Electrochemical Reduction of Carbon Dioxide to Syngas and Formate

Subscriber access provided by UNIVERSITY OF CONNECTICUT

Article

The Electrochemical Reduction of Carbon Dioxide to Syngas and Formate at Dendritic Copper-Indium Electrocatalysts Zachary B. Hoffman, Tristan S Gray, Kasey Babe Moraveck, T. Brent Gunnoe, and Giovanni Zangari ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.7b01161 • Publication Date (Web): 07 Jul 2017 Downloaded from http://pubs.acs.org on July 7, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Catalysis is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 11

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

The Electrochemical Reduction of Carbon Dioxide to Syngas and Formate at Dendritic Copper-Indium Electrocatalysts Zachary B. Hoffman1, Tristan S. Gray2, Kasey B. Moraveck3, T. Brent Gunnoe2, and Giovanni Zangari1,* Departments of Materials Science and Engineering1, Chemistry2, and Chemical Engineering3, University of Virginia, P.O. Box 400745, 395 McCormick Road, Charlottesville, VA 22904 - 4745 ABSTRACT: The ability to maintain high efficiencies while simultaneously tuning the selectivity of the electrochemical reduction of CO2 (ERC) using low cost electrodes has proven to be one of the greatest obstacles to the widespread commercialization of this technology. In this study, we electrodeposit dendritic copper-indium alloys of various compositions and investigate their catalytic activity towards the reduction of CO2. These electrocatalysts are increasingly dendritic with higher Bi fraction and, depending on composition, consist of mixed phases of Cu, In and Cu-In intermetallic phases. ERC at these electrodes produces formate at high efficiencies (up to 62% with a 80 at% alloy, -1 V)) while also tuning the CO/H2 ratio to achieve an ideal syngas composition with a 40 at% In alloy, -1 V. The observed product distribution as a function of alloy composition and applied potential is rationalized in terms of the relative adsorption strengths of CO and COOH intermediates at Cu and In sites, and their distinct variation with applied potential induced by the differences in electronic structure. This study highlights the opportunities of using alloys to enhance control over the product distribution and suggests that suitable alloys could be promising catalysts for the inexpensive and efficient production of fuels.

KEYWORDS: CO2 reduction, electrocatalysis, Cu-In alloys, syngas, formate other processes that convert them to higher value fuels and chemicals. A carbon neutral cycle, for example, can be INTRODUCTION realized where formate is oxidized back to CO2 in a direct Extensive carbon dioxide (CO2) discharge into the formate fuel cell, or where Fischer-Tropsch products are atmosphere has contributed to unprecedented climate changes combusted back to CO2, and the cycle restarts with the ERC. 1-2 that should be urgently mitigated . This realization has led to The electrochemical reduction of CO2 has been a strong interest in the utilization of waste CO2, in particular as thoroughly investigated using various classes of catalysts12; an inexpensive feedstock for the production of fuels and metallic electrodes, in particular, require minimal commodity chemicals. As carbon capture and sequestration manufacturing efforts and provide favorable electron transfer (CCS) becomes more prevalent, the availability of kinetics. The ERC potentially forms a series of products via concentrated CO2 and the opportunity to diminish its successive proton coupled electron transfer (PCET) reactions 3-4 concentration in the atmosphere have grown proportionally . (Figure 1) with relatively positive redox potentials13-14. The prospect of economical and sustainable CO2 Early on, Hori developed a taxonomy of the products reuse has gained widespread attention in the global green formed by different catalyst materials15; in brief, Au and Ag 5-7 energy community . Among all the possible technological reduce CO2 only to CO, sp post-transition metals such as In, routes to convert CO2 in useful products, the electrochemical Pb, Sn, produce mostly formate, and late transition metals reduction of CO2 (ERC) to fuels appears to be the most such as Pt, Ni, Fe form mainly H2. Copper is unique among practical, since it can be carried out through a carbon neutral metals as it has been shown to produce a wide range of cycle, using renewable energy from solar and wind sources, products, including CO, formate, hydrocarbons and alcohols. under atmospheric conditions, and with a limited capital Unfortunately, these transformations occur only at high investment. Financial and feasibility analyses support this overvoltage, where the hydrogen evolution reaction (HER) is option, showing the possibility to run this process at a profit8. in competition with CO2 reduction, leading to a decline of the The resulting increased availability of commodity chemicals overall efficiency and selectivity16-17. The ERC may form CO including hydrocarbons, syngas (CO and H2 gas), and formate or formate selectively with efficiency approaching 100%, would further improve the industrial infrastructure9. Potassium although it requires precious metallic catalysts18-19 or other formate in particular has gained recent prominence as a highly impractical conditions and materials20-21 to do so. Economical efficient feedstock for alkaline fuel cell applications10-11. On and efficient electrocatalysts that slow down HER, tune the the other hand, syngas has been used for decades in Fischerselectivity towards value-added products, and remain stable Tropsch reactors for the production of hydrocarbons, including over time, are essential to enable widespread deployment of a gasoline and diesel. Thus, effluent products from the ERC practical process for the electrocatalysis of CO2. could be directly commercialized, or could be supplied to

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 1: An electrocatalytic loop visualizing the ERC by consecutive PCET steps to form (1) HCOO-, (2) CO, and (3) CH4. CH4 is then oxidized back to CO2 generating electricity, and the cycle repeats. Adapted from Kortlever et al.14 The product distribution as a function of applied potential at Cu electrodes has been rationalized by Norskov’s group utilizing the Computational Hydrogen Electrode (CHE) approach, whereby a given reaction step may occur when the applied potential decreases the thermodynamic barrier for that step to zero. The model is purely thermodynamic, and assumes that the energy barrier for such step is negligible. According to this theory the products obtained with increasing applied overvoltage are H2, HCOO-, CO and CH413. The search for trends among different metallic catalysts is facilitated by considering separately C-bound and O-bound adsorbed intermediates and the linear scaling of their adsorption strength on various metals, with the result that the reaction with the largest overvoltage is the *CO (“*” indicates an adsorbed species) to *CHO (or *COH). Cu is unique because it exhibits the lowest overvoltage for this step. Furthermore, it is predicted that CO* cannot be kept adsorbed at high overpotential: a second metal with stronger oxygen affinity is needed to enable both adsorption of CO and H at suitable potentials22. Hori15 was the first to identify a variety of products from CO2 reduction; this was confirmed by Jaramillo, who was able to detect, using a Cu cathode and a custom electrochemical cell, products ranging from C1 to C323, showing a product distribution vs. potential similar to that theoretically predicted by Norskov13. Building on early work, Kanan provided an original approach to catalyst synthesis by showing that SnOx exhibited good activity, while metallic Sn did not; a composite of the two exhibited instead a synergistic behavior24. Expanding on this result, Cu electrodes reduced from its oxide were prepared, showing enhanced catalytic

