Research Article

Quantum Mechanical Screening of Single-Atom Bimetallic Alloys for the Selective Reduction of CO2 to C1 Hydrocarbons

The Joint Center for Artificial Photosynthesis, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States
Department of Chemistry, University of California at Berkeley, Berkeley, California 94720, United States
§ Department of Chemical and Biomolecular Engineering, University of California at Berkeley, Berkeley, California 94720, United States
ACS Catal., 2016, 6 (11), pp 7769–7777
DOI: 10.1021/acscatal.6b01393
Publication Date (Web): September 23, 2016
Copyright © 2016 American Chemical Society
OpenURL UNIV OF NORTH TEXAS
*E-mail: alexbell@berkeley.edu (A. T. Bell)., *E-mail: mhg@cchem.berkeley.edu (M. Head-Gordon).

Abstract

Abstract Image

Electrocatalytic reduction of CO2 to energy-rich hydrocarbons such as alkanes, alkenes, and alcohols is a very challenging task. So far, only copper has proven to be capable of such a conversion. We report density functional theory (DFT) calculations combined with the Poisson–Boltzmann implicit solvation model to show that single-atom alloys (SAAs) are promising electrocatalysts for CO2 reduction to C1 hydrocarbons in aqueous solution. The majority component of the SAAs studied is either gold or silver, in combination with isolated single atoms, M (M = Cu, Ni, Pd, Pt, Co, Rh, and Ir), replacing surface atoms. We envision that the SAA behaves as a one-pot tandem catalyst: first gold (or silver) reduces CO2 to CO, and the newly formed CO is then captured by M and is further reduced to C1 hydrocarbons such as methane or methanol. We studied 28 SAAs, and found about half of them selectively favor the CO2 reduction reaction over the competing hydrogen evolution reaction. Most of those promising SAAs contain M = Co, Rh, or Ir. The reaction mechanism of two SAAs, Rh@Au(100) and Rh@Ag(100), is explored in detail. Both of them reduce CO2 to methane but via different pathways. For Rh@Au(100), reduction occurs through the pathway *CO → *CHO → *CHOH → *CH + H2O(l) → *CH2 + H2O(l) → *CH3 + H2O(l) → * + H2O(l) + CH4(g); whereas, for Rh@Ag(100), the pathway is *CO → *CHO → *CH2O→ *OCH3 → *O + CH4(g) → *OH + CH4(g) → * + H2O(l) + CH4(g). The minimum applied voltages to drive the two electrocatalytic systems are −1.01 and −1.12 VRHE for Rh@Au(100) and Rh@Ag(100), respectively, at which the Faradaic efficiencies for CO2 reduction to CO are 60% for gold and 90% for silver. This suggests that SAA can efficiently reduce CO2 to methane with as small as 40% loss to the hydrogen evolution reaction for Rh@Au(100) and as small as 10% for Rh@Ag(100). We hope these computational results can stimulate experimental efforts to explore the use of SAA to catalyze CO2 electrochemical reduction to hydrocarbons.

1 Introduction


Global energy consumption is projected to triple by the end of the 21st century, relative to the present rate, because of population and economic growth.(1, 2) While this increased energy demand can be met through the use of fossil fuels (petroleum and coal), doing so is expected to triple the level of atmospheric CO2 by the end of the century.(1) Such a dramatic increase in atmospheric CO2 concentration in the atmosphere is likely to perturb global ecosystems on an unprecedented scale.(3) A potential means for reducing CO2 concentration and achieving carbon neutrality is the electrochemical reduction of CO2 to fuel using electricity produced from wind, solar radiation, and geothermal heat. To achieve this goal, it is necessary to discover electrocatalysts for CO2 reduction that are inexpensive (ideally composed of Earth-abundant materials), efficient (having low overpotentials), selective, and stable.(4-13) With regard to selectivity, the objective is to obtain high Faradaic efficiencies to the products of CO2 reduction while reducing the Faradaic efficiency for the hydrogen evolution reaction (HER), which can dominate, particularly at less negative applied voltages.(14-18)
In thermodynamic terms, the outcome of the electrocatalysis is determined by the stability of the first hydrogenation intermediates (*COOH or *OCHO for CO2RR, and *H for HER; see Scheme 1).(19-21) If *COOH or *OCHO is more stable than *H, it will occupy the active sites, and then the CO2RR is the primary pathway; on the other hand, if *H is more stable, then the HER will be the dominant pathway. Since the formation of these intermediates all involve the transfer of one (H+ + e) pair, variations in the applied voltage (assuming the interaction between the dipole of the adsorbate and the space-charge field is small and can be ignored)(22) or pH will not affect the relative stability of the *COOH, *OCHO, or *H intermediates. As a consequence, the key to enhancing the CO2RR and to suppressing the HER is discovery of electrocatalysts that bind *COOH or *OCHO more strongly than *H (ΔG*COOH or ΔG*OCHO < ΔG*H, where ΔG*COOH, ΔG*OCHO, and ΔG*H are the Gibbs free energies of formation for surface-bonded COOH, OCHO, and H species, respectively, referenced to a clean surface, H2(g), and CO2(g)).
figure
Scheme 1. Schematic Description of the First Two Hydrogenation Steps for the Electrochemical CO2 Reduction Reaction and Hydrogen Evolution Reactiona

aThe asterisk symbol (*) represents the surface, and “*A” represents A adsorbed on the surface.

Meeting this criterion has proven to be a very challenging task. Karamad et al. have used density functional theory (DFT) to predict the performance of 34 intermetallic alloys for CO2RR, relative to that for HER.(20) They found that ΔG*COOH (or ΔG*OCHO) > ΔG*H for all 34 materials, meaning that HER would be the dominant reaction pathway in all cases. Changing the electrocatalysis to consider the competition between the CO reduction reaction (CORR) versus HER on the same 34 catalysts showed that four of them can perform CORR preferentially to HER. In other words, four cases satisfy ΔG*CHO (or ΔG*COH) < ΔG*H, where *CHO and *COH are the first hydrogenated intermediates of CORR (see Scheme 2), and ΔG*CHO, ΔG*COH, and ΔG*H are their Gibbs free energies of formation, relative to a clean surface, and H2(g), and CO(g).(20) Based on those discoveries, the authors proposed a two-pot tandem reaction for CO2RR, in which CO2 is first reduced to CO by well-defined existing protocols such as the reverse water–gas shift reaction,(23) and then CO is further reduced to hydrocarbons by those alloys. However, the capital cost for the proposed two-pot process is expected to be higher than that for an integrated system. Therefore, there is an opportunity to explore new types of electrochemical catalysts that can first convert CO2 to CO, and then to hydrocarbons, while suppressing the HER.
figure
Scheme 2. Schematic Description of the First Hydrogenation Step for the Electrochemical CO Reduction Reaction and Hydrogen Evolution Reaction
In this study, quantum mechanical DFT calculations are used to screen 28 single-atom alloys (SAAs) as novel electrochemical catalysts for CO2RR. Those catalysts are composed of isolated single atoms, M (M = Cu, Ni, Pd, Pt, Co, Rh, and Ir), embedded into the (111) and (100) surfaces of Au or Ag (Figure 1). The total number of SAAs is 28 (7 × 2 × 2 = 28), and we will denote these catalysts as M@Au(111), M@Au(100), M@Ag(111), or M@Ag(100).
figure

