Subscriber access provided by NEW YORK UNIV
Letter
A Fundamental Relationship between Reaction Mechanism and Stability in Metal Oxide Catalysts for Oxygen Evolution Xi Rong, Jules Parolin, and Alexie M Kolpak ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.5b02432 • Publication Date (Web): 14 Jan 2016 Downloaded from http://pubs.acs.org on January 18, 2016
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 7
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
A Fundamental Relationship between Reaction Mechanism and Stability in Metal Oxide Catalysts for Oxygen Evolution Xi Rong, Jules Parolin,[a] and Alexie M. Kolpak* Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States. [a] Present address: MINES ParisTech, Paris, France. ABSTRACT: Rational design of efficient, stable oxygen evolution reaction (OER) catalysts is necessary for widespread adoption of electrochemical energy storage technologies. Achieving this goal requires elucidation of fundamental relationships between surface structure and reaction mechanism. Here we address this issue using ab initio computations to determine the surface structure and OER mechanism for LaNiO3, a perovksite oxide that exhibits high activity but low stability. We find a new OER mechanism in which lattice oxygen participation via reversible formation of surface oxygen vacancies is critical. We show that this mechanism has a lower reaction barrier compared to the generally proposed mechanism, leading to improved agreement with experimental data. Extending the study to La1-xSrxBO3 (B = transition metal), we demonstrate a transition to the lattice oxygen-mediated mechanism with decreasing catalyst stability. Our results suggest new approaches for next-generation catalysts design.
KEYWORDS. Electrocatalysis, oxygen evolution reaction mechanism, perovskite oxides, density functional theory, surface structure The design of active, stable and inexpensive electrocatalysts for the oxygen evolution reaction (OER) is crucial for the development of electrochemical energy conversion technologies such as electrolysis and metal-air batteries. Among the various classes of OER electrocatalysts, perovskite-based metal oxides with the chemical formula ABO3 (A = alkaline earth, B = transition metal) have attracted substantial attention due to reports of high activity comparable to or greater than state-of-the-art precious metal OER catalysts such as IrO21-5. Furthermore, the ability to synthesize perovskites with a wide range of compositions provides exciting possibilities for tuning catalytic properties. However, exploring these possibilities in a systematic manner requires a deeper understanding of compositionstructure-property relationships. A number of studies have investigated such relationships. Suntivich et al. experimentally measured OER activity for ABO3 catalysts with A = La or Sr with various B cations6, showing that OER activity peaks when the antibonding-like eg state is half-filled, and decreases for materials with either lower or higher eg occupation. Based on this work, mixed cation catalysts such as Ba0.5Sr0.5Co0.8Fe0.2O3-δ (BSCF)6, PrBaCo2O5+δ (PBCO)7 and SrNb0.1Co0.7Fe0.2O3-δ8 were found to have enhanced activity. However, materials with high OER activity tend to have low stability, exhibiting remarkable activity change in a few cycles9-11. Furthermore, experiments have shown that different perovskites with the same optimal eg occupation in fact exhibit very different activities5-8. These exceptions to the trend have been experimentally correlated to the bulk stability of the oxide3, 12, indicating that understanding the origin of the relationship between catalyst activity and stability will be crucial for enabling rational design of highly active cata-
lysts that can also undergo many thousands of cycles for applications. Developing this understanding is challenging due to the complexity of oxide surfaces in aqueous conditions, and, critically, the elusiveness of the OER mechanism, particularly for the most active ABO3 materials. DFT has been widely used to investigate surface reactivity of ABO3 catalysts, identifying correlations between activity and adsorption strength of reaction intermediates and electronic structure10, 13-16. However, most previous studies have considered model ABO3(001) surfaces terminated by stoichiometric AO or BO2 planes, which may not be the relevant surfaces17, 18. Further, computational studies have generally assumed the OER mechanism from Refs. 6 and 13, in which O2 (g) evolves via –OH, –O, –OOH, and –OO intermediate adsorbates bound to surface transition cations6, 19-21, although it was suggested that OER on the most active oxides may proceed via an alternate mechanism13. For