Page 2 of 11

activity and improved stability25. Oxide-derived Au nanoparticles were also investigated, resulting in the preferential formation of CO at low overpotential18. More recent work has focused on the effect of the exposed crystallographic facets using single crystals and deposits consisting of Cu nanocubes, showing much improved selectivity for ethylene with the latter; enhanced ethylene selectivity at low overpotential for Cu (100) and (211) facets was demonstrated26. In parallel, the use of nanostructured surfaces (nano-particles and –pores) has resulted in increased efficiency due to the enhanced presence of high activity facets with high Miller indices and higher surface energy27. Bimetallic catalysts have been utilized since 199128, and more recently efforts in this direction have multiplied. In particular, Cu-In cathodes were prepared by reducing an oxide derived Cu surface using an In solution, thus obtaining a bimetallic Cu-In surface with high surface area, which resulted in high efficiency of CO production (> 70%) at low overpotential (0.4 to -0.7 V vs. RHE), but the production rate was not specified29. Later work from this group showed improved crystallinity of the catalyst by electrochemical reduction of CuInO2 in KHCO3/CO2 30. A similar procedure in parallel with the coating of Cu nanoparticles by In oxide were used to prepare catalysts that evolved over time to form Cu-rich cores with an In(OH)3 shell 31, in agreement with the work of Bocarsly, showing that Indium forms metastable oxides under conditions where it is expected to be metallic 32. Finally, work on Au-Cu nanoparticles forming ordered monolayers demonstrated a composition dependence on the selectivity for various products, with limited efficiency at Au and Au-Cu electrodes33. As discussed above, copper promotes formation of various products, but only at high overpotential, decreasing its efficiency. In order to improve on this material, it may be useful to couple to Cu more oxyphilic materials from the sp metals block: Sn, In, Bi, Sb. This could result in more sluggish hydrogen evolution while enhancing the adsorption strength of species like CO and CHO to facilitate successive hydrogen addition. In this work alloying of In with Cu is investigated due to the inherent ability of In to suppress HER, even at high cathodic potentials34-36, while producing large amounts of formate as well as CO37-39. We report on the CO2 reduction at Cu-In electrodes with various compositions, electrodeposited under conditions conducive to the formation of dendritic surfaces. The ability of these electrocatalysts to selectively and efficiently convert CO2 to formate and syngas is examined. EXPERIMENTAL METHODS Synthesis and Preparation of Metallic Catalysts: Dendritic copper was electrodeposited on Cu (100 nm) sputtered on Si wafers, using a solution containing 10 mM CuSO4 (Alfa Aesar, 99.99%), 0.5 M H2SO4 (Sigma Aldrich, 98%) and 0.5 M NaCl (Fischer Scientific, 99.99%). Depositions were performed at -0.95 V vs MSE to a charge density of 1 C/cm2, corresponding to a nominal thickness of 650 nm40. Indium was electrodeposited on Au films (40 nm) sputtered on Si wafers, using an electrolyte made with 50 mM In2(SO4)3 (Alfa Aesar, 99.99%), 0.5 M H2SO4 (Sigma Aldrich, 98%) and 0.5 M NaCl (Fischer Scientific, 99.99%). Depositions were performed at -2.05 V vs MSE. A charge density of 9 C/cm2, corresponding to a nominal thickness of 2.75 µm, was used to obtain full coverage. Copper-indium alloys were electrodeposited on Au (100 nm) sputtered on Si,

ACS Paragon Plus Environment

Page 3 of 11

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

from an electrolyte containing 0.5 M H2SO4 (98%, Sigma Aldrich), 0.1 M In2(SO4)3 (Alfa Aesar, 99.99%), and 1 mM CuSO4 (Alfa Aesar, 99.99%)41. Films were grown at constant potential, between -1.45 V and -1.75 V vs MSE. All the above electrodeposition processes were carried out and monitored with a Biologic SP-150 potentiostat. Immediately before performing either electrodeposition or CO2 reduction, all surfaces were thoroughly rinsed with acetone and methanol, followed by a 1 minute etch with 10% H2SO4 and a final rinse. The geometrical area of the films was 3 cm2 (1.5 cm x 2 cm). Electrodeposition was performed in a three-electrode cell with vertical electrodes, using Pt-coated mesh as the counter and a MSE as the reference electrode. In order to prevent metallic contamination, the counter electrode was rinsed before use in concentrated HNO3. Milli-Q water (18.2 MΩ-cm resistivity) was used throughout. Materials Characterization The morphology of the electroplated films was analyzed using a FEI Quanta 650 Scanning Electron Microscope (SEM) using a SE-ETD (Secondary Electron – Everhart Thornley Detector) setup. An Energy Dispersive XRay Spectroscopy (EDS) instrument attached to the SEM, with AZtec (Oxford Instruments) software, was used to determine alloy composition. Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES) was used on selected samples to confirm the EDS measurements. X-Ray Diffraction (XRD) was used to identify the crystal structure and phase constitution of the various films. A Philips PANalytical X’Pert Pro MPD (Multi-Purpose Diffractometer) was used with a Cu X-ray tube source (λ Kα =1.54 Å). Continuous scans were performed from 20 to 65 degrees (2θ). PDF-4 (International Centre for Diffraction Data) software was used for identification of XRD patterns. High resolution transmission electron microscopy (HR-TEM, FEI Titan - 300 kV) was used to achieve better insight on the grain size, crystallographic facets and structure of the Cu-In electrocatalysts. Calculations of Fast Fourier transforms (FFT) complemented HR-TEM in determining planar d-spacings and in turn lattice constants and crystal structure. Electron Energy Loss Spectroscopy (EELS) was used to map compositional distribution of the catalyst materials. Electrochemical Cell Design and Implementation: A custom electrochemical cell was constructed, consisting of two cylindrical compartments on top of each other, separated by an anion exchange membrane (Tokuyama A-006). The top compartment contains a glass cone, vertex at the top, which confines the gaseous products that bubble up, forming a headspace. A 14/20 joint and septum closes off this

