metallic and nonmetallic substrates—selection of experimental conditions. Mater. Chem. Phys. 72, 360-365 (2001), which is incorporated by reference in its entirety. In view of this, the steel cloths were first treated by “striking” with nickel. A Wood's nickel strike solution, which consists of 1 M NiCl₂and 1 M HCl in water, was used. Typically, a piece of steel cloth that is 3 cm by 5 cm is taken and submerged to have 2.5 cm by 5 cm in the nickel strike solution. The cloth was used as the working electrode while a piece of nickel foil was used as the soluble counter electrode in an undivided beaker cell. The cloth was pretreated by applying an oxidative current of 15 mA cm_geom ⁻²for 30 seconds, immediately after which a reductive current of 30 mA cm_geom ⁻²was applied for 5 minutes obtain a nickel-plated stainless steel cloth. The cloth was thoroughly rinsed with DI water and dried in air at 80° C. The cell potential required for nickel plating was typically ˜1 V. Some of the current went toward hydrogen evolution, evidenced by gas evolution on the cloth; assuming 90% FE toward nickel plating, the resulting nickel layer is approximately 3 μm thick. The cloths visibly change colors after striking with nickel (FIGS. 18A-18B).
After striking the cloth with nickel, platinum can be deposited. The nickel-stricken cloth was cut into smaller pieces to submerge ˜1.5 cm by 1.5 cm into 10 mL of a platinum plating solution. The platinum plating solution used was a citrate-ammonium bath, chosen for its high current efficiency toward platinum plating, low current density required, and the non-hygroscopic nature of the platinum precursor. See, for example, Rao, C. R. K. & Trivedi, D. C. Chemical and electrochemical depositions of platinum group metals and their applications. Coord. Chem. Rev. 249, 613-631 (2005); and Baumgartner & Raub. The Electrodeposition of Platinum and Platinum Alloys. Platin. Met. Rev. 32, 188-197 (1988), each of which is incorporated by reference in its entirety. The bath contained 35 mM (NH₄)₂PtCl₆, 400 mM trisodium citrate, and 75 mM of NH₄Cl. The nickel-stricken cloth is used as the working electrode; a piece of platinum foil is used as a soluble counter electrode in a beaker cell, which is kept over a water bath at 90° C. It is possible that the (NH₄)₂PtCl₆will not fully dissolve until the solution reaches 90° C. A constant reductive current of 10 mA (˜5 mA cm_geom ⁻²) was applied to the cloth for 5 minutes. The cell potential required for platinum plating is typically ˜1.7-1.8 V. If the potential was lower than ˜1.7 V, then a higher current, up to 20 mA, was applied; this is common for fresh baths. The cloth should turn darker after platinum plating (FIGS. 18A-18B). The cloth was thoroughly rinsed with DI water and dried at 80° C. A 14 mm diameter circle was cut out to be used as the HOR electrode, and was typically used only once per experiment. The entire platinum deposition procedure is unoptimized and may require further improvement for practical applications.
Ammonia contamination in the cathode compartment from the Pt/SSC anode is unlikely as using Pt foils (in 3-compartment experiments as opposed to Pt/SSC in 4-compartment and undivided experiments) which have not been in contact with the ammonium bath as the anode also leads to production of ammonia in NRR experiments (FIG. 13B). Ar and isotope labeling experiments confirm ammonia production via nitrogen reduction (FIGS. 15A-15C).
Hydrogen oxidation experiments
Hydrogen oxidation experiments focused on studying HOR were performed in 3-compartment cells, the setup of which is detailed in Assembly of a gas diffusion cell. The cell parts and separators were washed with acetone (to remove residual ferrocene) and water and dried in an oven at 80° C. between experiments. A steel foil was used as the cathode, while Pt-coated steel cloths or Pt/C carbon paper disks were used as the anode; both the anode and cathode materials were fresh in every experiment and used only once. 10 sccm of H₂was flowed past the anode after saturating the electrolyte with H₂. Regardless of current applied, 7.2 coloumbs of charge were passed though the cell, after which hydrogen oxidation Faradaic efficiency was quantified as described below in Quantification of hydrogen oxidation. In certain experiments (FIG. 12B), no pressure control in the gas compartment was used: the hydrogen gas was vented directly to a continuously operating fume hood. Some control experiments were performed where N₂gas was fed instead of H₂gas to the anode; the HOR FE was found to be zero to within error (FIGS. 36A-36D).
Quantification of hydrogen oxidation via mass balance
In cases where the flowrate of gas past GDEs is low and the applied current is high, the conversion of the gas may be high (FIGS. 25A-25D). The nominal conversion 4 in this case is defined in Equation 3.
$\begin{matrix} ξ = \frac{I_{applied} \cdot FE}{n F \cdot Q_{applied} \cdot \frac{P_{0}}{{RT}_{0}}} & # (3) \end{matrix}$
In Equation 3, I_appliedis the applied current density, FE is the Faradaic efficiency towards the reaction of interest, n is the number of electrons involved in the electrochemical reaction (n=2 for HOR, n=6 for NRR), F is Faraday's constant, Q_appliedis the flowrate of gas set using the flow controller, R is the universal gas constant, T₀and P₀are the standard temperature and pressure (273K, 1 bar), respectively, used if the Q_appliedis given in units standard volumetric flow units (e.g. sccm).
The conversion can be estimated by measuring the outlet flowrate of gas and by using a mass balance over the gas in the gas compartment, given in Equation 4.
Q _in −Q _out =ξQ _in #(4)
From Equations 3 and 4, the Faradaic efficiency of gas conversion can be computed. Implicitly, one can assume that the gas compartment does not have leaks, and that gas dissolution into the electrolyte is negligible. Practically, this process may be used to quantify hydrogen oxidation, as the conversions can be very high.