example, BSCF and PBCO are known to exhibit fast oxygen bulk diffusion rates and surface exchange kinetics22, 23. This suggests the flexibility of their lattice oxygen involved into OER. Recently, a new mechanism involving surface lattice hydroxides was proposed5, 24 based on the hypothesis that surface oxygen is partially protonated during OER on LaNiO3, a perovskite with activity comparable to that of IrO2. In addition, experiments have observed direct lattice oxygen participation in OER over IrO225, RuO226 and Ru0.9Ni0.1O2-δ27 catalysts. When prepared in air, LaNiO3 is found to exhibit oxygen vacancies (V0), indicative of weakly bound lattice oxygens and low bulk stability28. Thus, surface lattice oxygens may be involved in OER on this material as well, as suggested in several different studies24, 29, 30. In this work, we thus employ our recently developed ab initio approach17 to deter-
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
mine the realistic surface structure and reaction mechanism(s) for OER on LaNiO3, particularly in alkaline electrolyte at room temperature. We find a new OER mechanism in which the lattice oxygen is an active participant. This mechanism, referred to below as the lattice oxygen mechanism (LOM), has a significantly smaller energy barrier than the generally proposed adsorbates evolution mechanism (AEM) on LaNiO3, making the predicted OER activity of LaNiO3 relative to other ABO3 consistent with experimental results. Extending to a series of ABO3 catalysts, we show that the OER mechanism transitions from AEM to LOM with decreasing ABO3 bulk stability, as formation of the surface VO intermediate becomes increasingly favorable due to ineffective electron donation by the B-site cation. The relationship between mechanism and ABO3 stability elucidated in this work provides insights into new approaches for design of stable and efficient ABO3 catalysts. To elucidate the OER mechanism on the surface of LaNiO3 (Figure S1), we begin by comparing the free energies of each AEM intermediate with various isomers. The isomers are constructed such that lattice oxygens participate in the AEM adsorbates (Figure S2). As the reaction of H2O(l) + e- = -H + OH-(aq) could be energetically favorable (Figure S3), surfaces are initially modeled with one -H adsorbate per 2x2 cell, i.e., ¼ monolayer (ML), on the (001) NiO2-termination. Here, -H indicates the H adsorbed onto the surface lattice O, which is thus defined as protonated. The surface protonation arises from the inability of the Bsite to effectively donate electrons to oxygens, particularly those in the surface layer, which are not fully coordinated. We also initially consider ¼ ML OER intermediates on the ¼ ML –H surfaces (Figure 1). These surfaces thus include all the necessary adsorbates for OER, enabling direct energetic comparison of isomeric configurations. Subsequently, we investigate the stable surface stoichiometry in more detail, constructing a surface phase diagram as a function of pH and applied potential. All free energies are computed using DFT calculations as implemented in the VASP31 code with PAW pseudopotentials32 and the RPBE-GGA functional33. As discussed below, we find that using the GGA+U34 functional does not qualitatively change our mechanism. Further details of the computations and surface models are in the supporting information (SI).
Figure 1. Configurations and relative stabilities of several isomers of the (a) –O and (b) –OH AEM intermediates on the partially hydrogenated NiO2-terminated LaNiO3(001) surface.
Page 2 of 7
Figure 1a shows the surface denoted A0, which has –O on ¼ of the surface Ni, as well as two of its isomers, A1 and A2, with different surface configurations. (We note that intermediates may occupy all surface Ni under OER conditions17, 35.) Starting from A0, an unprotonated oxygen is removed from the NiO2 surface plane and placed on top of –O to form A1, which thus has ¼ ML –OO, V0, and -H. Similarly, A2 is formed by removing a protonated oxygen (i.e., hydroxide) from A0 and forming –OOH. Thus, the surface configuration of A2 is ¼ ML –OOH and V0. Figure 1b illustrates a set of isomers based on the –OH AEM intermediate. Surface B0 has ¼ ML –OH and -H. As above, B1 is obtained from B0 by removing an unprotonated surface lattice oxygen, thereby forming –OOH, while B2 is obtained by removing a protonated oxygen to form adsorbed H2O229. Other sets of isomers (not shown), are similarly constructed starting from the –OO and -OOH AEM intermediates. In addition, we consider non-stoichiometric intermediates, e.g., A0 and B0 with surface V0, which could form due to electrochemical exchange with the solvent. As discussed below, we find that these are unstable.