cone from the atmosphere. A schematic of the cell can be found in the Supporting Information (Figure S1). At the start of each experiment the inner cone was filled with electrolyte, which was progressively replaced with gaseous products during electrolysis. At the conclusion of an ERC experiment, the gaseous products were syringed out, volumetrically measured, and then analyzed using gas chromatography (GC). The volume of the evolved gaseous headspace varied with the electrocatalyst composition and applied potential. Besides the cone, the top compartment housed also the electrocatalyst and reference electrode, placed approximately 0.5 cm apart. The bottom compartment held the Pt mesh counter electrode. The electrolyte used in the ERC experiments was 0.1 M KHCO3 (Alfa Aesar, 99.9% metals basis), pH 7.4, purified using recrystallization and vacuum filtration techniques. After saturation with high purity CO2 (99.99%, Praxair) the pH was 6.8. A total of 60 mL solution was placed in the top compartment, 120 mL in the bottom one; the volume of the outer cone headspace was consistently 35 mL. Electrolysis Procedure: The ERC experiments were carried out using a Biologic SP-150 potentiostat. A three electrode setup was utilized, including an Ag/AgCl reference electrode (0.6 V vs RHE at a pH of 6.8, constant during each experiment); all potentials were converted to RHE. An additional resistance (Ru) compensation beyond the automatic compensation from the potentiostat23 was carried out by running an LSV (linear sweep voltammetry), re-compensating for the resistance, and then finally running the ERC at a set potential. LSVs (10 mV/s) were performed from 0.6 to -0.8 V vs RHE to characterize the electrochemical response while reducing any copper and indium oxide layers that may have formed on the surface during preparation. The ERC experiments were performed at room temperature and pressure, terminating at a total charge of 55 ± 5 C for potentials of -0.8, -0.9, and -1.0 V vs RHE and 85 ± 5 C at -1.1 V vs RHE. The charge totals were specified in order to produce a usable volume of gaseous products for quantification using gas chromatography. Constant charge totals were preferred over constant run times as they provide more appropriate criteria for comparisons based on product efficiency and selectivity with respect to different potentials and electrocatalyst compositions. Product Identification and Quantification: Gas chromatography (GC) and proton nuclear magnetic resonance (NMR) were employed to identify and quantify the various gaseous and liquid reaction products, respectively. 1 mL aliquots of gaseous products were manually extracted at the end of each run and injected into the sample loop of a gas chromatograph. Both helium and argon carrier gases were used in order to enable

ACS Paragon Plus Environment

ACS Catalysis

a

b

Figure 2: SEM images of (a) dendritic Cu and (b) electroplated In.

quantification of all gaseous products. The CO2 reduction products were identified by GC-TCD (Gas Chromatography – Thermal Conductivity Detector) with a Shimadzu GC-2014 instrument equipped with a Restek RT-Qbond 30 m × 8 mm fused silica PLOT column. The instrument was equipped with a 500 mL injection loop in which the sample passed through three columns in series (Hayesep T 80/100 mesh 0.5 m × 2.0 mm, Supelco 60/80 Mesh 5 Å molecular sieve 2.0 m × 2.1 mm, and Hayesep Q 80/100 mesh 1.5 m × 2.0 mm). H2, CO, CH4, and C2H4 were identified by use of high purity standards (Praxair). Sample chromatograms and the associated retention times of products can be found in the Supporting Information (Figures S2 and S3). In this manual batch-type system a fraction of the gases dissolved in solution and needed to be accounted for, in order to obtain accurate gaseous product quantification. Tabulated Henry’s law constants42 were used to calculate the amount of dissolved gas43-44. Yields of products in the aqueous phase, taken from both sides of the membrane, were determined by 1H NMR (10% D2O, Varian 600 MHz) using trimethylamine N-oxide as an integration standard. Solvent suppression of the H2O signal was achieved by pre-saturation and double pulsed -field gradient spin echo pair after the last pulse with the 180° pulse replaced with a composite pulse. Trimethylamine N-oxide was used as the standard in micromolar quantities to provide direct quantification of any liquid product. The standard was chosen due to its placement between potential product peaks on the NMR spectra (Figure S4). Potassium formate was identified by use of high purity standards. No other liquid CO2 reduction products were observed by 1H NMR. RESULTS AND DISCUSSION Structure and Morphology of Catalyst Materials: Cu-In alloys grow spontaneously with a dendritic morphology; in order to meaningfully compare the pure metals with Cu-In therefore we also targeted deposition of dendritic Cu and In. Electrodeposited Cu showed extensive formation of dendritic features upon chloride addition40 (Figure 2a); electroplated Indium in contrast exhibited only limited roughening (Figure 2b), with a roughness factor of 1.1, as determined by AFM. The composition of Cu-In alloys was changed by varying the applied potential between -1.45 V and -1.75 V vs MSE; the composition vs. applied potential is reported in Figure 3,

showing an increase in In fraction with increasing overpotential. Oscillations in the applied potential were due to the intense hydrogen evolution; the alloy composition on the other hand varied slightly over different tests; a criterion of ±5 at% from 20, 40, 60, or 80 at% In was required for use in ERC tests. Due to the large difference in the redox potentials of Cu and In, In is codeposited only at a potential where Cu is reduced under diffusion limitations with simultaneous HER, leading to the formation of dendritic features45-46. The applied potential determines the degree of growth instability, resulting in more complex dendrites with increasing overpotential. Thus, differences in the dendritic structure are shown in Figure 4, where Cu-rich alloys show scattered dendritic islands and indium-rich alloys indeed reveal more extensive dendritic architectures. 100

