Abstract
A process for converting carbon dioxide into a carbon-based molecule catalyzes a direct-conversion reaction of a vapor-fed flow of the carbon dioxide to the carbon-based molecule using a tandem electrocatalyst integrated with the gas diffusion electrode. The tandem electrocatalyst is a nanostructure composed of two parts: a copper or a copper-based binary or ternary alloy, and a metal center coordinated to nitrogen-doped carbon (NC) or a NC containing macrocyclic organic compound. In one specific implementation, the tandem electrocatalyst consists of copper and nickel-coordinated nitrogen-doped carbon (NiNC), and the carbon-based molecule is ethylene. The copper or copper-based binary or ternary alloy may be in the form of nanoparticles, nanocubes, nanotubes, nanoflowers, nanorods, or nanoplates. The metal center coordinated to nitrogen-doped carbon (NC) or to a NC containing macrocyclic organic compound may be in the form of nanoflowers, nanotubes, nanocages, mesopores, macropores, nanofibers, nanospheres, or other nanostructure.
Claims
1. A process for converting carbon dioxide into a carbon-based molecule, the process comprising: applying a working voltage to a tandem electrocatalyst integrated with a gas diffusion electrode; providing a vapor-fed flow of the carbon dioxide to the tandem electrocatalyst integrated with the gas diffusion electrode; catalyzing a direct-conversion reaction of the vapor-fed flow of the carbon dioxide to the carbon-based molecule using the tandem electrocatalyst integrated with the gas diffusion electrode.
2. The process of claim 1 wherein the tandem electrocatalyst consists of copper and nickel-coordinated nitrogen-doped carbon (NiNC), and wherein the carbon-based molecule is ethylene.
3. The process of claim 2 wherein the copper is in the form of nanoparticles, nanocubes, nanotubes, nanoflowers, nanorods, or nanoplates.
4. The process of claim 1 wherein the tandem electrocatalyst is a nanostructure composed of 1) a copper-based binary or ternary alloy, and 2) a metal center coordinated to a) nitrogen-doped carbon (NC) or b) a NC-containing macrocyclic organic compound.
5. The process of claim 4 wherein the copper-based binary or ternary alloy is in the form CuX-Y, where each of X, Y is a transition or post-transition metal.
6. The process of claim 5 wherein each of X, Y is selected from the group consisting of Ag, Zn, Al, and Sn.
7. The process of claim 4 wherein the copper-based binary or ternary alloy is in the form of nanoparticles, nanocubes, nanotubes, nanoflowers, nanorods, or nanoplates.
8. The process of claim 4 wherein the metal center is composed of a transition or post-transition metal.
9. The process of claim 8 wherein the metal center is Fe, Co, Ni, Ag, Zn, Al, or Sn.
10. The process of claim 4 wherein the macrocyclic organic compound is porphyrins or phthalocyanines with modified ligands.
11. The process of claim 4 wherein the metal center coordinated to nitrogen-doped carbon (NC) or to a NC-containing macrocyclic organic compound is in the form of nanoflowers, nanotubes, nanocages, mesopores, macropores, nanofibers, nanospheres, or other nanostructure.
12. The process of claim 1 wherein the carbon-based molecule is ethylene, ethanol, acetate, or a straight chain hydrocarbon with at least three carbon atoms.
13. The process of claim 1 implemented using a membrane electrode assembly electrolyzer.
14. The process of claim 1 implemented using a flow electrolyte electrolyzer.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
[0033] FIG. 1A is a schematic illustration of GDE-based Cu/NiNC tandem electrocatalysis for CO.sub.2 reduction, according to an embodiment of the invention.
[0034] FIGS. 1B and 1C are scanning electron microscope (SEM) images of a uniform catalyst layer of tandem Cu/NiNC GDE, according to an embodiment of the invention.
[0035] FIG. 1D shows EDS elemental maps of Cu and Ni of the cross-section image in FIG. 1C.
[0036] FIG. 1E and FIG. 1F are x-ray photoelectron spectroscopy graphs of a.u. intensity vs binding energy, illustrating Cu.sup.0, Cu.sup.1+, and Cu.sup.2+ oxidation states indicating the presence of Cu.sub.2O.
[0037] FIG. 1G is an x-ray diffraction graph of a.u. intensity vs theta, showing the x-ray diffraction patterns of the evolution of crystalline structures of Cu and Cu/NiNC tandem electrodes before electrochemical reactions.
[0038] FIG. 2A is a Faradaic efficiency graph showing higher faradaic efficiency of ethylene (FE.sub.ethylene) with the Cu/NiNC tandem electrode across all tested potentials compared to the Cu only electrode
[0039] FIG. 2B is a graph of partial current density of ethylene (j.sub.ethylene) showing higher j.sub.ethylene of the tandem electrode towards the formation of C.sub.2H.sub.4 compared to that of the Cu only electrode.
[0040] FIG. 2C is a graph of the C.sub.2H.sub.4/CO selectivity ratio obtained with Cu/NiNC and Cu electrodes.
[0041] FIG. 3A shows the CO.sub.2 consumption rate towards the formation of C.sub.2H.sub.4, on Cu and Cu/NiNC tandem electrodes.
[0042] FIG. 3B shows a graph of the CO.sub.2 consumption rate towards the formation of uncoupled CO, and partial current density towards CO on Cu and Cu/NiNC tandem electrodes.
[0043] FIG. 3C is a schematic diagram illustrating two possible CO.sub.2 reduction pathways for C.sub.2H.sub.4 formation on the tandem catalyst surface on a gas diffusion layer.
[0044] FIG. 4A and FIG. 4D are 3D and top-down schematic views, respectively, of the gas-liquid interface in the simulation domain.