Figure 1. Surface models used to simulate the single-atom alloys: (a) M@Au(111), M@Ag(111), and (b) M@Au(100), M@Ag(100). Each model is composed of a three-layer slab, using a 3 × 3 periodic cell, where one Au or Ag atom on the topmost layer is replaced by M (M = Cu, Ni, Pd, Pt, Co, Rh, and Ir). This means that the doping concentration in the top layer is 11.1% (1/9). The periodic boundary is represented by the black dotted line.

Au and Ag are known for their ability to reduce CO2 to CO with high Faradaic efficiencies,(17, 18, 24-28) in preference to the HER, and are the major component of our proposed catalysts. Therefore, we expect that CO2 will first be selectively reduced to CO by the surrounding Au or Ag (Scheme 3). While Ag and Au cannot further reduce CO, because of its much stronger binding interaction with M than with Au or Ag,(29) the newly formed CO will be preferentially bound to the single atom alloy site, M. There CO may be further reduced to hydrocarbons, if the CORR can be favored versus HER.
figure
Scheme 3. Schematic Description of Our Proposed One-Pot Tandem Catalytic Reactiona

aThe proposed electrocatalyst is composed primarily of gold or silver (colored gold), alloyed with small amounts of M (M = Cu, Ni, Pd, Pt, Co, Rh, and Ir). The feedstock CO2 is first reduced to CO by gold or silver, and is subsequently captured and further reduced to C1 products by M.

There are at least two advantages for this type of catalyst. First, since CO2 reduction to CO, and the further reduction of CO to hydrocarbons are catalyzed by the same catalyst (the one-pot tandem catalyst), these SAAs are intrinsically simpler than a two-pot tandem approach, such as the one proposed by Karamad et al.(20) Second, because individual M atoms can accommodate only a single CO, and are isolated from each other by gold or silver, C–C coupling is unlikely. Therefore, the catalyst is expected to be selective for C1 hydrocarbons (such as CH4 and CH3OH), and may even be selective for a single C1 product, simplifying the challenge of product separation.
While our work is entirely computational, there are encouraging experimental precedents for the use of SAAs to catalyze a variety of chemical reactions, such as the electrochemical reduction of O2 to H2O2, the selective hydrogenation of organic compounds, and the Ullmann reaction of aryl halides.(30-39) For example, Jirkovsky et al. have synthesized Pd@Au SAA and used it for selective reduction of O2 to H2O2 with ∼95% selectivity.(38) Sykes and co-workers have synthesized Pd@Cu for the selective hydrogenation of styrene and acetylene(36) and Pt@Cu for 1,3-butadiene hydrogenation to butenes.(34) Also, Blaylock et al. have found Pd@Ag to be an efficient catalyst for the selective hydrogenation of acetylene and acrolein.(39)
The remainder of this paper is organized as follows. In Section 2, the computational approach that we employ is described. Our results are presented in Section 3, beginning with computational screening of 14 Au-based SAAs and 14 Ag-based SAAs, searching for alloys that exhibit high selectivity for CORR versus HER, as well as a relatively low applied bias to enable the endergonic CORR to proceed past the critical first reduction step. We then turn to a detailed mechanistic examination of two of the most promising alloys, showing that both of them appear to be selective for methane formation (over both methanol and hydrogen), as well as a discussion of some of the considerations that are relevant to possible future experimental preparation and study of these alloys.

2 Computational Details


The revised PBE functional (RPBE)(40) and the projector augmented wave pseudo-potentials,(41, 42) as implemented in the Vienna ab initio Simulation Package (VASP),(43-46) were employed for all slab and molecule calculations. The Gaussian-smearing technique (smearing temperatures kBT = 0.1 eV for slabs and 0.01 eV for molecules) was used to accelerate SCF convergence, after which all calculated values of energy were extrapolated to kBT = 0. A Monkhorst–Pack k-point net of 3 × 3 × 1 was chosen to sample the reciprocal space for the slab calculations, whereas only the gamma point was sampled for the molecule calculations. Test calculations (see Computational Details in the Supporting Information) indicated that a 2 × 2 × 1 Monkhorst–Pack k-point net yields chemically significant changes in binding energy (as large as 0.22 eV for ΔG*COH), while the changes upon extension to 5 × 5 × 1 are at least two times smaller. Extending the slab from three layers to four layers also yields similar energy changes of no more than ∼0.1 eV (see the Supporting Information), which are likely to be similar to inherent errors in the density functional itself.
To prevent interactions between the periodic replicas along the z-direction, a vacuum separation of at least 15 Å between adjacent images was used for the slab calculations, and a 20 Å × 20 Å × 20 Å box was used for molecular calculations. Spin-polarized wave functions were used for all surface calculations, while spin-restricted wave functions were used for molecular calculations. A dipole correction was not employed in this study, since test calculations showed that it has an insignificant effect on the binding energies of ΔG*H, ΔG*CHO, and ΔG*COH.
As already shown in Figure 1, the slab models for all SAAs were composed of three metal layers, where each layer contains a 3 × 3 periodic cell. One Au or Ag atom on the topmost layer is substituted by M, leading to a surface substituent concentration of 1/9 (or 11.1%). During the geometry optimization, the atoms on the bottom layer were fixed in their bulk positions (a separation of 4.073 Å for Au, and 4.086 Å for Ag), whereas the atoms on the two top layers and all adsorbate atoms were allowed to relax. Geometries were optimized in water, simulated using the Poisson–Boltzmann implicit solvation model with a dielectric constant of ε = 80.(47)
All energies discussed in this work are free energies. For slabs, they were calculated asThe vibrational frequencies were evaluated for only surface adsorbates and were calculated by evaluating the partial Hessian matrix by finite differences. Based on the calculated vibrational frequencies, the zero-point vibrational energy (ZPVE), and vibrational contributions to the internal energy and entropy at 25 °C were calculated. The free energy of molecules was evaluated aswhere n = 8 for nonlinear molecules and n = 7 for linear molecules. Eelecsolv was obtained using the same approach as that for the slab systems. However, the ZPVE, the vibrational component of internal energy, and Svib, Srot, and Strans were calculated by the Q-Chem package, using the same functional with the 6-31G* Gaussian-type basis set. Finally, it is well-known that gas-phase thermochemical reaction energies computed with the RPBE functional exhibit systematic deviations with experimental data.(48) Therefore, following previous theoretical work, a sensitivity analysis was performed for 21 chemical reactions,(49) and we found that a +0.38 eV correction for the electronic energy for CO2 is needed.
For the investigation of reaction mechanisms, we assumed that proton and electron transfer are coupled. We also assumed that the barriers for hydrogenation through electrochemical means are small and easily surmountable at room temperature; therefore, we only considered thermodynamics and ignored kinetics. This assumption is supported by recent theoretical work showing the kinetic barriers are small, particularly under the influence of the applied voltages.(50) To relate our calculated free-energy diagrams at U = 0.0 VRHE to different applied voltages, the computational hydrogen electrode (CHE) model proposed by Nørskov and co-workers was used.(51) In the CHE, the zero of voltage is defined to correspond to a proton and an electron in equilibrium with H2(g) at a pressure of 101 325 Pa. Using the standard relation between applied bias and free energy, the chemical potential of the proton and the electron at an applied bias (U) may then be expressed as μ(H+) + μ(e) = 1/2μ(H2(g)) – eU, where e is the electron charge.
To evaluate the performance of the DFT methods, we first calculated the equilibrium potentials for CO2 reduction to formic acid, CO, formaldehyde, methanol, and methane (Table 1), all of which are common intermediates during the reduction. As shown in Table 1, the equilibrium potentials were calculated to be −0.22, −0.11, −0.09, 0.03, and 0.17 VRHE for the reduction to formic acid, CO, formaldehyde, methanol, and methane, respectively. [Note: VRHE is the unit of applied voltage (vs RHE).] Those numbers are in excellent agreement with the experimental values of −0.20, −0.12, −0.07, 0.03, and 0.17 VRHE, respectively.(52)
Table 1. Experimental(52) and Calculated Equilibrium Potential for Some Two-, Four-, Six-, and Eight-Electron Reductions
  Equilibrium Potential, U (eV, vs RHE)
 reactionexperimenttheory
1CO2(g) + 2(H+ + e) → HCOOH(aq)–0.20–0.22
2CO2(g) + 2(H+ + e) → CO(g) + H2O(l)–0.12–0.11
3CO2(g) + 4(H+ + e) → CH2O(aq) + H2O(l)–0.07–0.09
4CO2(g) + 6(H+ + e) → CH3OH(aq) + H2O(l)0.030.03
5CO2(g) + 8(H+ + e) → CH4(G) + 2H2O(l)0.170.17