80

Atomic %, Indium

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 11

60

40

20

0 -1.80

-1.75

-1.70

-1.65

-1.60

-1.55

-1.50

-1.45

-1.40

Deposition Potential, V vs MSE

Figure 3: The compositional dependence of Cu-In alloys on deposition potential using the plating solution containing 1 mM Cu2+, 0.2 M In3+, 0.5 M H2SO4. Error bars represent standard deviation of experimentally observed variables. Using the secondary electron mode in the SEM, significant contrast is shown in the images in Figure 4. The dark contrast background represents a relatively smooth, continuous film while the bright features show the various dendritic constructions growing vertically out of the base film. Increasing the applied potential, the Cu-rich dendrites exhibited first {111} then {100} facets, with extensive twinning. Indium-rich dendrites in contrast show a more complex morphology with a variety of

ACS Paragon Plus Environment

Page 5 of 11

In(200) Cu11In9(711) Cu11In9(622) S In(103)

In(112)

In(101) Cu11In9(400) In(002) S In(110) Cu11In9(11-2) Cu11In9(313) Alloy(111) Cu9In4(332) S

Cu11In9(402)

Cu11In9(111)

Figure 4: SEM images of dendritic copper-indium electrocatalysts used in ERC experiments. Top images show overall morphology while bottom images at higher magnification highlight surface details and exposed facets. Indium content (at%) increases from left to right. topography of the various Cu-In electrocatalysts; the continuous layer at the base of the dendrites varies in thickness from 0.1 µm in the Cu-rich film to 1.1 µm in the In-rich film. Similarly, the thickness of the dendritic films ranges from 0.8 µm to >18 µm. These catalysts show significant durability after prolonged reductions tests from the structural (Figure S6) and compositional (Table S1) standpoint. at% In XRD patterns of Cu-In alloys are shown in (Figure 100 5). Copper rich alloys (up to 25 at% indium) show a dominant 83 peak at 2θ~ 43º, identified as the FCC (111) reflection of the 76 Cu solid solution. A small Cu11In9 (11-2) intermetallic peak is 60 also seen but its intensity is very low relative to the (111) 57 reflection. 37-38 at% In alloys exhibit more intermetallic 38 peaks, including Cu9In4 (332) (a triclinic structure), Cu11In9 37 (711) and (402) (with monoclinic structure), and a further shift 25 of the FCC (111) peak. 22 Figure 6a shows an enlargement of the region 2θ = Au Sub 40-44º, highlighting the FCC (111) alloy peak. Initially this peak shifts to lower 2θ angles with increasing In content due 20 30 40 50 60 2θ (degrees) to In incorporation above the solubility limit (thermodynamically indium is only soluble in Cu up to ~ 1 Figure 5: X-ray diffraction (XRD) patterns for electroplated Cu-In films. Solid lines show the various substrate peaks at%) 47. By assuming Vegard’s law, the actual fraction of In associated with Au or Si. Dashed lines show reflections from incorporated in the lattice is ~12.5 at%, at an overall indium the various phases. Note the logarithmic scale. content of 57 at%. Supersaturation of In within the Cu (111) FCC lattice is destabilized when the internal stresses achieve a critical level, such that nucleation of an intermetallic different facets, probably due to the formation of non-cubic compound (Cu11In9 or Cu9In4) becomes intermetallic phases. Figure S5 shows the cross sectional Intensity (Log)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

ACS Paragon Plus Environment

ACS Catalysis

a

b

at% In

100 83 76 60 57 38 37 25 22 Au Sub

Lattice Constant, Cu(111) (Angstroms)

Cu9In4(331)

Cu9In4(332)

Alloy(111)

Cu11In9(313)

3.735

Intensity (Log)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 11

3.730

3.725

3.720

3.715

3.710 40.5

41.0

41.5

42.0

42.5

43.0

43.5

44.0

20

30

2θ (degrees)

40

50

60

70

80

90

at% Indium

Figure 6: (a) A zoomed-in diffraction pattern of the Cu-In alloys. The overlaid black dashed line shows the shifting of the alloy (111) reflection with composition. (b) Lattice constant of the alloy (111) peak as a function of indium content in the electrodeposited Cu-In alloy. The blue dashed line shows the trend in lattice constant with average composition. energetically favorable41. This is illustrated in Figure 6a as overall In content increases from 38 to 60 at%, and the (111) reflection begins to revert back towards smaller and more thermodynamically stable lattice constants (low In content) of the solid solution, along with the emergence of Cu-In intermetallic reflections. This is also displayed by the dashed line in Figure 6b, where the (111) alloy phase reaches a critical lattice constant at 3.733 Å, before a sharp decrease is seen. Above 57 at% indium, higher order Cu-In intermetallics appear and the In (110) peak becomes much more defined. As total indium content continues to increase, the intensities of the (111) alloy and Cu-In intermetallic peaks decrease substantially in the wake of elemental indium reflections. Inrich alloy (83 at% indium), exhibit only peaks of elemental indium, with the exception of the Cu9In4 (332) reflection. HR-TEM images of Cu-In dendrites are shown in Figures 7 and 8. A low magnification view of a Cu-In dendrite (80 at% In) is seen in Figure 7a; uniform dispersion of Cu and In is confirmed via EELS/EDS mapping as shown in Figure S7. The dendrite tips (Figure 7a) reveal a consistent tip width of approximately ~50 nm. An atomic resolution image of a dendrite tip (Figure 7b)

a

reveals an amorphous layer ~5 nm thick on the surface, and a well defined crystalline structure in the interior. These features are confirmed by FFT analysis of the two regions, as shown in Figure 7c and 7d, respectively. The amorphous surface layer is an oxide film supposedly formed by exposure to air. Recently, it was shown that indium and indium oxide on the surface contribute to the catalytic activity and propensity to produce formate in the ERC48. In our case, any oxide layer that may be present is removed during the linear sweep voltammetry (LSV) prior to the ERC experiment, exposing the crystalline region to the solution. The FFT pattern in Figure 7(d) reveals measured d-spacings associated with the (400) and (313) reflections of Cu11In9 from XRD patterns shown above, without any evidence of the (111) alloy reflection. This should be expected due to the high In fraction. These assignments were confirmed by reconstructing the FFT images using PDF-4 crystallography software (ICDD, International Centre for Diffraction Data) as shown in Figure S8 (Supporting Information). Figure 8a shows the surface region of a 60 at% indium Cu-In dendrite. Again, an amorphous surface layer is distinguishable with a thickness of ~ 3 nm, less than in

b

c

d (313) (400) Figure 7: HR-TEM images of an 80 at% In catalyst: (a) a typical dendrite network, (b) a dendrite tip, and FFT images from both (c) the thin (~5 nm) amorphous surface, and (d) the Cu11In9 intermetallic crystalline bulk.

ACS Paragon Plus Environment

Page 7 of 11

the previous case, as expected due to the lower In fraction. The FFT pattern reveals a superposition of the Cu11In9 intermetallic and the (111) solid solution reflections. Highorder crystallographic planes are seen in the diffraction pattern, including the (311), (11-2), and (711). As before, the oxide layer is removed prior to CO2 electrolysis and is therefore neglected. The high curvature of the dendrite

size with increasing indium content. The alloy (111) reflection showed relatively constant grain size around 15 nm before rising to 25 nm from 60 to 80 at% indium. 50

Cu11In9 (11-2)

Average Grain Size (nm)

45

Alloy (111) Cu9In4 (332)

40

Cu11In9 (711) 35 30 25 20 15 10 0

20

40

60

80

100

at% Indium

Figure 8: (a) HR-TEM of a 60 at% indium dendrite tip and (b) the associated FFT image labeled with the relevant crystallographic reflections.

Figure 9: Grain size of various crystallographic reflections as a function of overall indium content in the film (at% Indium). Grain size was evaluated using the Scherrer equation.

tips is likely to expose a wide range of crystallographic facets to the electrolyte, including those with high Miller indices. It should be noted however that XRD and TEM diffraction are unsuitable to determine the facets exposed to the electrolyte, due to the fact that diffraction methods identify the orientation of crystallographic directions in the bulk, that generally do not coincide with the surface orientation. This identification has been unfortunately often mistakenly made in the literature. The grain size of various Cu-In reflections were estimated using the Scherrer equation, revealing grain size trends with composition (Figure 9). Higher order reflections such as Cu9In4 (332) and Cu11In9 (711) showed a larger grain

However, the Cu11In9 (11-2) reflection remained relatively stationary between 20 and 30 nm. Grain size determination from XRD analysis matched with direct TEM measurements of dendritic facets on various Cu-In alloys (length scale of 10s – 100s nm).

100

Faradaic Efficiency (%)

90

Electrolysis of CO2 - Product Distribution and Conversion Rates

90

80

80

70

70

60

60

50

50

40

40

30

30

20

20

10

10

90

b

Hydrogen CO Formate

0 0

100

Figure 10 shows product distribution as a function of fraction for various applied potentials.

In

100

a

0

Faradaic Efficiency (%)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

20

40

60

80

100 100

c

90

80

80

70

70

60

60

50

50

40

40

30

30

20

20

10

10

0

0

20

0

20

40

60

80

100

40

60

80

100

d

0 0

20

40

60

Atomic % Indium

80

100

Atomic % Indium

Figure 10: Product faradaic efficiency as a function of indium content at various potentials (a, -0.8 V vs RHE; b, -0.9 V vs RHE; c, -1.0 V vs RHE; d, -1.1 V vs RHE).

ACS Paragon Plus Environment

ACS Catalysis

applications, at -1.1 V using a 80 at% indium alloy. Formate peaks at 32% efficiency, as does CO at 17%, with a 60 at% In catalyst. Overall, a complex tradeoff between CO, formate, and H2 yields is realized at all potentials and alloy compositions. A synergistic effect between copper and indium is generally observed, as we note lower H2 efficiencies on the alloys, at all potentials, with respect to either elemental metal. We attempted to determine the actual surface area of our electrodes at a constant cross-sectional thickness by using electrochemical impedance spectroscopy (EIS), according to the method described in ref. [49]. However, at open circuit potential, the approach to a vertical line expected in the Nyquist plot at low frequency for porous electrodes is never achieved, and the roughness factors determined from the capacitance at low frequency are much lower than expected based on the SEM imaging of cross sections (Figure S5). We therefore had to resign to use geometric areas and nominal current densities, with the assumption that the surface areas of dendritic Cu and Cu-In would be similar due to the similar morphology in order to meaningfully compare the available data. Total and partial nominal current densities for the detected products (Figure 11) were determined using products yields, current data acquired during the electrolysis of CO2 on the various electrocatalyst materials, and the geometrical area. Pure Cu shows the largest total current density across all potentials, while for Cu-In alloys the nominal current decreases with increasing In content (Figure 11a). The trends for pure In generally mirrors that of the 60 and 80 at% In materials.