Figure 2. Projected DOS of the active surface Ni and ligand oxygens for A0, A1, and A2. Spin up and spin down pDOS are represented with positive and negative values, respectively.
As Fig. 1a shows, A1 is -0.76 eV/intermediate lower in energy than A0; this energy difference decreases to -0.9 eV/intermediate without the surface protonation. The origin of the greater stability of A1 is apparent in the projected density of states (pDOS) of the active surface Ni and its ligand O, shown in Fig. 2. The large overlap of the Ni and O pDOS for all of the surfaces indicates strong Ni-O covalent bonding character. The energy differences between the isomers are primarily due to different occupation of the bonding states. Due to the double bonded O adsorbate, the active Ni on the A0 surface has a nominal 4+ valence. This is manifest in the partial occupation of the spin-down bonding state of A0. On the A1 surface, however, the adsorbed –OO has an O-O bond distance of 1.28 Å, close to that of superoxide, indicating that it is bound to Ni via a single bond, giving a nominal surface Ni3+. As Ni has only two unpaired electrons in the outer shell (Figure S4), it is
ACS Paragon Plus Environment
Page 3 of 7
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
more stable in this oxidation state; indeed, A1 has a highly stable electronic configuration, with fully occupied (unoccupied) bonding (antibonding) state. We note that using the functional of GGA+U increases the stability of A1 even more (Figure S5). The A2 surface suggested in Ref. 24 is 0.3 eV higher in energy than A0, as the protonated oxygens on the A0 surface are in the preferred oxidation state; thus, rearranging the surface hydroxides onto the adsorbates does not stabilize the surface. We also find that isomers of the single bonded –OH intermediate, such as B1 and B2 suggested in Ref. 29, are not expected to form, as they do not have reduced bond numbers of adsorbates than B0. Similarly, isomers of the single bonded –OO and –OOH AEM intermediates are not more stable than the original configuration. Based on identification of the lower energy A1 surface, we propose the lattice oxygen participated mechanism (LOM) shown schematically in Fig. 3b. The AEM is shown in Fig. 3a for comparison. In both mechanisms, OER starts from the fully hydroxylated NiO2 termination identified in Refs. 6 and 24. In the first reaction step of the LOM, however, the initial hydroxylated surface becomes A1 instead of going to A0 as in the AEM. In this LOM step, the electrochemically driven dehydrogenation of adsorbed –OH tends to oxidize the surface Ni from 3+ to 4+, which thus simultaneously drives the chemical formation of A1, maintaining the surface Ni3+. Upon dehydrogenation, the O of the adsorbed –OH moves from the Ni site to the Ni-O bridge site to form peroxide-like intermediates, promptly drawing the lattice O out of the surface to form the superoxide-like A1 surface (Figure S6). Thus, the chemical force, arising from ineffective electron donation by Ni, drives the formation of –OO, the embryonic stage of O2 (g). In the second LOM step, the adsorbed –OO evolves back to –OH, giving away O2 (g). Once this occurs, the V0 becomes unstable, leading to the third step, in which OH-(aq) fills the V0 and protonates an adjacent surface lattice O, giving e- to the solvent (the electron leaving is induced by the electrode potential in the closed circuit). Finally (step 4), deprotonation takes place, restoring the initial surface. For the ABO3 with bulk O vacancies, OER could proceed from the third LOM step at the surface vacancy sites. For the ABO3 starting with stable surface protonation (not shown in Figure 3), deprotonation could take place at step 3 due to the formation of a surface O vacancy, followed by the vacancy filling step.