3 Results and Discussion


3.1Identification of Catalysts with a High Selectivity to CORR
As stated in the Introduction, the requirement for a catalyst to be selective for CORR over HER is that either ΔG*CHO < ΔG*H or ΔG*COH < ΔG*H, meaning that the catalyst binds with *CHO or *COH more tightly than with *H. Therefore, as the first step in our investigation, ΔG*CHO, ΔG*COH, and ΔG*H were calculated for all 28 SAAs and the corresponding applied voltages needed to hydrogenate *CO to *CHO and *COH were evaluated. We considered all possible adsorption sites; however, the values of ΔG*CHO, ΔG*COH, and ΔG*H reported in Table 2 are based on the most stable binding site (see Tables S1 and S2 in the Supporting Information).
Table 2. Calculated Binding Free Energy of *CHO (ΔG*CHO), *COH (ΔG*COH), and *H (ΔG*H), and the Applied Bias Necessary for the Reduction of *CO to *CHO (U*CHO) and *COH (U*COH)
catalystΔG*H (eV)ΔG*CHO (eV)ΔG*COH (eV)U*CHO (VRHE)U*COH (VRHE)
Cu@Au(111)0.370.731.22–0.61–1.10
Ni@Au(111)0.290.460.70–1.10–1.33
Pd@Au(111)0.380.521.06–0.69–1.23
Pt@Au(111)0.000.040.70–0.70–1.35
Co@Au(111)0.14–0.040.02–1.21–1.27
Rh@Au(111)–0.10–0.320.22–0.96–1.50
Ir@Au(111)–0.50–0.56–0.37–1.19–1.39
Cu@Au(100)0.230.641.34–0.75–1.44
Ni@Au(100)0.010.200.36–1.11–1.27
Pd@Au(100)0.090.240.89–0.69–1.34
Pt@Au(100)–0.14–0.210.20–0.75–1.16
Co@Au(100)–0.12–0.20–0.36–1.23–1.08
Rh@Au(100)–0.22–0.46–0.34–1.01–1.13
Ir@Au(100)–0.57–0.73–0.73–1.22–1.22
Cu@Ag(111)0.300.841.44–0.81–1.42
Ni@Ag(111)0.020.180.74–1.18–1.74
Pd@Ag(111)0.240.501.44–0.80–1.74
Pt@Ag(111)–0.14–0.060.93–0.71–1.70
Co@Ag(111)–0.10a–0.23a–1.50
Rh@Ag(111)–0.12–0.360.23–1.14–1.73
Ir@Ag(111)–0.56–0.71–0.36–1.28–1.63
Cu@Ag(100)0.310.771.69–0.88–1.80
Ni@Ag(100)–0.08–0.030.54–1.26–1.82
Pd@Ag(100)0.100.251.22–0.87–1.84
Pt@Ag(100)–0.15–0.240.78–0.83–1.84
Co@Ag(100)–0.26a–0.41a–1.44
Rh@Ag(100)–0.27–0.59–0.11–1.12–1.60
Ir@Ag(100)–0.63–0.89–0.63–1.26–1.52
a

Geometry optimization of *CHO leads to the formation of surface-bound H and CO species on Co.