[0045] FIG. 4B and FIG. 4C are plots of concentrations of CO and P, respectively, when the flux applied to catalyst particles is J=110.sup.6 kmol/m.sup.2s.
[0046] FIG. 4E and FIG. 4F are plots of dimensionless diffusive fluxes of CO and P, respectively.
[0047] FIG. 4G is a graph of the calculated area-averaged fluxes of CO from NiNC and higher-order products P through the gas-liquid interface.
[0048] FIG. 4H is a graph of the P/CO flux ratios of those fluxes as a function of the Cu.sub.2O/NiNC volume ratio.
[0049] FIG. 5A is a schematic diagram illustrating a practical utilization of the tandem catalysis using the Cu/NiNC GDE as the cathode in a MEA electrolyzer.
[0050] FIG. 5B is a graph of faradaic efficiency and cell voltage of Cu and tandem Cu/NiNC gas diffusion electrodes for CO.sub.2 reduction.
[0051] FIG. 5C is a graph of chemical energy efficiency of Cu and tandem Cu/NiNC gas diffusion electrodes.
[0052] FIG. 6 is a graph of x-ray diffraction patterns of Cu and Cu/NiNC GDEs before (pre) and after (post) electrochemical tests.
[0053] FIG. 7 is a schematic illustration of a vapor-fed CO.sub.2 flow electrolyzer setup for vapor-fed CO.sub.2 reduction using Cu/NiNC and Cu GDEs.
[0054] FIG. 8A, 8B, 8C are bar charts illustrating the distribution of gas phase products obtained with Cu.sub.xO gas diffusion electrodes with various catalyst loadings using 1 M KOH electrolyte for vapor-fed CO.sub.2 reduction. (Yellow, purple, and orange bars indicate H.sub.2, C.sub.2H.sub.4, and CO, respectively.)
[0055] FIG. 9 is a bar chart illustrating the potential dependent product selectivity evaluation of CO.sub.2 reduction of NiNC GDE in the vapor-fed CO.sub.2 flow electrolyzer.
[0056] FIG. 10A, 10B are cross-sectional SEM images of Cu GDE and Cu/NiNC GDE, respectively, used for electrochemical testing.
[0057] FIG. 11 is a bar chart illustrating Faradaic efficiency and partial current density of C.sub.2H.sub.4 obtained with the Cu GDE at various potentials.
[0058] FIG. 12A, 12B, 12C are bar charts showing product distributions obtained from Cu, Cu/NiNC-high, and Cu/NiNC-low GDEs, respectively.
[0059] FIG. 13A is a graph of Faradaic efficiency towards C.sub.2H.sub.4 obtained from Cu, Cu/NiNC-high, and Cu/NiNC-low GDEs, respectively.
[0060] FIG. 13B, 13C are graphs of partial current densities towards C.sub.2H.sub.4 and CO, respectively.
[0061] FIG. 13D is a graph of C.sub.2H.sub.4/CO selectivity ratio obtained with Cu and different tandem GDEs at various potentials.
[0062] FIG. 14A, 14B show pre-electrochemical SEM images of Cu GDE.
[0063] FIG. 14C, 14D show post-electrochemical SEM images of Cu GDE.
[0064] FIG. 14E, 14F show pre- and post-electrochemical SEM images of Cu/NiNC GDE.
[0065] FIG. 15 is a plot of cell voltage vs time, showing cell voltage stability of the optimized Cu/NiNC GDE.
[0066] FIG. 16 is a table showing catalytic performance of reported tandem catalysts in the literature.
DETAILED DESCRIPTION
[0067] In embodiments of the invention, increased ethylene formation during vapor-fed carbon dioxide reduction is achieved by employing tandem electrocatalyst consisting of nickel-coordinated nitrogen-doped carbon and cuprous oxide, which act as carbon monoxide generator and coupler, respectively. Performance evaluation was conducted in both flow electrolyte and membrane electrode assembly electrolyzers to demonstrate the potential for large-scale and direct carbon dioxide gas conversion into value-added chemicals.
[0068] A preferred embodiment of the invention provides a unique tandem gas diffusion electrode (GDE) comprising cuprous oxide nanocubes and Ni-coordinated N-doped carbons (NiNC). Described herein is also the evaluation of tandem catalysis under direct CO.sub.2 gas reduction by employing both vapor-fed flow electrolyte and membrane electrode assembly (MEA) electrolyzers. The Cu/NiNC-catalyst-loaded GDEs provide substantially increased mass transport that enables high product formation rates. Due to the different physicochemical properties of the two materials, we observed enhanced selectivity towards the formation of C.sub.2H.sub.4 during CO.sub.2R while mitigating mixing of the two under CO.sub.2 reduction conditions. Because NiNC is an excellent CO production catalyst, it significantly increases the local availability of CO near the Cu sites to promote increased CC coupling. In fact, the tandem Cu/NiNC GDE exhibits significantly higher selectivity and production rate towards the formation of C.sub.2H.sub.4 at lower overpotentials compared to the Cu-only GDE in the vapor-fed flow electrolyzer. Additionally, resolved 3-dimensional continuum simulations are performed to demonstrate the qualitative enhancement of C.sub.2+ product formations on GDEs with varying ratios of Cu and NiNC, showing that the increased internal CO concentration with higher loadings of NiNC leads to an increased C.sub.2+ flux. Lastly, the tandem GDE is evaluated in an MEA electrolyzer, demonstrating a very respectable energy efficiency towards C.sub.2H.sub.4, among the highest reported for direct CO.sub.2-to-ethylene conversion in the literature.