applied potential USHE assuming thermodynamic equilibrium between bulk LaNiO3, solvent molecules, and solvated La and Ni cations, as exchange of atoms at the oxide-water interface can significantly affect surface stoichiometry and adsorption of intermediates, changing reaction barriers17 (SI). For example, the presence of –H adsorbates could reduce the relative stability of A1 over A0 (Figure 1a and S6). The surface formation energy is ∑ , where ∆ is determined by DFT using ∑ ° , where ( ) is the free energy of the new (reference) surface, ° is the standard chemical potential of species A (A could be any atoms), and is the number of exchanged atoms of A. is the electrochemical energy for A to form the most stable H AO species at pH and ° USHE (Table S2), given by !"# $ ° 2.3 ( )*+, )* - ./012345 , where is the experimental formation energy relative to standard hydrogen electrode, ./012345 is the ion activity, and ( ( ) is the number of transferred electrons (protons).
Figure 4. (a) Computed surface phase diagram relevant to AEM for LNO in aqueous solution. Surface structures are labelled in each colored region. (b) Reaction pathway on R4. (c) Reaction pathway on R5. The active surface Ni and O ligands are shown for each intermediate. Figure 3. Schematics of (a) the adsorbates evolution mechanism and (b) the lattice-oxgyen participated mechanism (b). Red and black atoms are in/from the lattice and the solvent, respectively.
To compare AEM and LOM barriers, we first determine the surface structure of LaNiO3 as a function of pH and
Using the ideal NiO2 termination as the reference, we compare for a range of possible surfaces relevant to AEM (Figure S7), including different concentrations of –H and OER intermediates (-O and -OH) from ¼ to 1 ML. Figure 4a shows the resulting surface phase diagram. With
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
increasing USHE and pH, the stable surface is increasingly oxidized, evolving from La-deficient R1 to R2 (partially hydroxylated R1), R3 and R4 (fully hydroxylated NiO2 terminations with 1 and ½ ML of protonated surface lattice oxygens, respectively), and R5 and R6 (fully hydroxylated and –O covered NiO2 termination, respectively). The surface R3, with 1 ML –OH and 1 ML –H, can be regarded as the result of water dissociation. The retention of –H in both R3 and R4 is due to the significant H adsorption strength as electron donors to the lattice O, which itself increases with decreasing concentration. Upon oxidation, the protonation disappears in R5, followed by the formation of –O in R6 at URHE = 1.96 V; this is the first step of AEM. Critically, the diagram shows that R3, R4 and R5 lie above the condition of URHE = 1.23 V and below the first step of AEM, indicating that these surfaces could be the initial surfaces on which LOM occurs. Our calculations show that the heavily protonated R3 has the LOM overpotential only outside the condition for stable R3 (URHE < 1.40 V); it is thus not discussed here. Using R4 and R5 as the initial LOM and AEM surfaces, respectively, we compute the corresponding free energies of each LOM and AEM step at URHE = 1.23 V. The largest free energy is the estimated overpotential, η, in accordance with the procedure in Ref. 13 (Table S3). For OER on R4 (Figure 4b), step 1 is the potential-determining step for AEM, with η = 0.73 V. However, step 1 is thermodynamicdownhill for LOM, in which lattice O participates in the adsorbates, forming a surface oxygen vacancy, VO. Therefore, the LOM is the relevant mechanism. (Note that η for the AEM is different from Ref 13, as the latter is computed assuming lower coverage of intermediates, Table S3). Once the VO forms in LOM, the retained protonation from R4 becomes less stable and is ionized into the solvent (step 3). This is followed by filling of the vacancy site (step 4), which results in a switch in the protonation site on R4 during each OER cycle. This switch does not affect the energetics. The overpotential of 0.31 V given by LOM is well within the condition of stable R4 (1.40V