Analyzing the results, we found that there are 10 alloys that favor CORR via the *CHO intermediate (i.e., ΔG*CHO < ΔG*H): Co@Au(111), Rh@Au(111), Ir@Au(111), Pt@Au(100), Rh@Au(100), Rh@Ag(111), Ir@Ag(111), Pt@Ag(100), Rh@Ag(100), and Ir@Ag(100). Three alloys favor CORR via the *COH intermediate (i.e., ΔG*COH < ΔG*H): Co@Au(100), Co@Ag(111), and Co@Ag(100). For Ir@Au(100), ΔG*CHO and ΔG*COH are the same, and both of them are smaller than ΔG*H. Two interesting trends are observed. First, most of the alloys (except Pt@Au(100) and Pt@Ag(100)) that favor CORR have M = Co, Rh, and Ir. This means that choosing group 9 transition metals as M effectively enhances CO reduction and suppresses HER. Second, for most of the 28 SAAs (except Co@Au(100), Co@Ag(111), and Co@Ag(100)), *CHO is more stable than *COH. This means that *CO → *CHO is the most energetically favorable hydrogenation pathway for *CO, similar to what was observed in the Cu(211) system.(49)
Also, it should be mentioned that very recent theoretical studies using the surface charging technique to simulate the applied voltage show that the binding energy of *OCHO is voltage-dependent, although the variation is small.(53, 54) It is possible that the binding energies of *H, *CHO, and *COH may also be slightly potential-dependent. This effect may affect our results for Ir@Au(111), Pt@Au(100), and Pt@Ag(100), since the values of ΔG*CHO for these catalysts are only slightly smaller than that of ΔG*H (by 0.06, 0.07, and 0.09 eV, respectively), and the stability of *CHO and *H could consequently be reversed for very negative applied voltages (e.g., −1.0 VSHE).
To understand the reason why alloys with M = Co, Rh, and Ir favor CORR over the HER, we plotted the scaling relations between ΔG*H and ΔG*CHO (Figure 2), and ΔG*H and ΔG*COH (Figure 3). Figures 2 and 3 show that the slopes for ΔG*CHO vs ΔG*H and ΔG*COH vs ΔG*H are 1.61 and 2.10, respectively. This indicates that when alloys bind strongly with H, they also bind with CHO and COH even more strongly, making CORR the major pathway. The larger slope for ΔG*COH vs ΔG*H (2.10), compared to that for ΔG*CHO vs ΔG*H (1.61), can be rationalized by the number of bonds formed between the COH and CHO and the surface: COH can form three bonds to the surface, while CHO can only form one.(55)
figure

Figure 2. Gibbs free binding energy of *CHO (ΔG*CHO) against that of *H (ΔG*H). The equation for the linear fitting line is ΔG*CHO = 1.61 × ΔG*H + 0.07, and the correlation coefficient (R2) is 0.90.

figure

Figure 3. Gibbs free binding energy of *COH (ΔG*CHO) against that of *H (ΔG*H). The equation for the linear fitting line is ΔG*COH = 2.10 × ΔG*H + 0.54, and R2 = 0.71.

3.2Mechanisms of CO Reduction to Methane Catalyzed by Rh@Au(100) and Rh@Ag(100)
Having identified several promising SAAs for CORR, we then selected two of them to study the reaction mechanism for further reduction of *CHO or *COH to C1 products. Rh@Au(100) was chosen from the Au-based alloys for two reasons. First, Rh@Au(100) requires a smaller applied voltage for the reaction *CO → *CHO (−1.01 V vs RHE), which suggests that its energy consumption is less than the others. Second, *CHO on the surface of this catalyst is 0.24 eV more stable than *H, which guarantees that most of the electrical energy is directed to CORR, in preference to HER. From among the Ag-based SAAs, we chose Rh@Ag(100) for detailed mechanistic study, because *CHO for Rh@Ag(100) is 0.32 eV more stable than *H, guaranteeing selectivity for CORR over HER. In addition, although the required applied voltage for *CO → *CHO (−1.12 VRHE) for Rh@Ag(100) is higher than that for Pt@Ag(100) (−0.83 VRHE), the method to synthesize Rh@Ag is known.(56)
There are numerous possible CORR routes leading to the formation of either the 4 × (H+ + e) reduction product, CH3OH(aq), or the 6 × (H+ + e) reduction product, CH4(g), from *CO, as depicted in Scheme 4. The first hydrogenation products are *CHO and *COH, while the second hydrogenation products are *CH2O, *CHOH, and *C + H2O(l), and the third hydrogenation products include *OCH3, *CH2OH, and *CH + H2O(l). The fourth hydrogenation products are *O + CH4(g), *CH3OH, and *CH2 + H2O(l), the fifth hydrogenation products are *OH + CH4(g), and *CH3 + H2O(l), and the sixth hydrogenation product is CH4(g) + H2O(l). Desorption of surface-bound CH3OH (*CH3OH) results in the formation of CH3OH(aq). In the absence of experimental information, we considered the thermodynamics for all possible pathways for Rh@Au(100) and Rh@Ag(100). However, the activation barrier for each elementary step was not determined.
figure
Scheme 4. Schematic Description of Possible Reduction Pathways for the Surface-Bound CO (*CO)a

aFour-electron reduction of *CO leads to the formation of methanol, whereas six-electron reduction results in the formation of methane.

Numerous DFT-based studies have been devoted to the investigation of the mechanisms of CO2RR catalyzed by heterogeneous catalysts.(20, 21, 29, 49, 50, 57-64) Since copper is the only material that can reduce CO2RR to hydrocarbons with significant conversion,(26, 65) most of the previous theretical studies have focused on the mechanism of CORR on Cu.(49, 50, 57) Two main pathways have emerged from these studies, differing primarily in the reaction intermediate formed by the reduction of *CO. Based on their investigation of CO2RR catalyzed by the Cu(211) surface, Nørskov and co-workers proposed a reaction mechanism involving the following transformations:(colored blue in Scheme 4).(49) By contrast, based on their analysis of CORR on Cu(111), Janik and co-workers proposed that CH4 involves the following transformations:(colored red in Scheme 4).(57) Recently, Goddard and co-workers have used implicit solvation plus applied electric voltage to revisit CO2RR catalyzed by the Cu(111) surface.(50) They propose that the hydrogenation of *CO leads to the formation of *COH, in agreement with Janik;(57) however, the subsequent reduction of *COH leads to the formation of *CHOH, which is different from the pathways identified by Nørskov(49) and Janik.(57)
Our calculations show that along the most favorable pathway (Figure 4) for CORR catalyzed by Rh@Au(100), the first (H+ + e) pair is added to the carbon of *CO to form *CHO, the second one is added to the oxygen to form *CHOH, and the third one is added to the oxygen to liberate H2O(l), leading to the formation of *CH. (The free-energy surface for all possible reduction pathways can be found in Figure S1 in the Supporting Information). These steps are followed by further hydrogenation of *CH to *CH2, *CH3, and eventually to the formation of gaseous CH4(g) and the regeneration of the surface alloy site (*). For the *CO, *CHO, *CHOH, and *CH3 intermediates, the adsorbates prefer to bind on the top site of Rh. However, *CHOH prefers to bind in the 4-fold hollow site formed by three Au and one Rh, and *CH2 prefers to bind on the bridged site formed by Au and Rh. Analyzing the potential energy surface, we found that the most uphill step (the potential-determining step) is the initial reduction: *CO → *CHO with ΔG = 1.01 eV. Since each step involves only a single electron transfer, with an applied bias of −1.01 VRHE, all the steps become downhill or thermoneutral.
figure

Figure 4. Gibbs free-energy diagram for *CO electrochemical reduction to methane catalyzed by Rh@Au(100) at 0.00 VRHE (black) and −1.01 VRHE (red). At −1.01 VRHE, the free-energy change for each step is either zero or downhill.