Product distribution as a function of applied potential for the various compositions is displayed for completeness in Figure S9. Pure dendritic Cu produces more than 75% H2 at all potentials, while CO and formate are produced in significantly smaller amounts. Methane and ethylene are minor products, formed at 0.1-1% efficiency (Figure S9). An inclination to produce predominantly H2 and formate is revealed also on pure indium surfaces, with CO efficiencies below 10% across all potentials. As overpotential is increased, on pure In, formate reaches its peak efficiency at -1 V, just below 40%. As with pure copper, H2 is the dominant product on indium at all explored potentials, although not as overwhelming. At the most positive potentials of -0.8 and -0.9 V (Figure 10a & 10b), CO is produced steadily at 30% efficiency across a wide range of alloy compositions, while HER is concurrently suppressed to below 35% efficiency. Formate is produced in substantial quantities, with peak efficiencies above 60 % being achieved at 60 or 80 at% In alloys. At -1 V (Figure 10c), using a 40 at% In alloy composition, the proper ratio of syngas is obtained as well as 50% formate. These are the optimal conditions to enable utilization of the entire product stream, though separation would still be needed. CO peaks at 28% efficiency, using a 20 at% In alloy, before dropping off as In content increases. Furthermore, the highest reported formate efficiency (62%), in this study, is achieved using an 80 at% indium catalyst at this potential. At -1.1 V (Figure 10d), hydrogen evolution dominates over all ERC products and therefore is typically undesirable given the higher energy consumption and lower ERC product yield. That being said, syngas can still be produced, at an appropriate ratio for Fischer-Tropsch a

b 1.0

2

Partial Current Density (mA/cm )

2

Current Density, mA/cm

Pure Cu 20 at% In 40 at% In 60 at% In 80 at% In Pure In

0.9

10

1

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

-1.1

-1.0

-0.9

-1.1

-0.8

-1.0

-0.9

-0.8

Potential (V vs RHE)

Potential (V vs RHE)

c

d

0.7

10

2

2

Partial Current Density (mA/cm )

0.6

Partial Current Density (mA/cm )

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 11

0.5

0.4

0.3

0.2

0.1

1

0.1

0.0 -1.1

-1.0

-0.9

-0.8

Potential (V vs RHE)

-1.1

-1.0

-0.9

-0.8

Potential (V vs RHE)

Figure 11: Total geometric current density (a) and partial currents for (b) formate, (c) CO, and (d) H2 for Cu, In and the various CuIn electrocatalysts as a function of applied potential. The current density is normalized to the geometric surface area for each specific material. A logarithmic scale is used in (a) and (d). Alloying of In to Cu resulted in a slower kinetics towards water electroreduction, down by almost an order of magnitude. Formate partial current densities (Figure 11b) are

shown to increase as overpotential is increased, even up to -1.1 V vs RHE for some alloys where the efficiency of formate production dropped off only due to the increased HER at this

ACS Paragon Plus Environment

Page 9 of 11

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

potential. A peak for formate partial current density is seen on several Cu-In alloys at -1.0 V vs RHE, corresponding to the highest Faradaic efficiencies for formate production. CO partial current densities for the various Cu-In alloys (Figure 11c) show the lowest conversion rates among the various products and some scatter as a function of potential, but they all perform better than pure Cu and In. Thus, a synergistic effect in Cu-In alloys was observed: all alloys produced CO at a higher rate than either pure metallic component. With regard to the H2 partial current density jH2 (Figure 11d), the trends are intuitive – jH2 is increasing with larger overpotential on all electrocatalysts under study. Interestingly, pure electroplated In generally produces H2 at a larger rate than the 40, 60, and 80 at% In alloys, showing inhibition of HER upon alloying. Comparisons with other works available in the literature can be made, but they must be only qualitative due to the widely different conditions that have been used. The nominal total current density observed in this work is similar to those of refs. [18, 25, 33], but higher than ref.29, probably due to the more positive potential range used in the latter work. The product distribution is also similar to those reported in those references, with formation of mostly CO and/or formate. The work by Rasul29 showed a higher CO efficiency (up to 85%), probably due to the slower HER at more positive potentials. High activity facets on Cu nanofoams show a propensity to generate formate at higher efficiencies than on their smooth Cu counterparts, with efficiencies as high as 35%27. Thus, alloying Cu with In at a similar dendritic surface morphology, allows for higher formate efficiencies (62%) as demonstrated in this work. Furthermore, nanostructured SnO2 particles were prepared, by a different method than in ref24, and a formate efficiency over 80% was achieved50. These comparisons lead us to hypothesize that the oxide derived materials reported in refs.18, 24, 25, 29 afford higher efficiencies simply due to nanostructuring and resulting formation of high energy crystal facets with enhanced activity, as exemplified by this work and refs. [27, 33, 50]. We attempted to find correlations between the structural properties and composition of the electrocatalyst vs. product distribution. We noted an approximate relationship between the extent of formate production and the intensity of the high order reflections (400), (711), and (311) for the Cu9In4 and Cu11In9 phases. In contrast, as In content decreased the (111) Cu-In alloy reflection became stronger, correlating to a dominance of CO in the product stream, at potentials below -1.1 V. In addition, selectivity to formate was seen to increase with grain size; this may suggest that defects such as steps and grain boundaries may not be the active sites, but it should be noted that twins may not be detected by XRD but may still be active sites for the reactions of interest. No correlation was however found between the phase constitution and the product distribution, suggesting that no definite conclusion can be inferred about microstructural correlations from the available data. On the other hand, the observed product distribution can be more directly explained on the basis of surface composition. The Cu-In alloy will tend to be enriched in Indium at the surface due to the larger affinity to –OH; this enrichment however is difficult to quantify due to the complex surface morphology. An overview of the literature shows that CO exhibits a strong adsorption strength at Cu (∆Hads ~ -0.7 eV/at), when CO adsorbs on (100) facets through the C atom51. In contrast, very limited (or no) adsorption of CO2 or

CO occurs on In surfaces52. More recently, Lim et al.53 calculated by DFT the enthalpy of adsorption ∆Hads of COOH and CO on Cu and In (see Fig. 3 in [53]): ∆Hads of COOH on Cu was calculated to be -1.5 eV/at and on CO was -0.7 eV/at; in contrast ∆Hads of COOH on In was ~ -1.0 eV, and on CO was ~ 0. These data show that the presence of two different elements at the surface, one a d-metal, another a sp-metal, may enable modulation of the relative variation of the strength of adsorption of –COOH and –CO, resulting in a departure from the correlation discussed by Norskov between ∆Hads (COOH) and ∆Hads (CO). This in turn may result in a variation of product distribution vs composition, as experimentally observed. The pathway for the formation of –COOH (detected as HCOOK in our system) and CO has been described as follows13: (1)

CO + H  + e ⟶ ∗ COOH

(2)

∗ COOH + H  + e ⟶ CO + H

In order to form HCOOK the adsorption strength of –COOH must be small so that the product may quickly desorb; at high Cu content the probability to adsorb on a Cu atom would increase and CO tends to be formed according to eq. (2). At high In content on the contrary the CO tends to decrease, as observed. The delocalized p orbital on In has a relatively low (compared with other metallic p block dopants) radicalpreparation energy cost with regards to stabilizing the 2p orbital of C in the –COOH intermediate. The cost of stabilizing the covalent σ-bond of the p orbital of In with the 2p of –COOH is also relatively low, which ultimately steers the reaction to COOH and in turn HCOOK, after desorption. Furthermore, CO is formed in smaller quantities because an empty p orbital on In is required, for which the unbound electrons on the C in CO must bond53. A larger amount of energy is thus required to catalyze the –COOH → CO step, as shown above. The discussed trends correspond with our product distribution, where generally formate is the preferred product, over CO, as In content increases. The product distribution vs. applied potential on the other hand is more difficult to predict due to the dependence of ∆Hads on potential. Overall, the alloying of In with Cu inhibits HER and hinders the path of CO2 reduction reaction to CO (Figure 1). CONCLUSION The electrochemical reduction of CO2 (ERC) was performed at Cu-In alloy electrocatalysts of varying composition within a potential range of -0.8 to -1.1 V vs RHE. The ERC at these electrocatalysts has been shown to produce formate at high efficiency while simultaneously enabling adjustment of the ratio of syngas components. Specifically, at 40 at% Indium and -1.0 V vs RHE, we achieve high formate production (49% Faradaic Efficiency, FE) as well as an optimal ratio, of 2.6:1 H2 to CO, in the effluent syngas produced. Furthermore, a maximum of 62% FE for formate is obtained at this potential with an 80 at% indium electrode. Dendritic Cu-In alloys were obtained by electrodeposition; Cu-rich electrodes mainly consisted of a solid solution, while various intermetallics were formed at higher In content. TEM imaging evidenced the formation of dendrite tips of ~ 50 nm width, exhibiting a variety of surface facets including those with high surface activity. The

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

selectivity towards formate synthesis is linked to the indium fraction at the electrode surface. On the other hand, the greatest CO efficiencies are achieved at lower reduction potentials and decrease with higher indium content. The fraction of In (sp metal) and Cu (d metal) at the surface varies the relative adsorption strength of –COOH and –CO intermediates, and changes their trend with applied potential, which ultimately led to the observed product distribution. Finally, it is shown that dendritic copper-indium synergistically promotes the electrocatalysis of CO2 to formate and syngas at greater efficiencies than either pure metallic counterpart.

ASSOCIATED CONTENT Supporting Information. Additional information including additional microscopy and imaging, electrochemical characterization, sample chromatograms and NMR data, catalyst stability and durability, and cell design. This material is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATION Corresponding Author *Giovanni Zangari Email: [email protected] Phone: 434-243-5474 Dept. of Materials Science and Engineering, University of Virginia, Charlottesville, 22904-4745, USA

ACKNOWLEDGMENT This work was financially supported by the NSF grant CHE #1152778. The authors gratefully acknowledge partial support through the Environmental Resilience and Sustainability Fellowship awarded on behalf of the Office of Graduate Research, at the University of Virginia, the Jefferson Trust, and the AES Graduate Fellowship in Energy program. We would also like to thank Richard White and Helge Heinrich for assistance in materials characterization methodology, Dr. Jeff Ellena for assistance with NMR testing, and Dr. Jonathan Fillippi for many long technical discussions.

REFERENCES [1] Hansen, J.; Sato, M.; Ruedy, R.; Lo, K.; Lea, D. W.; & MedinaElizade, M. Proc. Natl. Acad. Sci. U.S.A., 2006, 103, 14288. [2] IPCC Fourth Assessment Report (AR4) http://www.ipcc.ch/publications_and_data/publications_ipcc_fourth_a ssessment_report_wg1_report_the_physical_science_basis.htm (accessed Jun 7, 2017). [3] Herzog, H. J. Environ. Sci. Tech., 2001, 35, 148A-153A. [4] The Global Status of CCS 2014 https://hub.globalccsinstitute.com/publications/global-status-ccs2014/global-status-ccs-2014 (accessed Jun 7, 2017). [5] Centi, G.; & Perathoner, S. Catal. Today, 2009, 148, 191-205. [6] Gattrell, M.; Gupta, N.;, & Co, A. J. Electroanal. Chem., 2006, 594, 1-19. [7] Ma, J.; Sun, N.; Zhang, X.; Zhao, N.; Xiao, F.; Wei, W.; & Sun, Y. Catal. Today, 2009, 148, 221-231. [8] Agarwal, A. S.; Zhai, Y.; Hill, D.; & Sridhar, N. ChemSusChem, 2011, 4, 1301-1310. [9] Hu, B.; Guild, C.; & Suib, S. L. J. CO2 Util., 2013, 1, 18-27. [10] Jiang, J.; & Wieckowski, A. Electrochem. Commun., 2012, 18, 41-43.

Page 10 of 11

[11] Bartrom, A. M.; Ta, J; Nguyen, T. Q.; Her, J.; Donovan, A.; & Haan, J. L. J. Power Sources, 2013, 229, 234-238. [12] Whipple, D. T.; & Kenis, P. J. J. Phys. Chem. Lett., 2010, 1, 3451-3458. [13] Peterson, A. A.; Abild-Pedersen, F.; Studt, F.; Rossmeisl, J.; & Nørskov, J. K. Energy Environ. Sci., 2010, 3, 1311-1315. [14] Kortlever, R.; Shen, J.; Schouten, K. J. P.; Calle-Vallejo, F.; & Koper, M. T. J. Phys. Chem. Lett., 2015, 6, 4073-4082. [15] Vayenas, C. G. Modern Aspects of Electrochemistry; Y. Hori; Springer US: Berlin, 2007; Vol. 42, p 89 – 189. [16] Bevilacqua, M.; Filippi, J.; Miller, H. A.; & Vizza, F. Energy Technology, 2015, 3, 197-210. [17] Lim, R. J.; Xie, M.; Sk, M. A.; Lee, J. M.; Fisher, A.; Wang, X.; & Lim, K. H. Catal. Today, 2014, 233, 169-180. [18] Chen, Y.; Li, C. W.; & Kanan, M. W. J. Am. Chem. Soc., 2012, 134, 19969-19972. [19] Lu, Q.; Rosen, J.; Zhou, Y.; Hutchings, G. S.; Kimmel, Y. C.; Chen, J. G.; & Jiao, F. Nat. Comm., 2014, 5. [20] Rosen, B. A.; Salehi-Khojin, A.; Thorson, M. R.; Zhu, W.; Whipple, D. T.; Kenis, P. J.; & Masel, R. I. Science, 2011, 334, 643644. [21] Sonoyama, N.; Kirii, M., & Sakata, T. Electrochem. Commun., 1999, 1, 213-216. [22] Peterson, A. A.; & Nørskov, J. K. J. Phys. Chem. Lett., 2012, 3, 251-258. [23] Kuhl, K. P.; Cave, E. R.; Abram, D. N.; & Jaramillo, T. F. Energy Environ. Sci., 2012, 5, 7050-7059. [24] Chen, Y.; & Kanan, M. W. J. Am. Chem. Soc., 2012, 134, 19861989. [25] Li, C. W.; & Kanan, M. W. J. Am. Chem. Soc., 2012, 134, 72317234. [26] Roberts, F. S.; Kuhl, K. P.; & Nilsson, A. Angew. Chem., 2015, 127, 5268-5271. [27] Sen, S.; Liu, D.; & Palmore, G. T. R. ACS Catal., 2014, 4, 30913095. [28] Watanabe, M.; Shibata, M.; Kato, A.; Azuma, M.; & Sakata, T. J. Electrochem. Soc., 1991, 138, 3382-3389. [29] Rasul, S.; Anjum, D. H.; Jedidi, A.; Minenkov, Y.; Cavallo, L.; & Takanabe, K. Angew. Chem., 2015, 127, 2174-2178. [30] Jedidi, A.; Rasul, S.; Masih, D.; Cavallo, L.; & Takanabe, K. J. Mater. Chem A, 2015, 3, 19085-19092. [31] Larrazabal, G.O.; Martin, A.J.; Mitchell, S.; Hauert, R.; & Perez-Ramirez, J. ACS Catal. 2016, 6, 6265-6274. [32] Pander, J.E.; Baruch, M.F., Bocarsly, A.B. ACS Catal. 2016, 6, 7824-7833. [33] Kim, D.; Resasco, J.; Yu, Y.; Asiri, A. M.; & Yang, P. Nat. Comm., 2014, 5. [34] Butler, J. N.; & Makrides, A. C. Trans. Faraday Soc., 1964, 60, 1664-1676. [35] Butler, J. N.; & Dienst, M. J. Electrochem. Soc., 1965, 112, 226232. [36] Metikoš-Huković, M.; Omanović, S. J. Electroanal. Chem., 1998, 455, 181-189. [37] Hori, Y.; Kikuchi, K.; & Suzuki, S. Chem. Lett., 1985, 14, 16951698. [38] Kapusta, S.; & Hackerman, N. J. Electrochem. Soc., 1983, 130, 607-613. [39] Jitaru, M.; Lowy, D. A.; Toma, M.; Toma, B. C.; & Oniciu, L. J. Appl. Electrochem., 1997, 27, 875-889. [40] Shao, W.; & Zangari, G. J. Phys. Chem. C, 2009, 113, 1009710102.

ACS Paragon Plus Environment

Page 11 of 11

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

[41] Liang, D.; Unveroglu, B.; & Zangari, G. J. Electrochem. Soc., 2014, 161, D613-D619. [42] Sander, R. Compilation of Henry's law constants for inorganic and organic species of potential importance in environmental chemistry. http://henrys-law.org/henry-3.0.pdf (accessed Jun 7, 2017). [43] Cox, C. R.; Lee, J. Z.; Nocera, D. G.; Buonassisi, T. Proc. Natl. Acad. Sci., 2014, 201414290. [44] Morris, A. J.; McGibbon, R. T.; Bocarsly, A. B. ChemSusChem, 2011, 4, 191-196. [45] Shin, H. C.; Liu, M. Adv. Funct. Mater., 2015, 15, 582-586. [46] Shin, H. C.; Dong, J.; & Liu, M. Adv. Mater., 2003, 15, 16101614. [47] Subramanian, P. R.; Laughlin, D. E. Bull. Alloy Phase Diagrams, 1989, 10, 554-568. [48] Detweiler, Z. M.; White, J. L.; Bernasek, S. L.; Bocarsly, A. B. Langmuir, 2014, 30, 7593-7600. [49] Rouya, E.; Cattarin, S.; Reed, M. L.; Kelly, R. G.; & Zangari, G. J. Electrochem. Soc., 2012, 159, K97-K102. [50] Zhang, S.; Kang, P.; & Meyer, T. J. Am. Chem. Soc., 2014, 136, 1734-1737 [51] Föhlisch, A.; Nyberg, M.; Bennich, P.; Triguero, L.; Hasselström, J.; Karis, O.; Nilsson, A. J. Chem. Phys., 2000, 112, 1946-1958. [52] Rossnagel, S. M.; Dylla, H. F.; & Cohen, S. A. J. Vac. Sci. Technol., 1979, 16, 558-561. [53] Lim, H. K.; Shin, H.; Goddard III, W. A.; Hwang, Y. J.; Min, B. K.; & Kim, H. J. Am. Chem. Soc., 2014, 136, 11355-11361.

FOR TABLE OF CONTENTS ONLY

ACS Paragon Plus Environment