The flowrate of gas leaving the compartment is difficult to quantify using a flow controller as it is directed to a pressure-controlling water column which releases the gas to the ambient environment and adds additional back pressure which may be difficult to control (FIGS. 32A-32D). Therefore, to quantify the flowrate of gas leaving the gas compartment, the time between bubbles of gas detaching the tubing in the water column (FIGS. 32A-32D) was measured by an electronic stopwatch. This method assumes that the volume of gas bubbles is fairly constant and independent of flowrate, i.e. that they only detach when they reach a critical volume. The critical bubble volume was found to change with tubing orientation and gas, and so only results in the same set can be compared directly. The bubble size can be “calibrated” by measuring the interval between bubble detachments at a given flowrate when no current is applied. Thus, the outlet flowrate was computed using Equation 5.
$\begin{matrix} Q_{out} = \frac{Q_{nominal in} \cdot t_{no current}}{t_{with current}} & # (5) \end{matrix}$
The HOR FE quantification experiments were performed using 3-compartment cells with a Pt/SSC anode, Daramic separator, and steel cathode. The electrolyte used was either 1 M LiBF₄/0.11 M EtOH/THF (FIGS. 24A-24B) or 1 M LiBF₄in 9:1 PC/DMC (FIGS. 26A-26D). First, the average time for bubble detachment was measured when flowing 0.5 sccm of the desired gas (N₂for controls and H₂for HOR quantification) without applying current. Then, 25 mA of current was applied without changing tubing configuration or flowrate, and the average time for bubble detachment was recorded.
It was found that when N₂is fed to the cell, the interval between bubbles does not significantly change for either tested electrolyte when current is applied (FIGS. 24A-24B and FIGS. 26A-26D), while the interval increases when H₂is fed and current is applied (FIGS. 24A-24B). The Faradaic efficiency for HOR can be estimated from these data; it was found to be 105±2% for oxidation using THF-based electrolyte (FIGS. 24A-24B and FIG. 26) and 112±19% for PC-based electrolyte (FIGS. 26A-26D).
A limitation with the method of mass balancing was found, however. When hydrogen is fed a rate of 0.2 sccm to a 3-compartment cell with a Pt/SSC anode and THF-based electrolyte and 25 mA of current are applied, there should be some gas leaving the gas compartment, as 25 mA corresponds to a hydrogen oxidation rate of at most 0.176 sccm at 100% FE. However, it was observed that no bubbles are evolved and the level of the total amount of gas decreases with time; a constant amount of gas with 25 mA of applied current is obtained when H₂is fed at a rate of 0.22-0.23 sccm. While this demonstrates the limitation of the mass balance method and possible unaccounted for sources of hydrogen depletion, hydrogen is oxidized at high rates in this system with close to unity FE.
eHB experiments and coupling to water splitting
A 4-compartment cell with a steel cloth cathode and Pt/SSC anode was assembled as described in Assembly of a gas diffusion cell. The operation of the 4-compartment cell was similar to operation of the 3-comparment cells in NRR and HOR experiments. 3.5 mL of 1 M LiBF₄, 0.11 M EtOH in THF was added to the 4-compartment cell (1.75 mL to each compartment), while 10 sccm of THF-rich N₂and 10 sccm of THF-rich H₂were fed to the cathode and anode compartments, respectively. The solutions were saturated with their respective gases for 10 minutes by flowing gas through the SSCs. The pressure gradient across the SSCs was lowered using a water column to 1 kPa, at which point the gas flowed past the SSC. 25 mA were applied to the cell for 4.8 minutes, after which the ammonia content of the cathode chamber was analyzed as described in Nitrogen reduction experiments.
In long term experiments (FIG. 13B), 20 mA were applied for 1 hour in a 3-compartment cell with a platinum foil anode or 1-2 hours in a 4-compartment eHB cell. Photographs of the anolyte solutions were taken. In long duration experiments to assess the efficacy of nitrogen reduction, eHB reactors were used. In certain experiments (FIGS. 27A-27B), the Daramic was replaced with a thinner Celgard separator to promote diffusion between the electrolyte compartments, or the separator was removed completely. 20 mA of current were applied to the reactor, while 5 sccm of N₂and H₂flowed past the cathode and anode GDEs, respectively.
To couple the eHB to water splitting, a commercially available water-splitting cell (Fuel Cell Technologies) was assembled. The electrodes were part of a membrane electrode assembly (MEA) purchased from FuelCellStore with an electrode area of 5 cm². The cathode side was platinum black with a loading of 3 mg cm⁻²; the anode side was iridium ruthenium oxide with a loading of 3 mg cm⁻²; both electrodes were on a Nafion 115 membrane. The bolts on the electrolyzer were tightened with a torque wrench with a torque of 40 lb-in.
Milli-Q water was fed continuously to the anode of the water splitting cell at ˜70 mL/min with a peristaltic pump. A constant current of 200 mA was applied across the electrolyzer; the voltage required was 1.59 V. This corresponds to an output H₂flowrate of 1.5 sccm. The cell was slightly angled to help oxygen bubbles to leave the anode compartment. The cathode compartment was sealed off at one end to force hydrogen to flow in a single direction. The hydrogen was first fed to a vial containing magnesium sulfate (MgSO₄) to capture some of the moisture in the gas stream, after which it was fed to a vial with THF and molecular sieves to saturate the gas with THF. The cell was then operated analogously to the way an eHB cell was, with the difference that the feed rate of H₂was 1.5 sccm, as defined by the water splitting current.