In contrast, CORR catalyzed by Rh@Ag(100) takes a very different pathway, one that is similar to that proposed by Nørskov and co-workers (Figure 5).(49) The free-energy surface for all possible reduction pathways is detailed in Figure S2 in the Supporting Information. In the beginning, four (H+ + e) pairs are successively added to the carbon of *CO to reduce it to *CHO, *CH2O, *OCH3, and then to * + free CH4(g). After those hydrogenation steps, a surface-bound oxygen (*O) is left behind, which is further reduced by two other (H+ + e) pairs, first to form *OH, and then to liberate H2O(l) and regenerate *. All of the adsorbates prefer to bind on the top site of Rh. From the calculated potential energy surface (PES), we found that the potential-determining step is *CO → *CHO with ΔG = 1.12 eV, and therefore, all the steps along the pathway become downhill or thermal neutral once a voltage of −1.12 VRHE is applied.
figure

Figure 5. Gibbs free-energy diagram for *CO electrochemical reduction to methane catalyzed by Rh@Ag(100) at 0.00 VRHE (black) and −1.12 VRHE (red). At an applied bias of −1.12 VRHE, the free-energy change for each step is either zero or downhill.

3.3Experimental Viability and Discussion
For the proposed M@Au (or Ag) SAAs, the single active site M is only involved in converting CO to C1 products, while the surrounding Au or Ag reduces CO2 and supplies CO for the M site. Therefore, another critical requirement for success of this type of one-pot tandem catalyst is whether the minimal electric voltage for CO reduction to CH4(g) or CH3OH(aq) by M matches the voltage for efficient CO2-to-CO reduction catalyzed by Au or Ag. If the minimum applied voltage required for the reduction CO to C1 products is too high or too low, compared to the voltage for the maximum Faradaic efficiency for CO2 reduction to CO, the SAAs will not be efficient and may not be selective for the CORR vs the HER.
For one of our proposed catalysts, Rh@Au(100), the minimum voltage needed to drive the CORR is −1.01 VRHE. Experimental measurements show that, at this voltage, the Faradaic efficiency for CO2 reduction to CO over Au is ∼60%.(25) Assuming all CO produced by surrounding Au is reduced by M, then this means that only ∼40% of the Faradaic efficiency is diverted to the HER. For Rh@Ag(100), the minimum bias needed to drive the CORR is −1.12 VRHE, for which the Faradaic efficiency for CO2 reduction to CO over Ag is as high as ∼90%.(17) This suggests that the HER channel may consume as little as 10% of the Faradaic efficiency (assuming that the reduction of CO on M is 100% efficient). These results suggest that our proposed one-pot tandem SAA catalysts are potentially efficient for CO2RR to CH4. A critical issue remaining unanswered by the present work is whether or not the rate of CO conversion to CH4 on the Rh sites embedded in the surface of Au or Ag is sufficiently active and sufficiently numerous to consume all, or a large fraction, of the CO generated on the host metal. To address this point, it will be necessary to determine the turnover frequencies (TOFs) for CO2 reduction to CO on Au and Ag, as well as the TOF for CO conversion to CH4. If the TOF for the latter reaction is sufficiently high, then it would be possible to achieve high overall Faradaic efficiencies for the reduction of CO2 to CH4 using Rh@Au or Rh@Ag catalysts.
Another important question about the viability of our proposed Rh@Au and Rh@Ag SAA catalysts is whether or not they can be synthesized and whether they would be stable. Rh and Au, and Rh and Ag are immiscible: while they can be mixed at high temperatures, phase separation occurs upon cooling.(66, 67) However, recent advances in synthetic techniques have overcome this difficulty.(56, 68, 69) In particular, Humphrey and co-workers have discovered a microwave-assisted method that enables the synthesis of stable, homogeneous Rh@Au and Rh@Ag nanoparticles with broadly tunable Rh:Au and Rh:Ag ratios.(56)
Subsurface segregation of M is another problem, which may deactivate the proposed SAA catalysts. Co (2.522 J/m2), Rh (2.659 J/m2), and Ir (3.048 J/m2) have larger surface free energies than Au (1.506 J/m2) and Ag (1.246 J/m2); therefore, they are likely to segregate into the subsurface under vacuum,(70, 71) leading to an inactive catalyst. However, since it is well-known that CO binds with Co, Rh, and Ir more strongly than Au and Ag do (indeed, that is the basis of the usefulness of the SAAs for CORR), those Group 9 metals may prefer to be present on the surface under CO-rich conditions.(29) Indeed, Sansa et al. have used DFT to study CO-induced segregation in gold-based alloys, and they have predicted that, under CO environments, Co, Rh, and Ir will remain on the surface of Au(111) and Au(100).(72)
The aggregation of active sites into islands is another potential problem for the stability of SAAs. To test whether or not this is the case, additional calculations were performed for Rh@Au(100), one of the two most promising SAAs for CO2RR. Rh@Au(100) was modeled by a three-layer 5 × 5 surface and the reciprocal space was sampled by a 2 × 2 × 1 Monkhorst-pack k-point net. During the geometry optimizations, the atoms in the bottom layer were fixed at their bulk positions, whereas the rest of atoms were allowed to relax. Two Au atoms on the top layer were replaced by Rh atoms, corresponding to an Rh surface concentration of 8%. Three surface configurations were considered: (a) two Rh atoms adjacent to each other along the y-axis, (b) two Rh atoms adjacent to each other along the xy-direction, and (c) two Rh atoms separated by Au atoms (see Figure 6). Our calculated results show that the three configurations are almost degenerate energetically, with configuration (c) being slightly more stable than configurations (a) and (b), by 0.01 and 0.02 eV, respectively. However, configuration (c) is expected to be more stable when entropy is taken into account. These results suggest that the proposed SAAs are stable to island formation.
figure

Figure 6. Configurations considered in evaluating the tendency of active sites toward island formation.

Finally, it is appropriate to revisit some of the uncertainties in the computational protocol itself. For example, as discussed in the Computational Details of the Supporting Information, we observe an ∼0.1 eV change in the difference in binding energy between H and CHO upon changing from a three-layer slab to a four-layer slab, as well as a similar shift upon increasing the slab thickness from three layers to four layers. While such shifts are not insignificant, they do not affect our main conclusions. The key point is that we find a free energy binding difference of nearly 0.4 eV in favor of CO2RR over HER for the best single-atom alloys (e.g., Rh@Au(100)). This gap is large enough that potential errors on the order of 0.1 eV will not qualitatively change it, even if they lead directly to stabilization of the HER pathway (which, of course, is not clear). The same argument suggests that our main conclusions are likely to be robust to other possible errors imposed by computational limitations, such as deficiencies of the RPBE density functional itself.