[0069] FIG. 1A is a schematic illustration of a GDE-based Cu/NiNC tandem electrocatalysis for CO.sub.2 reduction, according to an embodiment of the invention. CO.sub.2 flows to contact gas diffusion layer 100 which has a catalyst layer 102 on an opposite surface where there is an electrolyte 104. The tandem electrocatalysis involves one Cu-based catalyst 106 and one CO producer 108 for enhanced multi-carbon productions. Here, NiNC of 200 nm in diameter is applied as a CO producer 108 which effectively converts CO.sub.2 to CO, and then CO further reacts with Cu catalysts 106 for enhanced C.sub.2+ product formation in lower overpotentials. Also, the cuprous oxide (Cu.sub.xO) nanocube, with a 30 nm diameter is prepared as the catalyst which is expected to be reduced to a metallic copper (Cu) nanocube as the active oxidation state under the applied cathodic potentials during the CO.sub.2 reduction reaction. The active catalyst layer 102 is composed of two different materials, Cu nanocube and NiNC (Cu/NiNC), mixed and then loaded on the gas diffusion layer (GDL) 100 as the cathode in the CO.sub.2 vapor-fed flow cell setup. In the tandem reaction scheme, the increased local CO availability leads to increased CC coupling that is used for ethylene production. Specifically, NiNCs act as the CO generator to increase local CO concentration, while Cu acts as the CO promoter to further convert CO to C.sub.2+. NiNC has superior catalytic performance for converting CO.sub.2 to CO in the neutral electrolyte (>90% FE.sub.co) due to its well-dispersed single-atom Ni sites on the nitrogen-doped carbon scaffolds. In general, high-surface area Cu catalysts show high overall activity for producing ethylene, ethanol, n-propanol, n-butanol, and traces of acetate and ethane, at fairly low overpotentials.
[0070] A uniform catalyst layer of Cu/NiNC GDE is composed of as-prepared Cu.sub.xO nanocubes and NiNCs produced by thorough mixing during GDE fabrication as shown from top view in the SEM image in FIG. 1B. Similarly, FIG. 1C is the cross-section view SEM image of the tandem GDE showing uniformly distributed Cu.sub.xO nanocubes and NiNCs throughout the thickness of the catalyst layer. A small particle size of cuprous oxide nanocubes is a key to their incorporation into the interparticle spacings of the much larger NiNC, accomplished by a simple sonication during electrode fabrication, resulting in a homogeneously mixed catalyst layer that can facilitate CO to readily transport over to neighboring Cu.
[0071] FIG. 1D shows EDS elemental maps of Cu and Ni of the cross-section image in FIG. 1C. FIG. 1E shows Cu 2p, and FIG. 1F shows Cu LMM XPS spectra of Cu and Cu/NiNC GDEs, and FIG. 1G shows XRD patterns of pristine Cu and Cu/NiNC GDEs before electrochemical tests.
[0072] The energy-dispersive spectroscopy (EDS) elemental maps of Cu and Ni shown in FIG. 1D also corroborate uniform distributions of cuprous oxide nanocubes and NiNC throughout the thickness. Ex situ measurements were conducted to measure the oxidation state of Cu present in the tandem GDE. The Cu 2p XPS shows peaks that correspond to Cu.sup.0, Cu.sup.1+, and Cu.sup.2+ oxidation states with a weak satellite at 941-945 eV, which is a strong indication of the presence of Cu.sub.2O, the target phase during the synthesis of cuprous oxide nanocubes as shown in FIG. 1E. This is clearer in the XPS of the Cu LLM region where a strong signal corresponding to Cu.sub.2O is observed while CuO and Cu peaks are not observed as shown in FIG. 1F. Interestingly, the Ni XPS signal could not be clearly characterized, most likely due to its low concentration in the catalyst layer, consistent with the results for similar single-site catalysts reported in the literatures. In addition, the homogeneous mixing of NiNC with cuprous oxide nanocubes could have likely hindered the detection of Ni 2p XPS signal owing to some NiNC surfaces being physically blocked by Cu. Furthermore, FIG. 1G shows the ex situ XRD patterns of the evolution of crystalline structures of Cu and Cu/NiNC tandem electrodes before (FIG. 1G) and after electrochemical reactions (FIG. 6). In both Cu and Cu/NiNC electrodes, the (110), (111), and (200) plane peaks reveal the signatures of Cu.sub.2O nanocubes and abroad carbon signal around 25 attributed to the underlying GDL substrate. According to the prior reports, cuprous oxides or copper oxides have been demonstrated to reduce to Cu metal under reducing conditions in the presence of CO.sub.2 or CO. Specifically, no surface oxides are detected on the electrochemically reduced cuprous oxide surfaces at potentials negative of 0.4 V vs. RHE as revealed by X-ray absorption spectroscopy (XAS). Thus, in this work, the as-synthesized cuprous oxide nanocubes are assumed to be fully reduced to metallic Cu nanocube under the reduction conditions where tandem effects are observed.
[0073] The electrocatalytic performance of Cu/NiNC and Cu GDEs are evaluated in a vapor-fed CO.sub.2 flow electrolyzer using 1 M KOH as the electrolyte (FIG. 7), revealing potential-dependent selectivity of products produced such as CO, C.sub.2H.sub.4, and C.sub.2H.sub.5OH (FIG. 8A, 8B, 8C). Specifically, higher ethylene FEs are observed with the Cu/NiNC tandem electrode across all tested potentials compared to the Cu only electrode as shown in the Faradaic efficiency graph of FIG. 2A. Specifically, Cu/NiNC shows FEs for C.sub.2H.sub.4 increase concomitantly with the overpotential, reaching a peak of an average 45% at 0.6 V vs. RHE. In comparison, the Cu electrode reaches an average peak FE of 35% for C.sub.2H.sub.4 at a more negative potential of 0.8V vs. RHE. The high selectivity towards C.sub.2H.sub.4 in the potential range of 0.5 to 0.7 V vs. RHE achieved with the tandem electrode is attributed to the increased CO generated by NiNC, which is confirmed to be highly CO producing in the potential range of 0.2 to 0.6 V vs. RHE (FIG. 9). The increased local CO availability near Cu facilitates CC coupling due to the proximity achieved between the two catalyst components from the robust mixing performed during GDE fabrication. The cross-section SEM images show a similar catalyst layer thickness of approximately 1 m for both Cu and Cu/NiNC electrodes at the nominal catalyst loading of 1 mg cm.sup.2, corroborating that the enhanced C.sub.2H.sub.4 selectivity is likely due to the tandem effect rather than improved mass transport (FIGS. 10A, 10B). In particular, at low overpotentials, a relatively high range of C.sub.2H.sub.4 FE of 20 to 30% is observed with the tandem electrode at the potentials ranging from 0.3 to 0.4 V vs. RHE compared to the Cu electrode (15 to 20%) and other Cu based GDEs reported in the literature. This is indicative of the tandem catalysis effect occurring particularly at the low overpotentials where NiNC produces CO with >80% FE, allowing for an early onset of CC coupling.
[0074] In addition to the higher FEs, the tandem electrode is also found to exhibit higher partial current density (PCD.sub.geo, current normalized by the geometric area of the electrode) towards the formation of C.sub.2H.sub.4 at all tested potentials as shown in FIG. 2B. In particular, the tandem electrode shows the highest C.sub.2H.sub.4 PCD.sub.geo of 62 mA cm.sup.2 at 0.6 V vs. RHE, which is more than 60% higher than the C.sub.2H.sub.4 PCD.sub.geo of Cu (38 mA cm.sup.2) at the same potential. The C.sub.2H.sub.4 PCDs for both electrodes' plateau at more negative potentials likely due to the onset of mass transport limitations of CO.sub.2 in the catalyst layer. The effect of CO generation from NiNC on tandem catalysis is demonstrated by performing CO reduction on the Cu electrode, which resulted in similar C.sub.2H.sub.4 FEs and PCDs shown above for the tandem electrode under CO.sub.2 reduction (FIG. 11). To emphasize the utilization of the generated CO and the effectiveness of CC coupling, we use the C.sub.2H.sub.4/CO selectivity ratio as an indicator to show the electrode's ability to convert local CO into C.sub.2H.sub.4.
[0075] FIG. 2C is a graph of the C.sub.2H.sub.4/CO selectivity ratio obtained with Cu/NiNC and Cu electrodes at various potentials tested in 1 M KOH electrolyte. As shown in FIG. 2C, both the Cu/NiNC and Cu electrodes generally show an increasing trend in the C.sub.2H.sub.4/CO ratio with increasing overpotential. However, it is observed to be substantially accelerated for the tandem electrode particularly at the moderate overpotentials, reaching a maximum C.sub.2H.sub.4/CO of 5.5 at 0.7 V. At the same potential, the Cu electrode shows a maximum C.sub.2H.sub.4/CO ratio of only 3.5. This dramatic increase in the selectivity ratio is attributed to the CC coupling that is facilitated by Cu using the additional local CO that is generated by NiNC, highlighting the high utilization of CO which leads to relatively higher C.sub.2H.sub.4 selectivity.
[0076] In addition to the FEs and PCDs discussed above, the tandem effect is demonstrated in terms of Cu-mass-normalized CO.sub.2 consumption rate towards C.sub.2H.sub.4 and uncoupled CO as shown in FIG. 3A and FIG. 3B, respectively. FIG. 3A shows the CO.sub.2 consumption rate towards the formation of C.sub.2H.sub.4 (left y-axis), and partial current density towards CO (right y-axis) produced on Cu and Cu/NiNC tandem electrodes at various tested potentials. FIG. 3B shows a similar graph of the CO.sub.2 consumption rate towards the formation of uncoupled CO (left y-axis), and partial current density towards CO (right y-axis). The tandem electrode shows increased CO.sub.2 consumption rate towards C.sub.2H.sub.4 formation at all potentials tested, reaching a maximum CO.sub.2 consumption rate of 0.4 mol mg.sub.Cu.sup.1 s.sup.1 at 0.8 V vs. RHE, compared to the Cu electrode reaching only 0.15 mol mg.sub.Cu.sup.1s.sup.1 at the same potential. The increase in the CO.sub.2 consumption rate towards producing C.sub.2H.sub.4 with increasing overpotential is also observed to be much faster with the tandem electrode than the Cu electrode. Interestingly, the CO.sub.2 consumption towards the formation of uncoupled CO measured with the tandem electrode at all tested potentials are much lower than those measured with the Cu electrode, likely due to the majority of CO near the Cu surface that have further CC coupling to form ethylene. This could be explained by an additional CC coupling mechanism that may be unique to our tandem electrode system operating in vapor-fed environments, which could be occurring in parallel to the conventional CC coupling mechanism, which is discussed in the next paragraph. In terms of CO activity as shown in FIG. 3B, the Cu/NiNC electrode leads to much lower CO PCD compared to Cu at all tested potentials which is likely due to the increased consumption of CO through CC coupling. This again bolsters the earlier discussion that a high local CO availability is needed in the vicinity of Cu, which NiNC is able to provide, in order for tandem catalysis to occur efficiently and accelerate the formation of C.sub.2H.sub.4.