DISCUSSION

Computing the diffusion-limited current density for H₂oxidation
One can estimate the diffusion-limited current density for H₂oxidation and find that it is fairly close to the value one can obtain via direct measurement (FIG. 10B). The diffusion-limited current density can be computed via Equation 6. See, for example, Bard, A. J. & Faulkner, L. R. Electrochemical Methods. Fundamentals and Applications. (John Wiley & Sons, Inc, 2001), which is incorporated by reference in its entirety.
$\begin{matrix} J_{\lim} = \frac{2 D_{H_{2}} C_{H_{2}} F}{δ} & # (6) \end{matrix}$
The solubility of hydrogen in pure THF is 3.3-3.4 mM.^9,10See, for example, Gibanel, F., López, M. C., Royo, F. M., Santafé, J. & Urieta, J. S. Solubility of nonpolar gases in tetrahydrofuran at 0 to 30° C. and 101.33 kPa partial pressure of gas. J. Solution Chem. 22, 211-217 (1993); and Brunner, E. Solubility of Hydrogen in 10 Organic Solvents at 298.15, 323.15, and 373.15 K. J. Chem. Eng. Data 30, 269-273 (1985), each of which is incorporated by reference in its entirety. However, the properties of the solvent change with addition of large amounts of electrolyte, leading to a “salting-out” effect, decreasing the solubility of the gas. See, for example, Weisenberger, S. & Schumpe, A. Estimation of Gas Solubilities in Salt Solutions at Temperatures from 273 K to 363 K. AIChE J. 42, 298-300 (1996), which is incorporated by reference in its entirety. One can estimate that the solubility of hydrogen in the electrolyte is close to half of its pure solvent solubility, approximately 1.7±0.8 mM, following Lazouski et al. The diffusivity of hydrogen in the electrolyte is also an estimate, computed by using an approximate value of the viscosity of solution, and assumed to be 3.8±0.8-10⁻⁹m²s⁻¹. The diffusion boundary layer thickness was previously measured and found to be 50±15 μm;¹corrections to the diffusion boundary layer thickness due to differences in diffusion coefficients of various species are not used due to the already large uncertainties in estimates of other parameters. Combining these assumptions, the estimated diffusion-limited current density for H₂oxidation is 2.5±1.5 mA cm⁻², which is fairly close to the experimentally measured value (˜2.75 mA cm⁻², FIG. 10B).
Development of the gas-liquid interface across the SSC
FIGS. 17A-17G depict the development of the gas-liquid interface across the vertical standing SSC and carbon-cloth (CC) based GDEs. As the figures may not tell the complete picture, the process is described in words below.
Initially, the gas compartment and electrolyte compartments contain no electrolyte and are separated by a vertically standing SSC or CC GDE. When gas is fed to the gas compartment, it predominantly leaves through the GDE if any resistance to flow (by means of water column or otherwise) is applied to the outlet of the gas compartment. When a small volume of electrolyte (880 uL) is added, the SSC and CC GDEs behave differently.
In the case of the CC GDE, the electrolyte completely stays within the electrolyte compartment and wets the GDE, and gas begins to flow out through the gas compartment outlet if there is not sufficient pressure to force the gas through the GDE. Adding additional electrolyte (for a total of 1.75 mL) does not qualitatively change the picture.
In the case of a SSC GDE, the electrolyte actually goes through the vertically standing GDE into the gas compartment (FIGS. 17A-17G). The gas usually escapes through the top of the SSC that has not been wet by electrolyte if any resistance to flow is applied to the outlet of the gas compartment. When additional electrolyte is added (1.75 mL total), there is sufficient electrolyte between the two compartments to fully wet the SSC and prevent gas flow through the GDE until a pressure gradient with a magnitude greater than or equal to the magnitude of the Laplace pressure of the GDE is reached. The gas therefore temporarily accumulates in the gas compartment, and increases the absolute pressure in the gas compartment, evidenced by motion of the gas-liquid boundary in the pressure controlling burette (FIGS. 32A-32D). This elevated pressure actually pushes the electrolyte in the gas compartment through the SSC back into the electrolyte compartment. If the pressure gradient across the SSC reaches the Laplace pressure of the SSC, gas begins to flow through the SSC (FIGS. 17A-17G and FIGS. 32A-32D). If it is lower, gas flows past the SSC out of the gas compartment (FIGS. 17A-17G and FIGS. 32A-32D).
Computing the diffusion-limited current density of N₂reduction
The diffusion-limited current density for nitrogen reduction in a 1 M LiBF₄in THF electrolyte has been estimated in Lazouski et al. By following a procedure similar to the one outlined above and in the aforementioned work, one can estimate the diffusion-limited current density in an aqueous electrolyte at a flooded electrode by using Equation 7.
$\begin{matrix} J_{\lim} = \frac{6 D_{N_{2}} C_{N_{2}} F}{δ} & # (7) \end{matrix}$
The diffusion coefficient and solubility of nitrogen in pure water are well known. At 25° C. and 1 bar of N₂partial pressure, the diffusivity of N₂in water is 2.01±0.1·10⁻⁹m²s⁻¹, while the solubility is 0.66 mM. See, for example, Ferrell, R. T. & Himmelblau, D. M. Diffusion coefficients of nitrogen and oxygen in water. J. Chem. Eng. Data 12, 111-115 (1967); and Battino, R., Rettich, T. R. & Tominaga, T. The Solubility of Nitrogen and Air in Liquids. J. Phys. Chem. Ref Data 13, 563-600 (1984), each of which is incorporated by reference in its entirety. The diffusion boundary layer thickness depends heavily on the hydrodynamics of the electrolyte; typical values for CO₂reduction in an aqueous electrolyte are 60-160 μm; the boundary layer thickness may be thinner is well-defined and vigorous hydrodynamics are observed, such as in systems utilizing rotating disk electrodes (RDEs). See, for example, Weng, L. C., Bell, A. T. & Weber, A. Z. Modeling gas-diffusion electrodes for CO₂reduction. Phys. Chem. Chem. Phys. 20, 16973-16984 (2018), which is incorporated by reference in its entirety. From these values, it was found that the diffusion limited current density for nitrogen reduction in water is a mere 0.48-1.27 mA cm⁻².
Source of ammonia in two phases
It was found that produced NH₃can be found in both the solution and the gas phases (FIGS. 12C and 12D). The rate at which NH₃is produced in the gas phase is almost independent of the applied current density, which implies that the FE for NH₃in the gas phase decreases with applied current density. It is unlikely that the NH₃is formed directly in the gas phase, as both the lithium and proton source required for NH₃formation are in solution. The source of NH₃in the gas phase may be stripping of the solution of NH₃by the gas flowing past the SSC. This would imply that NH₃is much more concentrated at the electrode than in the bulk. The general lack of forced mixing in the electrolyte supports this hypothesis. No ammonia is found in the gas phase when gas flows through the electrode (FIG. 12D). Any gas phase ammonia would be diluted in the bulk solution and thoroughly mixed, minimizing the amount of stripping.
The lack of stripping of ammonia from the bulk solution is likely due to strong interactions between Li⁺ and NH₃, stronger than between Li⁺ and THF. See, for example, Kaufmann, E., Gose, J. & Schleyer, P. v. R. Thermodynamics of solvation of lithium compounds. A combined MNDO and ab initio study. Organometallics 8, 2577-2584 (1989), which is incorporated by reference in its entirety. The formation Li—NH₃complexes in solution increases the solubility of ammonia in the electrolyte (FIGS. 39A-39D), reducing the ammonia partial pressure at a given NH₃concentration.
Estimating energy efficiency and consumption
A detailed description for estimating the energy efficiency of the process and sources of energy losses has been described in Lazouski et al. Briefly, the energy efficiency is computed as follows by the formula given in Equation 8.
$\begin{matrix} η = \frac{U_{NH_{3}} \cdot FE}{V_{total}} & # (8) \end{matrix}$
In Equation 8, U_NH ₃is the standard potential for the reaction of ammonia oxidation, i.e. the one likely to be used to extract useful work (4NH₃+3O₂→2N₂+6H₂O), V_totalis the total applied potential across the cell, and FE is the Faradaic efficiency. In order to compute the energy consumption, Equation 9 was used.
$\begin{matrix} P = \frac{3 F \cdot V_{total}}{FE \cdot M_{NH_{3}}} & # (9) \end{matrix}$
In Equation 9, F is Faraday's constant (96485 C/mol), V_totalis the total applied cell voltage, FE is the ammonia Faradaic efficiency, and M_NH ₃is the molar mass of ammonia (17 g/mol).

TABLE 1

Compilation of literature highest reported rates obtained in nonaqueous electrolytes
at close-to-ambient conditions. The data are plotted in FIG. 13A.

			Highest NH₃	NH₃FE at
Cathode			production rate,	highest	Reference
material	Electrolyte	Conditions	nmol cm⁻²s⁻¹	rate, %	(listed below)

Ag—Au@ZIF	0.2M LiCF₃SO₃, ~1%	Ambient	0.010	18.0	17
	EtOH, THF
Stainless steel	[C₄mpyr][eFAP]	Ambient	0.021	34.0	18
cloth, flooded
Copper foil	1M LiBF₄, 0.2M	Ambient	7.88	15.2	1
	EtOH, THF
Silver foil	0.2M LiClO₄, 0.18M	Ambient	0.58	8.4	19
	EtOH, THF
Iron foil	0.2M LiClO₄, 0.18M	50 bar N₂,	3.99	57.7
	EtOH, THF	25° C.
Molybdenum	0.2M LiClO₄, 0.18M	Ambient	0.22	7.5	20
foil	EtOH, THF
Nickel foil	0.1M LiCl in EDA	Ambient	0.036	17.2	21
Porous nickel	0.01M H₂SO₄in 9:1	Ambient	0.015	0.85	22
	2-propanol/water
Polyaniline on	0.03M H₂SO₄, 0.1M	50 bar N₂,	0.44	16	23
platinum	LiClO₄, MeOH	25° C.
Nickel	LiClO₄, EtOH, THF	Ambient	0.26	3.8	24
Nickel	LiClO₄, THF, EtOH,	Ambient	0.11	1.7
	PYR-14 TFSI Ionic
	Liquid
Stainless steel	1M LIBF4, 0.1M	Ambient	30.4	35.3	Present work
cloth, GDE	EtOH

TABLE 2

Raw data for current density variation for hydrogen oxidation
on various substrates. The data are plotted in FIG. 12A.

Current		Ferrocenium
density,		concentration,	Ferrocenium	HOR
mA cm⁻²	Electrode/condition	mM	FE, %	FE, %

5	Pt/C, 0 kPa	5.87	78.21	21.79
5	Pt/C, 0 kPa	7.39	98.57	1.43
10	Pt/C, 0 kPa	7.82	104.33	−4.33
10	Pt/C, 0 kPa	6.27	83.56	16.44
15	Pt/C, 0 kPa	6.57	87.57	12.43
15	Pt/C, 0 kPa	6.35	84.65	15.35
20	Pt/C, 0 kPa	6.87	91.67	8.33
20	Pt/C, 0 kPa	5.25	69.99	30.01
25	Pt/C, 0 kPa	5.90	78.62	21.38
25	Pt/C, 0 kPa	7.48	99.82	0.18
5	Pt/C 5 kPa	7.41	98.81	1.19
5	Pt/C 5 kPa	8.22	109.80	−9.80
10	Pt/C 5 kPa	8.76	116.95	−16.95
10	Pt/C 5 kPa	7.14	95.32	4.68
15	Pt/C 5 kPa	6.83	91.14	8.86
15	Pt/C 5 kPa	6.34	84.55	15.45
20	Pt/C 5 kPa	8.30	110.84	−10.84
20	Pt/C 5 kPa	6.90	92.03	7.97
25	Pt/C 5 kPa	8.26	110.33	−10.33
25	Pt/C 5 kPa	6.13	81.76	18.24
5	Pt/C 20 kPa	0.03	−0.05	100.05
5	Pt/C 20 kPa	0.03	0.00	100.00
10	Pt/C 20 kPa	0.34	4.11	95.89
10	Pt/C 20 kPa	0.40	4.95	95.05
15	Pt/C 20 kPa	1.79	23.51	76.49
15	Pt/C 20 kPa	1.47	19.25	80.75
20	Pt/C 20 kPa	3.90	51.80	48.20
20	Pt/C 20 kPa	3.15	41.83	58.17
25	Pt/C 20 kPa	5.28	70.32	29.68
25	Pt/C 20 kPa	5.32	70.88	29.12
5	Pt/SCC, 1 kPa	0.060	0.37	99.63
5	Pt/SCC, 1 kPa	0.047	0.20	99.80
10	Pt/SCC, 1 kPa	0.040	0.10	99.90
10	Pt/SCC, 1 kPa	0.035	0.03	99.97
15	Pt/SCC, 1 kPa	0.036	0.05	99.95
15	Pt/SCC, 1 kPa	0.038	0.08	99.92
20	Pt/SCC, 1 kPa	0.081	0.65	99.35
20	Pt/SCC, 1 kPa	0.129	1.30	98.70
25	Pt/SCC, 1 kPa	0.100	0.90	99.10
25	Pt/SCC, 1 kPa	0.043	0.14	99.86

TABLE 3

Raw data for pressure gradient variation for hydrogen
oxidation on Pt/SSC. The data are plotted in FIG. 12B.

Pressure	Ferrocenium
gradient,	concentration,	Ferrocenium	HOR
kPa	mM	FE, %	FE, %

0	6.54	87.28	12.72
0	7.24	96.63	3.37
0.5	0.08	0.60	99.40
0.5	0.34	4.14	95.86
1	0.02	−0.12	100.12
1	0.08	0.65	99.35
1.5	0.07	0.49	99.51
1.5	0.10	0.90	99.10
1.9	0.05	0.21	99.79
1.9	0.11	1.01	98.99

TABLE 4

Raw data for current density variation for nitrogen reduction
on steel cloths. The data are plotted in FIG. 12C.

Current	NH₃
density,	solution	NH₃gas	NH₃total	NH₃solution rate,	NH₃gas rate,	NH₃total rate,
mA cm⁻²	FE, %	FE, %	FE, %	nmol cm⁻²s⁻¹	nmol cm⁻²s⁻¹	nmol cm⁻²s⁻¹

5	22.59	9.65	32.25	3.90	1.67	5.57
5	17.51	10.58	28.09	3.02	1.83	4.85
5	21.78	15.15	36.92	3.76	2.62	6.38
10	28.88	4.92	33.81	9.98	1.70	11.68
10	37.66	9.94	47.60	13.01	3.43	16.44
10	35.80	9.01	44.81	12.37	3.11	15.48
15	36.13	7.50	43.63	18.72	3.89	22.61
15	36.74	4.10	40.84	19.03	2.12	21.16
15	31.84	5.74	37.58	16.50	2.98	19.47
20	30.76	5.38	36.14	21.25	3.71	24.97
20	40.23	4.85	45.07	27.79	3.35	31.14
20	29.46	3.27	32.74	20.35	2.26	22.62
20	33.47	8.14	41.62	23.13	5.62	28.75
25	36.09	5.52	41.61	31.17	4.76	35.93
25	33.43	4.88	38.31	28.87	4.22	33.09
25	26.70	2.25	28.94	23.05	1.94	24.99
25	30.01	2.14	32.15	25.92	1.85	27.77

TABLE 5

Raw data for pressure gradient variation across the electrode for nitrogen
reduction on steel cloths. The data are plotted in FIG. 12D.

Pressure
gradient,	NH₃solution	NH₃gas	Total	NH₃solution rate,	NH₃gas rate,	NH₃total rate,
kPa	FE, %	FE, %	FE, %	nmol cm⁻²s⁻¹	nmol cm⁻²s⁻¹	nmol cm⁻²s⁻¹

0	2.32	—	2.32	1.20	—	1.20
0	4.22	—	4.22	2.19	—	2.19
0.5	44.00	6.20	50.20	22.80	3.21	26.01
0.5	41.72	3.12	44.84	21.62	1.62	23.24
1.0	20.67	6.74	27.41	10.71	3.49	14.20
1.0	36.13	7.50	43.63	18.72	3.89	22.61
1.0	36.74	4.10	40.84	19.03	2.12	21.16
1.5	31.84	5.74	37.58	16.50	2.98	19.47
1.5	32.36	5.85	38.21	16.77	3.03	19.80
1.7	40.11	3.01	43.12	20.78	1.56	22.34
1.7	42.15	0.09	42.25	27.93	0.08	28.01

References, each of which is incorporated by reference in its entirety.

1. Lazouski, N., Schiffer, Z. J., Williams, K. & Manthiram, K. Understanding Continuous Lithium-Mediated Electrochemical Nitrogen Reduction. Joule 3, 1127-1139 (2019).
2. Nielander, A. C. et al. A Versatile Method for Ammonia Detection in a Range of Relevant Electrolytes via Direct Nuclear Magnetic Resonance Techniques. ACS Catal. 9, 5797-5802 (2019).
3. Stoychev, D., Papoutsis, A., Kelaidopoulou, A., Kokkinidis, G. & Milchev, A. Electrodeposition of platinum on metallic and nonmetallic substrates—selection of experimental conditions. Mater. Chem. Phys. 72, 360-365 (2001).
4. Rao, C. R. K. & Trivedi, D. C. Chemical and electrochemical depositions of platinum group metals and their applications. Coord. Chem. Rev. 249, 613-631 (2005).
5. Baumgartner & Raub. The Electrodeposition of Platinum and Platinum Alloys. Platin. Met. Rev. 32, 188-197 (1988).
6. Gagne, R. R., Koval, C. A. & Lisensky, G. C. Ferrocene as an Internal Standard for Electrochemical Measurements. Inorg. Chem. 19, 2854-2855 (1980).
7. Singh, A., Chowdhury, D. R. & Paul, A. A kinetic study of ferrocenium cation decomposition utilizing an integrated electrochemical methodology composed of cyclic voltammetry and amperometry. Analyst 139, 5747-5754 (2014).
8. Bard, A. J. & Faulkner, L. R. Electrochemical Methods. Fundamentals and Applications. (John Wiley & Sons, Inc, 2001).
9. Gibanel, F., López, M. C., Royo, F. M., Santafé, J. & Urieta, J. S. Solubility of nonpolar gases in tetrahydrofuran at 0 to 30° C. and 101.33 kPa partial pressure of gas. J. Solution Chem. 22, 211-217 (1993).
10. Brunner, E. Solubility of Hydrogen in 10 Organic Solvents at 298.15, 323.15, and 373.15 K. J. Chem. Eng. Data 30, 269-273 (1985).
11. Weisenberger, S. & Schumpe, A. Estimation of Gas Solubilities in Salt Solutions at Temperatures from 273 K to 363 K. AIChE J. 42, 298-300 (1996).
12. Ferrell, R. T. & Himmelblau, D. M. Diffusion coefficients of nitrogen and oxygen in water. J. Chem. Eng. Data 12, 111-115 (1967).
13. Battino, R., Rettich, T. R. & Tominaga, T. The Solubility of Nitrogen and Air in Liquids. J. Phys. Chem. Ref Data 13, 563-600 (1984).
14. Weng, L. C., Bell, A. T. & Weber, A. Z. Modeling gas-diffusion electrodes for CO₂reduction. Phys. Chem. Chem. Phys. 20, 16973-16984 (2018).
15. Kaufmann, E., Gose, J. & Schleyer, P. v. R. Thermodynamics of solvation of lithium compounds. A combined MNDO and ab initio study. Organometallics 8, 2577-2584 (1989).
16. Lobaccaro, P. et al. Effects of temperature and gas-liquid mass transfer on the operation of small electrochemical cells for the quantitative evaluation of CO₂reduction electrocatalysts. Phys. Chem. Chem. Phys. 18, 26777-26785 (2016).

17. Lee, H. K. et al. Favoring the unfavored: Selective electrochemical nitrogen fixation using a reticular chemistry approach. Sci. Adv. 4, eaar3208 (2018).

18. Zhou, F. et al. Electro-synthesis of ammonia from nitrogen at ambient temperature and pressure in ionic liquids. Energy Environ. Sci. 10, 2516-2520 (2017).
19. Tsuneto, A., Kudo, A. & Sakata, T. Lithium-mediated electrochemical reduction of high pressure N₂to NH₃ . J. Electroanal. Chem. 367, 183-188 (1994).
20. Andersen, S. Z. et al. A rigorous electrochemical ammonia synthesis protocol with quantitative isotope measurements. Nature 570, 504-508 (2019).
21. Kim, K., Yoo, C., Kim, J., Yoon, H. C. & Han, J. Electrochemical Synthesis of Ammonia from Water and Nitrogen in Ethylenediamine under Ambient Temperature and Pressure. J. Electrochem. Soc. 163, F1523-F1526 (2016).
22. Kim, K. et al. Communication—Electrochemical Reduction of Nitrogen to Ammonia in 2-Propanol under Ambient Temperature and Pressure. J. Electrochem. Soc. 163, F610-F612 (2016).
23. Köleli, F. & Röpke, T. Electrochemical hydrogenation of dinitrogen to ammonia on a polyaniline electrode. Appl. Catal. B Environ. 62, 306-310 (2006).
24. Pappenfus, T. M., Lee, K., Thoma, L. M. & Dukart, C. R. Wind to Ammonia: Electrochemical Processes in Room Temperature Ionic Liquids. in ECS Transactions vol. 16 89-93 (ECS, 2009).

Another example of a standalone electrode architecture is depicted in FIG. 42A. A version of the drawing with see-through edges is shown in FIG. 42B. The pieces, left to right, are: gas compartment 100, compartment O-ring 102, current collector 103, GDE material (mesh) 104, holder O-ring 106, and holder 108. These components form an enclosed gas electrode as describe herein. Ports 110 provide gas access to the electrode surfaces, allowing exposure to reactants and removal of products.
Referring to FIG. 43, a flow in electrode configuration can include a power connection 120 can enter the electrode to access the electrode surface within a flow-in electrode 140. A gas inlet 150 provides a fluid connection of gas through port 110. A pressure regulator 160 can provide control of pressure within the standalone electrode. For example, the pressure regulator can be a water column or other fluid column.
Referring to FIG. 44, a flow past electrode configuration can include a power connection 120 can enter the electrode to access the electrode surface within a flow-past electrode 190. A gas inlet 150 provides a fluid connection of gas through a port 110. A pressure regulator 160 can provide control of pressure within the standalone electrode. For example, the pressure regulator can be a water column or other fluid column. Gas outlet 180 can exit a port 110 and through the pressure regulator 160.
Referring to FIG. 45, an exemplary electrochemical reaction system utilizing a flow-in standalone GDE electrode and a flow-past standalone GDE electrode is shown. The system 200 can include a flow-in standalone GDE electrode 140 and a flow-past standalone GDE electrode 190, which are immersed in an electrolyte 220. Flow-in electrode 140 can transform a gas flow to a reactant gas that can then be used in conversion of a reactant to a product in the flow-past electrode 190.
An example is further described here.
Stand-alone electrode for utilizing sparingly soluble gases
Metallic meshes and other porous materials as gas diffusion electrodes can generally utilize sparingly soluble gases in electrochemical reactions in aqueous and nonaqueous electrolytes. However, in some applications, the gas diffusion electrodes are utilized in a custom, parallel-electrode architecture. While the architecture is efficient and convenient for both testing and synthetic applications, there may be applications that require utilization of sparingly soluble gases in other architectures. Examples include rapid resting of reactions in easy-to-setup beaker cells, synthetic reactions which require large volumes of solvent, and those which use gases in counter and balancing reactions. For these and other applications, a stand-alone electrode is described herein that uses the nonaqueous GDEs for utilizing sparingly soluble gases in electrochemical reactions.
The basic standalone architecture can be seen in FIGS. 42A and 42B. The architecture consists of a gas compartment, a current collector, the active gas diffusion electrode, and a holder piece, with O-rings between relevant parts of the setup. The gas compartment is meant to provide a separate reactive gaseous phase at the electrode. The current collector is immediately in contact with the gas compartment via an O-ring. The gas compartment contains an electrical connection—a wire of copper, aluminum, iron, nickel, or other metal, permanently or non-permanently fixed to the inside walls of the gas compartment. The current collector goes around the entire edge of the gas compartment opening and is slightly wider at the outer edge than the O-ring to maintain a good seal. The GDE is then in immediately contact with the current collector. The inner edge of the current collector is smaller than the GDE to have uniform electrical contact with the GDE around the edge. The connector has a groove to hold another O-ring which surrounds the GDE. The GDE is supported by a lip on the connector piece—the construction is effectively gas-tight due to O-ring contacting, and allows gas only through the electrode. The assembly can be held together with bolts made of metal or plastic, preferably plastic so as to avoid contamination of metals in the electrolyte. Other bolt-less configurations, some where the second, holder piece of plastic is screwed on using threading, are possible.
There are a number of possible configurations of the standalone electrode. In particular, the electrode may be in a “flow in” configuration, where the gas is fed through one inlet and used up by the reactions occurring at the electrode (FIG. 43), in a “flow past” configuration, where the gas enters the gas compartment through an inlet and exits as either unreacted gas or gaseous products through a separate outlet (FIG. 44), or a “flow through” configuration, where the gas is forced through the GDE into the electrolyte solution. In all these configurations, the electrical contact may enter the gas compartment either through one of the gas inlets, if chemical compatibility allows, or through a separate, gas- and liquid-tight inlets, e.g. via a septum. In FIGS. 43 and 44, the electrical contact is depicted to enter through the gas inlet.
One of the key requirements for invention operation is controlling the pressure gradient across the GDE. The location and configuration of the pressure control differ somewhat in the three gas flow configurations. In the “flow-through” configuration, the GDE itself establishes the necessary pressure gradients, and simply flowing gas into the gas compartment at a high enough rate and/or positive pressure is sufficient.
In the “flow-past” configuration, the pressure in the gas compartment is controlled after the gas leaves standalone electrode (FIG. 44). In one implementation, a simple water column is sufficient to control the pressure inside the gas compartment. Typically, the gas is continuously flowed into and out of the gas compartment into a water column, which adds additional positive pressure to the gas. More advanced pressure regulation methods may be used, such as a backpressure regulator. Further operations with the outlet can be performed after pressure control, assuming they do not add additional back-pressure to the gas.
In the “flow-in” configuration, the pressure in the gas compartment is controlled prior to entering electrode (FIG. 43). In this case, the gas stream is branched, with one stream entering the electrode, while the other is used to control the gas pressure of all the gas in the system. A static water column or more advanced pressure regulators can be used to maintain a constant pressure of gas. If the gas flow into the electrode is equal to its rate of utilization in the electrochemical reaction (and any leaks), then gas does not have to leave via the pressure-controlling system. However, if excess gas enters the system, then the excess gas can be expelled through the pressure regulator, as to avoid “flow-through” behavior at the GDE. In this case, the GDE will utilize a fraction of the gas fed to the system by pulling it into the electrode compartment.
Demonstration of standalone electrode use
A flow-in standalone electrode with a platinum-coated stainless-steel cloth GDE was assembled to demonstrate the capability of the electrode to utilize gases in electrochemical reactions in various solvents. The standalone electrode was used as the anode, to which either nitrogen or hydrogen gas was fed. A platinum foil was used as the cathode and an Ag/AgCl electrode was used as a reference. Two electrolyte compositions were used sequentially with the same electrode to demonstrate the solvent-agnostic nature of the electrode: 0.1 M tetrabutylammonium tetrafluoroborate with 0.05 M hexafluoroisopropyl alcohol in acetonitrile was used as a nonaqueous electrolyte, while a 0.05 M Na₂SO₄in water solution was used as an aqueous electrolyte. Either nitrogen or hydrogen gas was first flowed into the electrode in a flow-through configuration to fill the gas compartment of the electrode and saturate the electrolyte solution, after which the pressure in the electrode gas compartment was decreased to put the electrode in a flow-in configuration. A linear-sweep voltammogram was measured by sweeping the potential from open circuit voltage to a high oxidation potential, first with nitrogen in the gas compartment, followed by having hydrogen in the compartment. A significantly higher current was obtained when using hydrogen as the feed gas when compared to using nitrogen, which demonstrates that hydrogen oxidation is occurring. The currents obtained significantly exceeded the diffusion limited hydrogen oxidation current, demonstrating the use of the standalone electrode as a gas diffusion electrode. Results of the exemplary reaction are shown in FIGS. 46A-46B.
Details of one or more embodiments are set forth in the accompanying drawings and description. Other features, objects, and advantages will be apparent from the description, drawings, and claims. Although a number of embodiments of the invention have been described, it will be understood that various modifications may be made without departing from the spirit and scope of the invention. It should also be understood that the appended drawings are not necessarily to scale, presenting a somewhat simplified representation of various features and basic principles of the invention.

Claims

What is claimed is:

1. An electrochemical system comprising:

a housing including a chamber;

an electrode within the housing; and

a gas permeable metal on a surface of the electrode in contact with the chamber.

2. The system of claim 1, further comprising a gas inlet to the housing.

3. The system of claim 1, further comprising a first outlet of the housing to release a product from the housing.

4. The system of claim 1, wherein the gas permeable metal includes a metal mesh.

5. The system of claim 4, wherein the metal mesh includes 100, 200, 300, 400, 500, 1000, 1500, or 2000 fibers per inch.

6. The system of claim 1, wherein the gas permeable metal includes openings of between 1 and 200 micrometers, preferably between 2 and 100 micrometers.

7. The system of claim 1, wherein the gas permeable metal includes metal fibers or a porous metal.

8. The system of claim 1, wherein the gas permeable metal includes stainless steel, steel, nickel, iron, copper, silver, gold, or platinum.

9. The system of claim 1, wherein the gas permeable metal includes a catalytic metal, metal oxide, metal sulfide, or metal phosphide.

10. The system of claim 1, wherein gas permeable metal is exposed to a pressure gradient.

11. A method of supplying a gas to an electrochemical system comprising:

contacting a gas with a gas permeable metal on a surface of an electrode in a chamber of a housing.

12. The method of claim 11, wherein the gas is a sparingly soluble gas.

13. The method of claim 11, further comprising supplying a pressure of the gas in the chamber to create a pressure differential at the electrode.

14. The method of claim 11, further comprising applying a voltage to the electrode.

15. The method of claim 11, wherein the gas permeable metal includes a metal mesh.

16. The method of claim 15, wherein the metal mesh includes 100, 200, 300, 400, 500, 1000, 1500, or 2000 fibers per inch.

17. The method of claim 11, wherein the gas permeable metal includes openings of between 1 and 200 micrometers, preferably between 2 and 100 micrometers.

18. The method of claim 11, wherein the gas permeable metal includes metal fibers or a porous metal.

19. The method of claim 11, wherein the gas permeable metal includes stainless steel, steel, nickel, iron, copper, silver, gold, or platinum.

20. A method of oxidizing or reducing a gas comprising:

contacting a gas with a gas permeable metal on a surface of an electrode.

21. The method of claim 20, wherein the gas is a sparingly soluble gas.

22. The method of claim 20, wherein the gas is hydrogen.

23. The method of claim 20, wherein the gas is nitrogen.

24. The method of claim 23, wherein the ammonia is produced at a Faradaic yield of at least 30% or at least 40%.

25. The method of claim 20, wherein supplying a pressure of the gas in the chamber to create a pressure differential at the electrode.

26. An electrochemical system comprising:

a first electrode including:

a housing including a chamber;

an electrode within the housing; and

a gas permeable metal on a surface of the electrode in contact with the chamber; and

a second electrode including a gas inlet to a housing including a gas permeable metal on a surface of an electrode and a first outlet to release a product from the system.

27. The system of claim 26, wherein each gas permeable metal includes a metal mesh.

28. The system of claim 27, wherein each metal mesh includes 100, 200, 300, 400, 500, 1000, 1500, or 2000 fibers per inch.

29. The system of claim 26, wherein at least one gas permeable metal includes openings of between 1 and 200 micrometers, preferably between 2 and 100 micrometers.

30. The system of claim 26, wherein at least one gas permeable metal includes metal fibers or a porous metal.

31. The system of claim 26, wherein each gas permeable metal is exposed to a pressure gradient.