4 Conclusions


Twenty-eight (28) single-atom alloys (SAAs) were studied and evaluated as catalysts for the CO2 electrochemical reduction reaction to energy-rich hydrocarbons such as alkanes and alcohols in aqueous phase, using density functional theory (DFT) calculations together, with the Poisson–Boltzmann implicit solvation model. To be useful for the CO2 reduction reaction (CO2RR), the SAAs should contain isolated single-atom surface sites of M (M = Cu, Ni, Pd, Pt, Co, Rh, and Ir), as surface substitutions in Au or Ag. Initial reduction of CO2 to CO occurs on the Au or Ag host, and CO then binds preferentially to M. We found that 14 SAAs favor the formation of *CHO or *COH over *H, which, for those systems, means that CO reduction is preferred over proton reduction. This suggests that, for these catalysts, CO2 reduction is the major pathway, and the competitive evolution of hydrogen is suppressed. Most of the 14 promising SAAs contain M = Co, Rh, or Ir. This indicates that it is the identity of M rather than the surrounding metal that determines the performance of the SAA catalysts. The fact that M can bind only a single CO molecule also means that the SAAs will be selective for C1 products, relative to larger hydrocarbons.
We also performed a detailed thermodynamic study of the reaction mechanism for two of the most promising SAAs: Rh@Au(100) and Rh@Ag(100). We found that both of these SAAs reduce CO2 to methane, although the reduction occurs via different pathways. For Rh@Au(100), the reaction proceeds through the pathwaywhereas, for Rh@Ag(100), it isThe minimum applied voltages needed to drive the two electrocatalytic systems are −1.01 and −1.12 VRHE for Rh@Au(100) and Rh@Ag(100), respectively, for which the Faradaic efficiencies for CO2 reduction to CO are 60% for gold and 90% for silver. This suggests that SAA can efficiently reduce CO2 to methane with as little as 40% Faradaic efficiency loss in the hydrogen evolution reaction for Rh@Au(100) and as little as 10% for Rh@Ag(100). We believe that this computational work provides encouraging motivation for experimental efforts to prepare and characterize SAAs for the electrochemical reduction of CO2.

Supporting Information


The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscatal.6b01393.

  • Additional tables and details regarding computational methods (PDF)

The authors declare no competing financial interest.

Acknowledgment


This material is based on work performed in the Joint Center for Artificial Photosynthesis, a DOE Energy Innovation Hub, supported through the Office of Science of the U.S. Department of Energy, under Award No. DE-SC0004993.

References


This article references 72 other publications.