[0077] In attempts to maximize the local CO availability, tandem GDEs with varying ratios of Cu and NiNC are fabricated and evaluated (FIG. 12A, 12B, 12C, 13). Interestingly, tandem GDEs with 4 increased NiNC loading (Cu/NiNC-low) with a fixed Cu loading results in lower C.sub.2H.sub.4FE across the entire tested potential range compared to the final tandem GDE. This suggests that there exists an optimal CO.sub.2:CO ratio in the gas stream that is needed for achieving the maximized tandem effect. For instance, an earlier study reported that CO.sub.2/CO co-feeds with a moderate CO partial pressure (i.e., CO.sub.2:CO=1:1) result in the enhancements on the C.sub.2H.sub.4 production rate compared to pure CO.sub.2 and CO feeds. Similarly, in terms of C.sub.2H.sub.4 PCD, tandem GDEs with increased NiNC loadings (Cu/NiNC-high) show decreased C.sub.2H.sub.4 PCDs due to the imbalance of CO and CO.sub.2, which suggests that the CO partial pressure is important for facilitating the tandem reaction mechanisms.
[0078] FIG. 3C is a schematic diagram illustrating two possible CO.sub.2 reduction pathways for C.sub.2H.sub.4 formation on the tandem catalyst surface 306, which is fabricated on a gas diffusion layer 304. In a first pathway 300, Cu facilitates all reaction steps. In a second pathway 302, CO produced on NiNC a few bond lengths away is transported to the Cu surface for CC coupling.
[0079] CC bond formation on pure Cu catalysts has been a major area of investigation both experimentally and computationally. Pathway 300 through COCO coupling can occur on all of our studied catalysts, including pure Cu and tandem Cu/NiNC. It is well known that establishing a higher local concentration of CO at the interface can expedite the CC coupling process. It is here where our vapor-fed, tandem catalyst system can result in greater efficacy. Specifically, the NiNC catalyst generates CO nearly exclusively near the Cu surface, which makes use of the greater CO concentration to produce C.sub.2H.sub.4, as depicted in pathway 302. Mechanistically, Langmuir-Hinshelwood (LH) and Eley-Rideal (ER) pathways have been proposed. Generally, it is believed that Cu favors surface adsorbed CO to facilitate COCO coupling, either forming OCCO* or OCCO*, through the LH reaction pathway. However, there is likely a competitive adsorption of CO.sub.2 and CO on Cu surfaces during reaction, which can potentially impede coupling between two neighboring CO molecules. On the other hand, the ER reaction pathway allows for CC coupling between an adsorbed CO and a solvated CO at the outer Helmholtz layer, avoiding competition for adsorption sites for CO intermediates. While our results cannot elucidate which of these two is dominant, it is clear that the local concentrations of CO at the interface can improve the kinetics of CC coupling. Additionally, the vapor-fed reaction environment from using GDEs in this work circumvents the need to co-feed CO as the solvated CO is sufficiently generated from CO.sub.2R on NiNC. A possible bifunctional mechanism involving both Cu and Ni catalyst sites cannot be completely disregarded either, however, further investigation through in-situ electrochemical probing would help identify reaction intermediates and mechanisms at play. Below, we present computational modeling that reveals increased transport of CO to the Cu surface in the tandem system compared to the Cu only system.
[0080] To further elucidate the underlying mechanism of the tandem catalysis process, we perform mass transfer simulations using STAR-CCM+ with varying ratios of Cu and NiNC to examine the CO.sub.2RR product distributions in the microenvironment near the catalyst surfaces. Here, we study the flux of two species, 1) CO and 2) higher-order products, denoted as P, on a model system designed to elucidate potential mass transfer mechanisms involved the tandem system.
[0081] FIG. 4A is a 3D schematic of the simulation domain. FIG. 4D is a top-down schematic view of the gas-liquid interface in the simulation domain. Briefly, a single NiNC particle is represented as a hemisphere sitting on the gas-liquid interface, and the Cu particles are represented as smaller hemispheres on the NiNC surface.
[0082] FIG. 4B and FIG. 4C are plots of concentrations of CO and P, respectively, when the flux applied to catalyst particles is J=110.sup.6 kmol/m.sup.2s. FIG. 4E and FIG. 4F are plots of dimensionless diffusive fluxes of CO and P, respectively. These figures show the particular case when the angle () of the simulation domain is 90 degrees, i.e., four Cu.sub.2O particles are considered. FIG. 4G is a graph of the calculated area-averaged fluxes of CO from NiNC and higher-order products P through the gas-liquid interface. FIG. 4H is a graph of the ratios of those fluxes as a function of the Cu.sub.2O/NiNC volume ratio. Open circular symbols with solid lines show simulation results, and dashed lines show the results when the tandem effect is inactive.
[0083] By simulating only a periodic sector of azimuth angle with a single Cu particle, the number of Cu particles is then varied by changing from 0 to 2in other words, there are 2/ Cu particles. In this model, the NiNC is assumed to produce only CO with a flux J, and the Cu is assumed to be perfectly adsorbing for CO. Additionally, we assume that the flux of P from Cu has two contributions: (1) P is produced with a flux equal to the rate of consumption of CO, i.e. the tandem effect (Cu reacts with CO generated from NiNC to form P) and (2) P is produced with a flux J, since Cu also produces CO, and we assume that this is immediately converted to P. FIG. 4G shows averaged fluxes of CO and P through the gas liquid interface as a function of the Cu/NiNC volume ratio, nondimensionalized by J. The solid and dashed lines in FIG. 4G show fluxes when the tandem effecteffect (1) aboveis turned on and off, respectively. When the tandem effect is off, the flux through the gas liquid interface is simply based on the flux of CO from NiNC and the flux of P from Cu via mass conservation; the fluxes of CO and P decrease and increase, respectively, due to the geometric constraint that the available active surface area of NiNC decreases as the number of Cu particles increases. When the tandem effect is active, and the Cu particle is converting the CO produced by the NiNC into product P, we see an enhancement of the flux of P through the gas-liquid interface and a corresponding decrease in the flux of CO. This implies that the tandem effect causes a portion of the CO produced to be converted to P through CC coupling and illustrates how the catalyst composition can be tuned to achieve the desired product selectivity.
[0084] In FIG. 4H, this same data is plotted as a ratio, highlighting the extent that the tandem effect enhances conversion as the volume ratio increases. These simulation results corresponded to our experimental results in which higher Cu/NiNC ratios show the highest C.sub.2H.sub.4FE as well as C.sub.2H.sub.4 PCD. To further probe the microenvironment near the catalyst, in FIG. 4B and FIG. 4C, the concentration fields of CO and P are plotted for an example simulation. The CO concentration is fully depleted on the Cu surface, leading to an accumulation of P concentration, due to the conversion of CO to C.sub.2+ products via CC coupling on Cu; in contrast, the CO concentration is the highest and the P concentration is diminished on the NiNC particle, where CO is produced. In FIG. 4E and FIG. 4F, the local fluxes of CO and P on the gas-liquid interface are shown. Both fluxes are maximized on the gas-liquid interface right next to the catalyst particle, in other words, closest to where the species are produced; mass transfer into the gas stream is thus maximized here.
[0085] To further demonstrate the practical utilization of the tandem catalysis, the performance of the Cu/NiNC GDE is evaluated as the cathode in a MEA electrolyzer as shown in FIG. 5A which is a schematic illustration of the employed MEA cell setup. MEA performance evaluation of Cu and tandem Cu/NiNC gas diffusion electrodes for CO.sub.2 reduction is illustrated in FIG. 5B which is a graph of faradaic efficiency, and FIG. 5C which is a graph of chemical energy efficiency tested with 1.0 M KHCO.sub.3 anolyte and 50 sccm CO.sub.2.
[0086] Unlike in the vapor-fed flow electrolyzer demonstrated in the above section, the MEA uses a solid polymer-based membrane as the electrolyte, which is compressed between the cathode and the anode, which helps to improve energy efficiency with significantly lower single-cell voltages for potential commercial CO.sub.2 electrolyzer applications. At the applied current densities of 100 to 150 mA cm.sup.2, the Cu/NiNC tandem electrode in the MEA shows 3.0 and 3.2 V (non-iR compensated), respectively, with negligible voltage degradation over a total of one hour testing (FIG. 15). With increasing current density, C.sub.2H.sub.4 FE is observed to increase from 30 to 40%, while the CO FE decreases from 20 to 12%. The increasing C.sub.2H.sub.4 and decreasing CO selectivity trends with current density are attributed to the tandem effect in which the local CO near the Cu surface under cathodic potentials readily undergoes CC coupling. In comparison to the tandem electrode, the non-tandem Cu electrode tested in the identical MEA conditions show relatively lower C.sub.2H.sub.4 FEs of 18 and 24% at and 100 and 150 mA cm.sup.2, respectively. Interesting, the CO FE of 43% obtained at 100 mA cm.sup.2 with the Cu electrode was relatively higher than that of the tandem electrode, while 7% obtained at 150 mA cm.sup.2 was relatively lower. This decrease in the CO selectivity with increased current density observed with the Cu electrode is likely due the specific microenvironment formed at the active interface of an MEA that is different from that of a flow electrolyte cell. The performance metrics demonstrated by the tandem MEA are competitive compared to recently published results. In particular, the single-cell voltage of 3.2 V obtained at 150 mA cm-z in this study is significantly lower compared to the reported voltages which range from 3.6 to 3.9 V for C.sub.2H.sub.4 producing MEA-based CO.sub.2 electrolyzers. In terms of energy efficiency (EE), which is one of the key performance metrics related to the total energy that is required to drive the electrolyzer, the tandem MEA resulted in the C.sub.2H.sub.4EE of 10.7% and 14.5% at the applied current densities of 100 and 150 mA cm.sup.2, respectively. These results are among the highest based on EE values reported in the literature for the direct CO.sub.2-to-ethylene conversion, while the Cu MEA has resulted in C.sub.2H.sub.4 EE of 13.4%, highlighting the practical feasibility of MEA-based tandem CO.sub.2 electrolyzers for potential large-scale applications.
[0087] In summary, we have successfully demonstrated a highly selective ethylene formation at low overpotentials using a non-alloy tandem catalyst made from cuprous oxide nanocubes combined with nickel-containing nitrogen-doped carbon (NiNC) in a vapor-fed CO.sub.2 electrolysis system. Based on the product analysis, a significantly enhanced Faradaic efficiency of ethylene is achieved, resulting in 55% FE at 0.6 V vs. RHE compared to 30% obtained with the electrode made from only cuprous oxide. Additionally, a 4.5 times higher C.sub.2H.sub.4/CO selectivity ratio was observed, demonstrating a higher utilization of CO. Through varying the Cu and NiNC ratio, ethylene selectivity is maximized, which is attributed to the increased local CO coverage near the Cu surface by CO that is readily produced by NiNC. As a result, CO utilization was improved substantially for CC coupling, as indicated by low CO partial current densities in the potential range where high ethylene partial current densities are observed. To corroborate experimental observations, 3-dimensional continuum simulations were performed, verifying that CO generated from NiNC increases the CO flux to the Cu surfaces, thereby increasing CO concentration to for greater production of multi-carbon products. Furthermore, we demonstrated the practical feasibility of the tandem electrode performing CO.sub.2 conversion in a membrane electrode assembly (MEA), resulting in ethylene FE of 40% at 3.2 V cell voltage compared to 36% obtained with the Cu only electrode. Both the experimental and computational results provided in this study highlight the advantages of a tandem catalysis scheme based on Cu combined with a non-metallic CO promoter, NiNC, for the formation of ethylene. Tandem approaches can help accelerate the implementation of large-scale CO.sub.2 electrolyzers in the energy grids.
Synthesis of Nickel-Coordinated Nitrogen-Doped Carbon (NiNC) Powder
[0088] The synthesis of polyacrylonitrile (PACN)-based catalysts may use procedures in published literature. In one example, the synthesis of polymer particle, the polymerization reaction of 2 mL acrylonitrile (2 mL) is initialized by 2 mg azobisisobutyronitrile and Ni(NO.sub.3).sub.2.Math.6H.sub.2O in 2 mL acetone at 70 C. under N.sub.2 atmosphere for 12 h. Then the solution was vacuum dried and the powdered Ni-PACN was collected. The generated powder was heated in air at 230 C. for 2 h at a heating rate of 0.1 C./min. After that, the powder is further heated in N.sub.2 at 900 C. at a heating rate of 5 C./min. The final product is black Ni-PACN catalyst.
Synthesis of Cuprous Oxide Nanocubes
[0089] In another example, the synthesis process of 30 nm Cu.sub.2O nanocubes is performed using slight modifications to a published procedure. Typically, 1.152 g SDS (sodium dodecyl sulfate) powder is added to 181.2 mL water in a vial. After the solution is well mixed, 2 mL 1 M Cu.sub.2SO.sub.4 is added into the continuously stirring SDS solution. 0.8 mL 1 M NaOH is added to solution and consequently, Cu(OH).sub.2 is generated. A 16 mL 0.2 M sodium ascorbate solution is quickly injected into the Cu(OH).sub.2 solution. We allowed the solution to stir for 5 min, followed by aging for 10 minutes (no stirring) as color turns from light yellow to bright yellow. The solution is then centrifuged at 12000 RPM for 10 min and the resulting orange yellow powder is collected. The precipitate is washed twice with 20 mL 1:1 volume ratio of ethanol and water.
Preparation of Tandem Catalyst Layer Loaded Gas Diffusion Electrode (GDE)
[0090] To prepare tandem catalyst layer with different weight ratio (1:0.5, 1:1 and 1:2), 1.87 mg Cu.sub.2O powder and different amount of Ni-PACN (0.935 mg, 1.87 mg, and 3.74 mg) with 4 ul Nafion (5 wt %) solution were dispersed in 1:1 H.sub.2O:EtOH solution. The solution is sonicated for 1 h. The solution was spray-coated onto a gas diffusion layer (1.32.5 cm.sup.2, Sigracet 39BC) which is heated at 70 C. Finally, the catalyst layer is annealed in the oven to evaporate remaining solvent.
Electrochemical Measurements of GDE in Flow Electrolyte Cell
[0091] Catalyst coated electrodes were tested in a custom-built 3-compartment cell, in which a third chamber was added behind the typical catholyte chamber of the two-compartment electrochemical cell reported previously as shown in FIG. 13A, 13B, 13C, 13D. The working and counter electrode areas are 1 cm.sup.2, respectively. CO.sub.2 was continuously flowed through this third chamber to provide a vapor-phase feed to the prepared gas diffusion electrodes. Both the working and counter electrode electrolyte compartments were filled with 1M potassium hydroxide (KOH) as the catholyte for experiments at pH 14 and the vapor-fed 3.sup.rd compartment were fed with CO.sub.2 (20 sccm). IrO.sub.x GDE was used as a counter electrode, a leakless Ag/AgCl electrode (ET072-1, eDAQ) as the reference electrode. An anion exchange membrane (AEM) was a Fumasep FAA-3-50 placed between the two electrolyte-containing chambers for alkaline environments and a Selemion AMV was placed for neutral environments. Catalysts were tested using chronoamperometric (CA) measurements at defined electrochemical potentials which were corrected for 85% of the uncompensated resistance between the reference electrode and working electrode using built-in functions of the potentiostat (Biologic, VMP-300). Scan rates were set to 50 mVs.sup.1. Gas flow rates of the flow controller (gas inlet) and flow meter (gas outlet) were recorded by the external device inputs of the potentiostat.
Electrochemical Measurements of GDE in Membrane Electrode Assembly (MEA) Cell
[0092] To prepare the cathode (1 cm.sup.2), the same GDE as in the flow cell was used to spray an overlayer of the Sustainion ionomer (Dioxide Materials) with a loading of 0.05 mg cm.sup.2 to improve contact with the solid electrolyte membrane. IrO.sub.x GDE (Dioxide Materials) and Sustainion membrane (Dioxide Materials) activated in 1 M CsOH were used as the anode (1 cm.sup.2) and membrane (4.5 cm.sup.2), respectively. A commercial MEA cell assembly (Fuel Cell Technologies) was used to compress the MEA at 30 N m. The MEA electrolyzer testing was performed at room temperature with 30 sccm CO.sub.2 gas (99.999% Airgas) fed through a room temperature humidifier. The chronopotentiometry measurements were performed by holding the applied current for at least 6 minutes before stepping to the next current. Gas products from the MEA cell were quantified by injecting the outlet stream at the 5-minute mark of each applied current.
Gas Supply and Detection
[0093] Carbon dioxide was provided to the electrochemical cell and its flow rate was controlled with a flow controller. The dry carbon dioxide stream could be supplied as dry gas. Effluent gas from the vapor-fed working electrode compartment was fed to a gas chromatograph (SRI Instruments), and two gas chromatograph injections were taken per electrode potential for product quantification to determine Faradaic efficiencies. After electrochemical measurements, the liquid electrolyte was collected and tested by 1H nuclear magnetic resonance spectroscopy (NMR; Varian Inova 600 MHz) to quantify selectivity towards liquid-phase products. To calculate faradaic efficiencies and current densities, measurements were taken 3 times to ensure repeatability for catalyst synthesis and CO.sub.2 reduction testing of all tandem GDEs at each electrode potential.
Calculation of Energy Efficiency
[0094] The energy efficiency (EE) was calculated using the following equation:
[00001]
where E.sup.0.sub.product is the thermodynamic equilibrium potential of a product, FE.sub.product is the faradaic efficiency of a product, and E.sub.cell is the measured cell voltage between the cathode and the anode.
Computational Details
[0095] Simulations of mass transfer are performed using STAR-CCM+ (Siemens). The simulation domain consists of a periodic sector of a cylinder of radius 400 nm and height 10 m with NiNC and Cu.sub.2O particles (100 and 30 nm radius, respectively) on the bottom. As described previously, the angle of the sector is varied to vary the Cu.sub.2O:NiNC ratio. The concentrations of species i{CO, P} are governed by the steady diffusion equation
.sup.2C.sub.i=0.
On the NiNC particle surface, we have the following boundary conditions,
[00002]
where J is assumed to be 110.sup.6 kmol m.sup.2 s.sup.1 and the diffusivities D of both CO and P are assumed to be both equal to 110.sup.9 m.sup.2 s.sup.1. However, it is important to note that the specific values chosen do not ultimately affect the dimensionless trends. On the Cu.sub.2O particle surface, we have the following boundary conditions,
[00003]
[0096] On the gas-liquid interface of the cylinder, the species are assumed to be perfectly adsorbing into the gas stream, i.e.,
C.sub.i=0.
[0097] On the periphery of the cylinder, no flux is assumed, i.e., C.sub.i.Math.n=0, and at the top of the cylinder, far away from the catalyst particles and gas liquid interface, there is assumed to be no species C.sub.i=0. Care is taken to select a simulation domain that is large enoughboth in radius and in heightsuch that nearly all of the produced species leaves through the gas-liquid interface. The averaged dimensionless flux through the gas-liquid interface is calculated as
[00004]
where A is the surface area of the gas-liquid interface.
Physical Characterization
[0098] Catalyst GDEs were characterized ex situ by scanning electron microscopy (SEM; FEI Magellan 400 XHR, 5 kV), X-ray Photoelectron Spectroscopy (XPS; PHI Versaprobe 3 with Multipak data processing), and Grazing incidence (GI)-XRD diffraction (X'Pert Pro PANalytical Materials Research diffractometer). The SEM images and EDS results of the Cu.sub.2O, Cu.sub.2O/NiNC and NiNC gas diffusion electrodes were obtained using the EDAX/Ametek TEAM EDS system with an Octane Plus detector on a ThermoFisher Helios 6001 FIB/SEM operated at 15 KV. When taking SEM images, we used a 5 kV electron beam with 21 pA. The X-ray Photoelectron Spectroscopy (XPS) spectra were collected by a PHI Versaprobe 3 Scanning XPS Microscope with Multipak data processing. Grazing incidence (GI)-XRD (=1) characterization of gas diffusion layer (GDL), Cu.sub.2O (pre- and post-experimental) GDE, and Cu.sub.2O/NiNC (pre- and post-experimental) GDE was performed using an X'Pert Pro PANalytical Materials Research diffractometer with a Cu K (=0.154 nm) x-ray source. All samples were fixed on a standard glass slide substrate holder for XRD measurements. Diffractograms were collected at a step size of 0.03 or 0.05 and a time per step of 1 s. The beam mask and divergence slit were varied to maximize the x-ray spot size while maintaining =1.
[0099] FIG. 6 is a graph of XRD patterns of Cu and Cu/NiNC GDEs before and after electrochemical tests. FIG. 7 is a schematic illustration of a flow electrolyzer setup for vapor-fed CO.sub.2 reduction
[0100] FIG. 8A, 8B, 8C are bar charts illustrating the distribution of gas phase products obtained with Cu.sub.xO gas diffusion electrodes with various catalyst loadings using 1 M KOH electrolyte for vapor-fed CO.sub.2 reduction. (Yellow, purple, and orange bars indicate H.sub.2, C.sub.2H.sub.4, and CO, respectively.)
[0101] FIG. 9 is a bar chart illustrating the potential dependent selectivity evaluation of CO.sub.2 reduction of NiNC GDE in the vapor-fed CO.sub.2 flow electrolyzer.
[0102] FIG. 10A, 10B are cross-section SEM images of Cu GDE and Cu/NiNC GDE, respectively, used for electrochemical testing.
[0103] FIG. 11 is a bar chart illustrating Faradaic efficiency and partial current density of C.sub.2H.sub.4 obtained with the Cu GDE at various potentials tested in 1M KOH electrolyte for CO reduction reaction.
[0104] FIG. 12A, 12B, 12C are bar charts showing product distributions obtained from Cu, Cu/NiNC-high, and Cu/NiNC-low, respectively, for electrocatalytic CO.sub.2 reduction in 1 M KOH.
[0105] FIG. 16 is a table that shows catalytic performance of reported tandem catalysts in the literature.
[0106] FIG. 13A is a graph of Faradaic efficiency towards C.sub.2H.sub.4. FIG. 13B, 13C are graphs of partial current densities towards C.sub.2H.sub.4 and CO, respectively, obtained with different tandem GDEs at various potentials. FIG. 13D is a graph of C.sub.2H.sub.4/CO selectivity ratio obtained with different tandem GDEs at various potentials.
[0107] Top-view and cross-section SEM images are shown in FIGS. 9A-9F, where FIG. 14A, 14B show pre-experimental Cu GDE, where FIG. 14C, 14D show post-experimental Cu GDE, where FIG. 14E, 14F show pre- and post-experimental Cu/NiNC GDE under 1 M KOH electrolyte.
[0108] FIG. 15 is a plot showing cell voltage stability of the optimized Cu/NiNC performed in 1.0 M KHCO.sub.3(aq) obtained at 100 and 150 mA cm.sup.2 current densities.