  1. 1.
    Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 1572915735, DOI: 10.1073/pnas.0603395103
  2. 2.
    Cook, T. R.; Dogutan, D. K.; Reece, S. Y.; Surendranath, Y.; Teets, T. S.; Nocera, D. G. Chem. Rev. 2010, 110, 64746502, DOI: 10.1021/cr100246c
  3. 3.
    Thomas, C. D.; Cameron, A.; Green, R. E.; Bakkenes, M.; Beaumont, L. J.; Collingham, Y. C.; Erasmus, B. F. N.; de Siqueira, M. F.; Grainger, A.; Hannah, L.; Hughes, L.; Huntley, B.; van Jaarsveld, A. S.; Midgley, G. F.; Miles, L.; Ortega-Huerta, M. A.; Peterson, A. T.; Phillips, O. L.; Williams, S. E. Nature 2004, 427, 145148, DOI: 10.1038/nature02121
  4. 4.
    Rakowski Dubois, M.; Dubois, D. L. Acc. Chem. Res. 2009, 42, 19741982, DOI: 10.1021/ar900110c
  5. 5.
    Morris, A. J.; Meyer, G. J.; Fujita, E. Acc. Chem. Res. 2009, 42, 19831994, DOI: 10.1021/ar9001679
  6. 6.
    Appel, A. M.; Bercaw, J. E.; Bocarsly, A. B.; Dobbek, H.; DuBois, D. L.; Dupuis, M.; Ferry, J. G.; Fujita, E.; Hille, R.; Kenis, P. J. A.; Kerfeld, C. A.; Morris, R. H.; Peden, C. H. F.; Portis, A. R.; Ragsdale, S. W.; Rauchfuss, T. B.; Reek, J. N. H.; Seefeldt, L. C.; Thauer, R. K.; Waldrop, G. L. Chem. Rev. 2013, 113, 66216658, DOI: 10.1021/cr300463y
  7. 7.
    White, J. L.; Baruch, M. F.; Pander, J. E., III; Hu, Y.; Fortmeyer, I. C.; Park, J. E.; Zhang, T.; Liao, K.; Gu, J.; Yan, Y.; Shaw, T. W.; Abelev, E.; Bocarsly, A. B. Chem. Rev. 2015, 115, 1288812935, DOI: 10.1021/acs.chemrev.5b00370
  8. 8.
    Qiao, J. L.; Liu, Y. Y.; Hong, F.; Zhang, J. J. Chem. Soc. Rev. 2014, 43, 631675, DOI: 10.1039/C3CS60323G
  9. 9.
    Li, C. W.; Ciston, J.; Kanan, M. W. Nature 2014, 508, 504507, DOI: 10.1038/nature13249
  10. 10.
    Schouten, K. J. P.; Qin, Z. S.; Gallent, E. P.; Koper, M. T. M. J. Am. Chem. Soc. 2012, 134, 98649867, DOI: 10.1021/ja302668n
  11. 11.
    Schouten, K. J. P.; Kwon, Y.; van der Ham, C. J. M.; Qin, Z.; Koper, M. T. M. Chem. Sci. 2011, 2, 19021909, DOI: 10.1039/c1sc00277e
  12. 12.
    Jedidi, A.; Rasul, S.; Masih, D.; Cavallo, L.; Takanabe, K. J. Mater. Chem. A 2015, 3, 1908519092, DOI: 10.1039/C5TA05669A
  13. 13.
    Rasul, S.; Anjum, D. H.; Jedidi, A.; Minenkov, Y.; Cavallo, L.; Takanabe, K. Angew. Chem., Int. Ed. 2015, 54, 21462150, DOI: 10.1002/anie.201410233
  14. 14.
    Elgrishi, N.; Chambers, M. B.; Fontecave, M. Chem. Sci. 2015, 6, 25222531, DOI: 10.1039/C4SC03766A
  15. 15.
    Zhang, Y. J.; Sethuraman, V.; Michalsky, R.; Peterson, A. A. ACS Catal. 2014, 4, 37423748, DOI: 10.1021/cs5012298
  16. 16.
    Li, C. W.; Kanan, M. W. J. Am. Chem. Soc. 2012, 134, 72317234, DOI: 10.1021/ja3010978
  17. 17.
    Hatsukade, T.; Kuhl, K. P.; Cave, E. R.; Abram, D. N.; Jaramillo, T. F. Phys. Chem. Chem. Phys. 2014, 16, 1381413819, DOI: 10.1039/C4CP00692E
  18. 18.
    Kim, D.; Resasco, J.; Yu, Y.; Asiri, A. M.; Yang, P. D. Nat. Commun. 2014, 5, 4948, DOI: 10.1038/ncomms5948
  19. 19.
    Cheng, M. J.; Kwon, Y.; Head-Gordon, M.; Bell, A. T. J. Phys. Chem. C 2015, 119, 2134521352, DOI: 10.1021/acs.jpcc.5b05518
  20. 20.
    Karamad, M.; Tripkovic, V.; Rossmeisl, J. ACS Catal. 2014, 4, 22682273, DOI: 10.1021/cs500328c
  21. 21.
    Tripkovic, V.; Vanin, M.; Karamad, M.; Björketun, M. E.; Jacobsen, K. W.; Thygesen, K. S.; Rossmeisl, J. J. Phys. Chem. C 2013, 117, 91879195, DOI: 10.1021/jp306172k
  22. 22.
    Rossmeisl, J.; Logadottir, A.; Norskov, J. K. Chem. Phys. 2005, 319, 178184, DOI: 10.1016/j.chemphys.2005.05.038
  23. 23.
    Carrasquillo-Flores, R.; Ro, I.; Kumbhalkar, M. D.; Burt, S.; Carrero, C. A.; Alba-Rubio, A. C.; Miller, J. T.; Hermans, I.; Huber, G. W.; Dumesic, J. A. J. Am. Chem. Soc. 2015, 137, 1031710325, DOI: 10.1021/jacs.5b05945
  24. 24.
    Lu, Q.; Rosen, J.; Zhou, Y.; Hutchings, G. S.; Kimmel, Y. C.; Chen, J. G. G.; Jiao, F. Nat. Commun. 2014, 5, 3242, DOI: 10.1038/ncomms4242
  25. 25.
    Hahn, C.; Abram, D. N.; Hansen, H. A.; Hatsukade, T.; Jackson, A.; Johnson, N. C.; Hellstern, T. R.; Kuhl, K. P.; Cave, E. R.; Feaster, J. T.; Jaramillo, T. F. J. Mater. Chem. A 2015, 3, 2018520194, DOI: 10.1039/C5TA04863J
  26. 26.
    Hori, Y.; Wakebe, H.; Tsukamoto, T.; Koga, O. Electrochim. Acta 1994, 39, 18331839, DOI: 10.1016/0013-4686(94)85172-7
  27. 27.
    Chen, Y. H.; Li, C. W.; Kanan, M. W. J. Am. Chem. Soc. 2012, 134, 1996919972, DOI: 10.1021/ja309317u
  28. 28.
    Zhu, W. L.; Michalsky, R.; Metin, O.; Lv, H. F.; Guo, S. J.; Wright, C. J.; Sun, X. L.; Peterson, A. A.; Sun, S. H. J. Am. Chem. Soc. 2013, 135, 1683316836, DOI: 10.1021/ja409445p
  29. 29.
    Peterson, A. A.; Norskov, J. K. J. Phys. Chem. Lett. 2012, 3, 251258, DOI: 10.1021/jz201461p
  30. 30.
    Lucci, F. R.; Lawton, T. J.; Pronschinske, A.; Sykes, E. C. H. J. Phys. Chem. C 2014, 118, 30153022, DOI: 10.1021/jp405254z
  31. 31.
    Lucci, F. R.; Marcinkowski, M. D.; Lawton, T. J.; Sykes, E. C. H. J. Phys. Chem. C 2015, 119, 2435124357, DOI: 10.1021/acs.jpcc.5b05562
  32. 32.
    Pei, G. X.; Liu, X. Y.; Wang, A. Q.; Li, L.; Huang, Y. Q.; Zhang, T.; Lee, J. W.; Jang, B. W. L.; Mou, C. Y. New J. Chem. 2014, 38, 20432051, DOI: 10.1039/c3nj01136d
  33. 33.
    Pei, G. X.; Liu, X. Y.; Wang, A. Q.; Lee, A. F.; Isaacs, M. A.; Li, L.; Pan, X. L.; Yang, X. F.; Wang, X. D.; Tai, Z. J.; Wilson, K.; Zhang, T. ACS Catal. 2015, 5, 37173725, DOI: 10.1021/acscatal.5b00700
  34. 34.
    Lucci, F. R.; Liu, J. L.; Marcinkowski, M. D.; Yang, M.; Allard, L. F.; Flytzani-Stephanopoulos, M.; Sykes, E. C. H. Nat. Commun. 2015, 6, 8550, DOI: 10.1038/ncomms9550
  35. 35.
    Boucher, M. B.; Zugic, B.; Cladaras, G.; Kammert, J.; Marcinkowski, M. D.; Lawton, T. J.; Sykes, E. C. H.; Flytzani-Stephanopoulos, M. Phys. Chem. Chem. Phys. 2013, 15, 1218712196, DOI: 10.1039/c3cp51538a
  36. 36.
    Kyriakou, G.; Boucher, M. B.; Jewell, A. D.; Lewis, E. A.; Lawton, T. J.; Baber, A. E.; Tierney, H. L.; Flytzani-Stephanopoulos, M.; Sykes, E. C. H. Science 2012, 335, 12091212, DOI: 10.1126/science.1215864
  37. 37.
    Zhang, L.; Wang, A.; Miller, J. T.; Liu, X.; Yang, X.; Wang, W.; Li, L.; Huang, Y.; Mou, C. Y.; Zhang, T. ACS Catal. 2014, 4, 15461553, DOI: 10.1021/cs500071c
  38. 38.
    Jirkovsky, J. S.; Panas, I.; Ahlberg, E.; Halasa, M.; Romani, S.; Schiffrin, D. J. J. Am. Chem. Soc. 2011, 133, 1943219441, DOI: 10.1021/ja206477z
  39. 39.
    Aich, P.; Wei, H. J.; Basan, B.; Kropf, A. J.; Schweitzer, N. M.; Marshall, C. L.; Miller, J. T.; Meyer, R. J. Phys. Chem. C 2015, 119, 1814018148, DOI: 10.1021/acs.jpcc.5b01357
  40. 40.
    Hammer, B.; Hansen, L. B.; Norskov, J. K. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 74137421, DOI: 10.1103/PhysRevB.59.7413
  41. 41.
    Blochl, P. E. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 1795317979, DOI: 10.1103/PhysRevB.50.17953
  42. 42.
    Kresse, G.; Joubert, D. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 17581775, DOI: 10.1103/PhysRevB.59.1758
  43. 43.
    Kresse, G.; Furthmuller, J. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 1116911186, DOI: 10.1103/PhysRevB.54.11169
  44. 44.
    Kresse, G.; Furthmuller, J. Comput. Mater. Sci. 1996, 6, 1550, DOI: 10.1016/0927-0256(96)00008-0
  45. 45.
    Kresse, G.; Hafner, J. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 558561, DOI: 10.1103/PhysRevB.47.558
  46. 46.
    Kresse, G.; Hafner, J. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 49, 1425114269, DOI: 10.1103/PhysRevB.49.14251
  47. 47.
    Mathew, K.; Hennig, R. G. ArXiv 2016, 1601.03346
  48. 48.
    Blaylock, D. W.; Ogura, T.; Green, W. H.; Beran, G. J. O. J. Phys. Chem. C 2009, 113, 48984908, DOI: 10.1021/jp806527q
  49. 49.
    Peterson, A. A.; Abild-Pedersen, F.; Studt, F.; Rossmeisl, J.; Norskov, J. K. Energy Environ. Sci. 2010, 3, 13111315, DOI: 10.1039/c0ee00071j
  50. 50.
    Xiao, H.; Cheng, T.; Goddard, W. A.; Sundararaman, R. J. Am. Chem. Soc. 2016, 138, 483486, DOI: 10.1021/jacs.5b11390
  51. 51.
    Nørskov, J. K.; Rossmeisl, J.; Logadottir, A.; Lindqvist, L.; Kitchin, J. R.; Bligaard, T.; Jonsson, H. J. Phys. Chem. B 2004, 108, 1788617892, DOI: 10.1021/jp047349j
  52. 52.
    Oloman, C.; Li, H. ChemSusChem 2008, 1, 385391, DOI: 10.1002/cssc.200800015
  53. 53.
    Steinmann, S. N.; Michel, C.; Schwiedernoch, R.; Sautet, P. Phys. Chem. Chem. Phys. 2015, 17, 1394913963, DOI: 10.1039/C5CP00946D
  54. 54.
    Steinmann, S. N.; Michel, C.; Schwiedernoch, R.; Filhol, J. S.; Sautet, P. ChemPhysChem 2015, 16, 23072311, DOI: 10.1002/cphc.201500187
  55. 55.
    Abild-Pedersen, F.; Greeley, J.; Studt, F.; Rossmeisl, J.; Munter, T. R.; Moses, P. G.; Skulason, E.; Bligaard, T.; Norskov, J. K. Phys. Rev. Lett. 2007, 99, 016105, DOI: 10.1103/PhysRevLett.99.016105
  56. 56.
    Garcia, S.; Zhang, L.; Piburn, G. W.; Henkelman, G.; Humphrey, S. M. ACS Nano 2014, 8, 1151211521, DOI: 10.1021/nn504746u
  57. 57.
    Nie, X. W.; Esopi, M. R.; Janik, M. J.; Asthagiri, A. Angew. Chem., Int. Ed. 2013, 52, 24592462, DOI: 10.1002/anie.201208320
  58. 58.
    Back, S.; Kim, H.; Jung, Y. ACS Catal. 2015, 5, 965971, DOI: 10.1021/cs501600x
  59. 59.
    Shi, C.; O’Grady, C. P.; Peterson, A. A.; Hansen, H. A.; Norskov, J. K. Phys. Chem. Chem. Phys. 2013, 15, 71147122, DOI: 10.1039/c3cp50645b
  60. 60.
    Luo, W. J.; Nie, X. W.; Janik, M. J.; Asthagiri, A. ACS Catal. 2016, 6, 219229, DOI: 10.1021/acscatal.5b01967
  61. 61.
    Karamad, M.; Hansen, H. A.; Rossmeisl, J.; Norskov, J. K. ACS Catal. 2015, 5, 40754081, DOI: 10.1021/cs501542n
  62. 62.
    Hirunsit, P.; Soodsawang, W.; Limtrakul, J. J. Phys. Chem. C 2015, 119, 82388249, DOI: 10.1021/acs.jpcc.5b01574
  63. 63.
    Lim, H. K.; Shin, H.; Goddard, W. A.; Hwang, Y. J.; Min, B. K.; Kim, H. J. Am. Chem. Soc. 2014, 136, 1135511361, DOI: 10.1021/ja503782w
  64. 64.
    Bernstein, N. J.; Akhade, S. A.; Janik, M. J. Phys. Chem. Chem. Phys. 2014, 16, 1370813717, DOI: 10.1039/C4CP00266K
  65. 65.
    Kuhl, K. P.; Hatsukade, T.; Cave, E. R.; Abram, D. N.; Kibsgaard, J.; Jaramillo, T. F. J. Am. Chem. Soc. 2014, 136, 1410714113, DOI: 10.1021/ja505791r
  66. 66.
    Zarkevich, N. A.; Tan, T. L.; Johnson, D. D. Phys. Rev. B 2007, 75, 104203, DOI: 10.1103/PhysRevB.75.104203
  67. 67.
    Davies, R. H.; Dinsdale, A. T.; Gisby, J. A.; Robinson, J. A. J.; Martin, S. M. CALPHAD: Comput. Coupling Phase Diagrams Thermochem. 2002, 26, 229271, DOI: 10.1016/S0364-5916(02)00036-6
  68. 68.
    Essinger-Hileman, E. R.; DeCicco, D.; Bondi, J. F.; Schaak, R. E. J. Mater. Chem. 2011, 21, 1159911604, DOI: 10.1039/c0jm03913f
  69. 69.
    Kusada, K.; Yamauchi, M.; Kobayashi, H.; Kitagawa, H.; Kubota, Y. J. Am. Chem. Soc. 2010, 132, 1589615898, DOI: 10.1021/ja107362z
  70. 70.
    Tyson, W. R.; Miller, W. A. Surf. Sci. 1977, 62, 267276, DOI: 10.1016/0039-6028(77)90442-3
  71. 71.
    Vitos, L.; Ruban, A. V.; Skriver, H. L.; Kollar, J. Surf. Sci. 1998, 411, 186202, DOI: 10.1016/S0039-6028(98)00363-X
  72. 72.
    Sansa, M.; Dhouib, A.; Guesmi, H. J. Chem. Phys. 2014, 141, 064709, DOI: 10.1063/1.4891869

Metrics

Article Views: 1,181 Times
Received 18 May 2016
Published online 23 September 2016
Published in print 4 November 2016
Learn more about these metrics Article Views are the COUNTER-compliant sum of full text article downloads since November 2008 (both PDF and HTML) across all institutions and individuals. These metrics are regularly updated to reflect usage leading up to the last few days.

The Altmetric Attention Score is a quantitative measure of the attention that a research article has received online. Clicking on the donut icon will load a page at altmetric.com with additional details about the score and the social media presence for the given article. Find more information on the Altmetric Attention Score and how the score is calculated.
+
More Article Metrics
Explore by: