Surface Proton Transfer Promotes Four-Electron Oxygen Reduction on

May 11, 2017 - Four-electron oxygen reduction reaction (4e-ORR), a key pathway in energy conversion, is preferred over the two-electron reduction path...
0 downloads 15 Views 6MB Size
Download PDF
Article pubs.acs.org/JACS

Surface Proton Transfer Promotes Four-Electron Oxygen Reduction on Gold Nanocrystal Surfaces in Alkaline Solution Fang Lu,†,§ Yu Zhang,‡,§ Shizhong Liu,‡ Deyu Lu,† Dong Su,† Mingzhao Liu,† Yugang Zhang,† Ping Liu,‡ Jia X. Wang,‡ Radoslav R. Adzic,*,‡ and Oleg Gang*,†,#,⊥ †

Center for Functional Nanomaterials, Energy & Photon Sciences Directorate, Brookhaven National Laboratory, Upton, New York 11973, United States ‡ Chemistry Division, Energy & Photon Sciences Directorate, Brookhaven National Laboratory, Upton, New York 11973, United States # Department of Chemical Engineering, Columbia University, New York, New York 10027, United States ⊥ Department of Applied Physics and Applied Mathematics, Columbia University, New York, New York 10027, United States S Supporting Information *

ABSTRACT: Four-electron oxygen reduction reaction (4e-ORR), a key pathway in energy conversion, is preferred over the two-electron reduction pathway that falls short in dissociating dioxygen molecules. Gold surfaces exhibit high sensitivity of the ORR pathway to its atomic structures. A long-standing puzzle remains unsolved: why the Au surfaces with {100} sub-facets were exceptionally capable to catalyze the 4e-ORR in alkaline solution, though limited within a narrow potential window. Herein we report the discovery of a dominant 4e-ORR over the whole potential range on {310} surface of Au nanocrystal shaped as truncated ditetragonal prism (TDP). In contrast, ORR pathways on single-crystalline facets of shaped nanoparticles, including {111} on nano-octahedra and {100} on nanocubes, are similar to their singlecrystal counterparts. Combining our experimental results with density functional theory calculations, we elucidate the key role of surface proton transfers from co-adsorbed H2O molecules in activating the facet- and potential-dependent 4e-ORR on Au in alkaline solutions. These results elucidate how surface atomic structures determine the reaction pathways via bond scission and formation among weakly adsorbed water and reaction intermediates. The new insight helps in developing facet-specific nanocatalysts for various reactions.

■

on Au(100)8 and the high-index Au(910) and Au(11,1,1) single-crystal surfaces with a large fraction of the {100} subfacet9 while a 2e reduction to hydroperoxide ions (O2H−) proceeds on Au(111)10 and other surfaces. It still remains elusive why the {100} facets can activate the 4e-ORR on Au in alkaline media at high potentials, but unfortunately only in a confined potential region beyond which the 2e-ORR takes over. A clear answer to this long-standing question can enhance fundamental understanding of the reaction mechanisms for facet-dependent electrocatalytic reactions. The traditional method of preparing single-crystal surfaces involves orienting and cutting bulk crystals followed by polishing and annealing the surfaces. A small miscut can cause significant deviation from the desired high-index surfaces. Shape-controlled syntheses of metallic nanocrystals developed in recent years11−14 provided a new route in preparing highquality single-crystal facets for studying structural sensitive

INTRODUCTION The surface atomic structures of metal catalysts play key roles in determining the kinetics and mechanisms of catalytic reactions.1,2 Bulk single-crystal surfaces have been employed extensively as model systems for studying the intrinsic correlations between surface atomic structures and various catalytic activities. In oxygen reduction reaction (ORR), a key reaction in energy conversion devices such as alkaline fuel cells,3 a complete four-electron (4e) ORR is preferred because of two-fold efficiency of energy conversion in the comparison with the partial two-electron (2e) ORR.4 On active metal catalysts, such as platinum and palladium, the 4e-ORR dominates regardless of their surface atomic structures because the molecular oxygen (O2) reactant dissociates by the strong metal−oxygen interaction. The 4e-ORR activity then is limited by desorption of the reaction intermediates that occupy active surface sites. In contrast, gold is not active enough to dissociate O2 because of its weak interaction with O2.5−7 However, it has been known for long time that a 4e reduction to hydroxide ions (OH−) in alkaline solutions occurs in a limited potential region © 2017 American Chemical Society

Received: February 18, 2017 Published: May 11, 2017 7310

DOI: 10.1021/jacs.7b01735 J. Am. Chem. Soc. 2017, 139, 7310−7317

Article

Journal of the American Chemical Society catalysts.2,15,16 It is also important from practical perspective due to the high surface area of nanoparticle-based materials. However, a challenge to the bottom-up synthetic approaches is associated with the surface cleanness and its heterogeneity. For example, the elimination of the stabilizers and/or capping agents17 used during syntheses requires extraordinary cleaning procedures,18,19 as well as the surface structure might degrade after extensive post-treatment. Surfactants and ionic impurities, could detrimentally block the adsorption of reactants to the catalyst surface, and, consequently, inhibit the ORR activities.20 Nevertheless, as we show here, both requirements, on the facet structure and surface cleanness can be satisfied, and that permits the observation of novel catalytic behavior. In this study, we synthesized three monofacet Au nanocrystals via controlling the particles’ shape. The octahedra and cubes enclosed with {111} and {100} surfaces, respectively, were used to demonstrate that clean single-crystalline surfaces can be obtained for generating ORR behaviors similar to those on bulk Au single-crystal surfaces. The truncated ditetragonal prism (TDP) particles with all 12 sides being the {310} facets were identified capable to activate the 4e-ORR over the full potential range that has not been seen. Electrochemical studies in 1990s9 suggested that specific hydroxyl (OH−) adsorption at the four-fold hollow sites might be associated with the 4e-ORR on Au{100} in alkaline solutions because the 4e-ORR potential region coincides with a small rise of current in the voltammetry curve below the onset potential for Au surface oxidation or OH− adsorption. In 2006, a study of the ORR on polycrystalline Au using in situ surfaceenhanced Raman spectroscopy (SERS)21 found a peak at 820 cm−1 that was tentatively assigned to the AuO-H bending; its intensity stepped down near the potential where the 4e pathway switches to 2e on {100}. However, it remains unclear how adsorbed OH can promote the 4e-ORR.22,23 A possible explanation comes if we consider water, instead of OH, as the adsorbed species because H2O is the source of protons for the ORR in alkaline solution, so the co-adsorbed water may promote the 4e-ORR via its involvement as an active proton donor. In recent years, such an idea has been attracting attention to the effects of water adsorption on catalytic reactions.24,25 Our experimental observations of the 4e-ORR process can be understood in the light of the DFT-calculated free energy diagrams presented here for the ORR on the three Au surfaces without and with co-adsorbed water, which elucidate how the 4e-ORR is activated on Au {100} and {310} by surface proton transfers. Thus, the study elucidates the important role of the water adsorption.

Figure 1. Self-assembly of gold nanocrystals. (a1−c1) Typical largearea scanning electron microscopy (SEM) images of Au nanocrystals. (a2−c2) Locally magnified SEM images of Au nanocrystals. (a3−c3) Schematic representation of nanocrystals’ polyhedral shapes. (a) Au octahedral nanocrystals, (b) Au cubic nanocrystals, and (c) Au truncated ditetragonal prism nanocrystals.

Au surfaces does not require harsh treatments often applied.26,27 We ensured sufficient protection of the {111} facets during the growth of Au octahedra by using a high concentration of the surfactant (CPC) and a low concentration of the reducing agent (ascorbic acid).28,29 For preparing the Au cubes, we added KBr and increased the reaction temperature to 32 °C. In synergy with CPC, the cationic surfactant with Br− counterions stabilizes the Au {100} facets, in a way similar to that of the surfactant cetyltrimethylammonium bromide (CTAB).30 The Au TDPs were obtained by adding HCl and a small amount of Ag+. A monolayer of Ag may form on Au, serving as a short-lived facet-blocking adsorbate during a crystal growth.27 Since the amount of reducing agent is less than that required for the complete reduction of all the metal ions present, the Ag monolayer is removed via galvanic replacement by the excess Au ions at the end of synthesis. Energy-dispersive X-ray spectroscopy (EDX) analysis confirmed that no trace of Ag was left on the final TDP surface (Figure S2). While previously we prepared the Au TDPs at 25 °C,27 we found in this study that raising the temperature to 50 °C can speed up the synthesis without compromising quality. The yields of all three types of Au nanocrystals are higher than 95%, and the standard deviations in particle size distribution are below 4%. (See Materials and Methods and Table S1 in the Supporting Information for details on the particle synthesis and product quality.) The high uniformity of particle sizes and shapes assures that these nanocrystals assemble readily into ordered, densely packed structures. The scanning electron microscopy (SEM) images, shown in Figure 1, reveal that the packing types are primarily determined by the particle shapes. For example, the Au octahedra (∼42 nm edge length, Figure 1a2) with eight {111} faces (Figure 1a3) assemble only via incomplete face-toface contact with their neighbors: the bottom layer forms a

■

RESULTS AND DISCUSSION Synthesis and Characterization. Our solution-phase synthesis methods (Supporting Information) create the monofaceted nano-objects (octahedra, cubes, TDPs) (Figures 1 and S1) with a narrow size distribution (Table S1) and with a high-fidelity terminating surface. The typical synthesis procedures yield single-crystalline nanoparticles with clean surfaces after a simple washing, so enabling us to precisely probe the facet-dependent ORR behaviors. In shape-controlling synthesis of Au nanocrystals, we used the multistep procedure to prepare various polyhedral nanocrystals with edge lengths above 40 nm. The Au octahedra were formed in an aqueous solution of cetylpyridinium chloride (CPC), via reducing the AuCl4− ions by ascorbic acid at 25 °C. The surfactant CPC is weakly adsorbed, and accordingly, removing it to obtain clean 7311

DOI: 10.1021/jacs.7b01735 J. Am. Chem. Soc. 2017, 139, 7310−7317

Article

Journal of the American Chemical Society

Figure 2. Structural characterization of gold nanocrystals. (a1−c1) Transmission electron microscopy (TEM) images of Au nanocrystals. (a2−c2) TEM images of single Au nanoparticles. (a3−c3) Selected area electron diffraction (SAED) of Au nanocrystals. (c4) Scanning transmission electron microscopy (STEM) image of a single Au TDP nanoparticle. (c5) High-resolution high-angle annular dark field (HAADF)-STEM image of the TDP surface, showing its sub-facets of {110} and {100} planes. (c6) Tilt-view atomic model of the {310} facet. The red and yellow spheres respectively represent surface and bulk atoms. (a) Au octahedra, (b) Au cubes, and (c) Au TDPs.

simple hexagonal arrangement biased by a flat substrate surface (Figure 1a2). Then, a Minkowski-like structure forms after a few staggered layers, which increases the density of three-dimensional packing31 as depicted in Figure 1a1. Au cubes with six {100} faces (Figure 1b3) have an average edge length of ∼48 nm (Figure 1b2) and form closely packed layers with in-plane square symmetry (Figure 1b1,b2). The Au TDPs enclosed by 12 {310} facets (Figure 1c3) with the long edge ∼45 nm act like spheres, and thus, assemble into a quasi-hexagonal packing structure within the layer (Figure 1c1,c2). We further characterized the atomic lattice structures of these monofaceted nanocrystals using transmission electron microscopy (TEM) and selected area electron diffraction (SAED). In Figure 2, TEM images of multiple (a1−c1) and single (a2−c2) particles of Au octahedra (a), cubes (b), and TDPs (c) are shown with the SAED patterns (a3−c3) obtained from the corresponding single particles. The sharp diffraction dots in the

characteristic SAED patterns confirm the nanocrystals’ single crystallinity. The Au TDP nanocrystals are bounded by 12 high-indexed {310} facets that can be considered as a stepped surface with the vector sum of one {110} facet and two {100} facets (Figure 2c6). Typically, when viewed along the [001] direction, the projection profile of TDP appears di-tetragonal, while the correspondingly measured inner angles are close to those calculated from the ideal TDP model with {310} side facets. Orienting the Au TDPs along the axis of the [001] zone can drive their {310} facets parallel to the electron beam, so allowing us to directly observe the atomic arrangement by highresolution scanning transmission electron microscopy (STEM) with a high-angle annular dark field (HAADF) detector. As displayed in Figure 2c5, the sub-facets of the {110} and {100} planes clearly are present in the {310} facet oriented in the [100] direction, matching well the atomic model (Figure 2c6). 7312

DOI: 10.1021/jacs.7b01735 J. Am. Chem. Soc. 2017, 139, 7310−7317

Article

Journal of the American Chemical Society

Figure 3. Oxygen reduction reaction on shaped Au nanocrystals in alkaline solutions. Color scheme: Au octahedra, dark gray; cubes, blue; and TDPs, red. (a) Ring and (b) disk currents measured in O2-saturated 0.1 M NaOH at 50 mV·s−1 and 1600 rpm. Au nanocrystal loading and Au surface-toelectrode area ratio: 100 μg·cm−2 and 12.3 for octahedra, 130 μg·cm−2 and 8.4 for cubes, and 84 μg·cm−2 and 5.8 for TDPs. (c) Average number of transferred electrons (n) during the ORR as a function of potential calculated from the fraction of disk-to-ring flux in (a) and (b). The inset shows the 2e and 4e pathways for the ORR in alkaline electrolyte, where the color scheme of oxygen atoms illustrates the two oxygen sources. (d) Falsecolor SEM images of Au nanocrystals used for the ORR measurements: octahedra, cubes, and TDPs. Scale bar: 50 nm. (e) Cyclic voltammograms in deaerated 0.1 M NaOH at 50 mV·s−1. (f) Voltammetry curves for Tl UPD in deaerated 5 mM TlNO3 + 0.1 M NaOH at 50 mV·s−1.

Electrochemical Analysis and ORR Performance. We carried out electrochemical measurements on the three shaped Au nanocrystals in alkaline electrolyte. The results are shown in Figure 3 using the same color scheme for the curves as that for the SEM images (Figure 3d) of Au octahedra (dark gray), cubes (blue), and TDPs (red). After loading the Au nanocrystals on electrodes, we washed them in ethanol and water to remove the surfactant residuals. Prior to the evaluation of the ORR performance, we checked the cleanness and crystallinity of these monofaceted nanocrystals using cyclic voltammograms and thallium underpotential deposition (Tl UPD) by comparing to the known features on single-crystal surfaces. The voltammogram of Au cubes in deaerated 0.1 M NaOH (blue curve in Figure 3e) exhibits features similar to that on the Au(100) single-crystal surface.32 While there is no reported voltammogram for a Au(310) single-crystal surface to compare, the absence of any anodic current peak below 1.2 V on Au TDPs confirms that the surface was free of Ag and organic species. UPD of Tl exhibits characteristic current peaks on well-oriented clean Au(111)33 and Au(100),34 which are associated with the phase transitions between long-range ordered adlayer structures identified by surface X-ray diffraction. Figure 3f shows the Tl UPD current peaks on all three shaped nanocrystals, albeit with peak broadening due to

influences from the edges and corners of nanoparticle facets. These current peaks indicate the formation of ordered Tl and Tl−OH co-adsorbed phases on the monofacet surfaces. Above 0.9 V, the rising anodic currents signal the vanishing of ordered co-adsorbed Tl-OH adlayer33 Since clean and atomic flat surfaces are required for ordered adlayers, these Tl UPD current peaks and those obtained with narrower potential windows (Figure S4) further verified the high crystallinity and cleanness of the nanocrystals samples. We probed the ORR activities of Au nanocrystals using the techniques of rotating disk electrode (RDE) and rotating ring disk electrode (RRDE) in O2-saturated 0.1 M NaOH. The ORR currents on the disk electrode (Figure 3b) for the octahedra and cubes, consonant with other groups’ results,35,36 emulate those on the Au(111) and Au(100) single-crystal surfaces,9 respectively. Same as on the extended surfaces of Au single crystals, the Au octahedron is the least active among the three shaped nanocrystals for the ORR as evidenced by its onset potential (0.88 V) and limiting current (−3 mA·cm−2) being the lowest. Since the current limited by the mass transport for the 2e-ORR is ca. −3 mA·cm−2, half of that for the 4e-ORR (see Supporting Information for details on diffusionlimiting current density), the Au octahedron is not active for the 4e-ORR. In contrast, Au cubes and TDPs exhibit currents 7313

DOI: 10.1021/jacs.7b01735 J. Am. Chem. Soc. 2017, 139, 7310−7317

Article

Journal of the American Chemical Society

Figure 4. DFT study of the ORR on Au surfaces in alkaline solutions. (a−d) Reaction free energy diagrams for the 2e- (red) and 4e-ORR (bluegray) at the hydrogen reversible potential on three Au surfaces without (a) and with (b−d) adsorbed water. The lower energy pathways are shown by solid lines as compared to those shown by dashed lines. Protonation from solution is indicated by +H. (b1,b2) Optimized surface structures before (b1) and after (b2) adding a hydrogen (marked by H) to the surface that binds with an adsorbed O2 (marked by O2) on Au {100} with co-adsorbed water molecules. A surface proton transfer (marked by H1) from an adsorbed water to O1 coincides with the dissociation of O2(ads). (d1,d2) Optimized surface structures before (d1) and after (d2) adding one hydrogen (marked by H) per O2H(ads) on water-adsorbed Au {310}. The O−O bond breaks are promoted by the surface proton transfers from co-adsorbed water molecules (marked by H1, H2, and H3). Blue dashed lines show the hydrogen-bonding network that promotes O2−water co-adsorptions and surface proton transfers.

cubes at potential below 0.6 V, which means a higher 4e-ORR selectivity in this potential region than that on {100} facet. A precaution about determining the electron-transfer number for the nanocatalysts using the RRDE method is worth mentioning. RRDE studies of the ORR on carbon-supported Pt nanoparticles detected the 2e-ORR product with low Pt loadings indicating that the average electron-transfer number was lowered from that on the extended Pt surface (complete 4e) on small and highly dispersed particles.37 Also, in a thinlayer flow cell, the H2O2 yield increased with increasing flow rate or decreasing coverage of Pt.38 Because single-crystal studies support complete 4e-ORR on Pt, the 2e-ORR product detected under high ratio of oxygen flow to Pt surface area indicates that insufficient loading of nanoparticles can cause a lower electron-transfer number than that measured on extended metal surfaces. In our measurements, the ring currents observed for the Au cubes and TDPs are unlikely due to nanosize effect or insufficient loading because (1) the Au

exceeding the 2e-ORR limiting current and, thus, are active for the 4e-ORR. To evaluate the selectivity toward the 4e- or 2eORR on Au cubes and TDPs, we determined the average number of electrons transferred per O2 molecule using the RRDE method. As shown in the inset to Figure 3c, the 2e-ORR pathway in alkaline electrolyte leads to the production of O2H−. Using an additional ring electrode, the RRDE method detects the oxidation current of O2H− concomitantly produced via the 2e-ORR on the disk electrode. For Au cubes, the ring current is zero above 0.7 V (blue curve in Figure 3a); with decreasing potentials (or increasing overpotentials), the rise of the ring current of O2H− oxidation coincides with a decrease in the disk current (Figure 3b) from the maximum around 0.6 V to the plateau below 0.4 V. The same behavior was observed with Au(100) single crystals.9 Interestingly, on Au TDPs, the ring current is smaller and the disk current is larger than those on 7314

DOI: 10.1021/jacs.7b01735 J. Am. Chem. Soc. 2017, 139, 7310−7317

Article

Journal of the American Chemical Society

with a proton transfers from adsorbed water to O1. The synergetic effect is remarkable that the 4e-ORR is activated on Au {100} at the potentials as high as on Pt. Unfortunately, a switch from 4e- to 2e-ORR occurs around 0.6 V, which we attributed to the weakened water adsorption at low potentials. Details on the experimental support for water adsorption on Au {100} facets are given in the following paragraph. Figure S5 shows that the 0.6−1 V potential window for the 4e-ORR corresponds to a small increase of currents in the voltammetry curve measured in Ar-saturated solution. This voltammetry feature9,21,49 and an 820 cm−1 peak observed in SERS9,21 were ascribed to the site-specific OH adsorption at early dates before a vacuum vibration spectroscopic study of water adsorption on Au surfaces in 2010.50 The water molecules adsorbed from gas phase on Au surfaces exhibit a hydrogen bending mode at 730 cm−1 (a frustrated rotation around an axis perpendicular to the C2-axis of the H2O molecule), which is the most intense on Au(100) and weakens on Au(11,1,1) and Au(5,1,1) as the {100} terrace width decreases on the high-index surfaces.50 These results indicate that the SER peak at 820 cm−1 observed in the 4e-ORR potential region likely is associated with the adsorbed H2O, instead of adsorbed OH, on the {100} facets of the polycrystalline Au surfaces.21 For the close-packed {111} facets, previous studies of water adsorption found an ice-like bilayer structure,51 we adapted this structural model for calculating the effect of adsorbed water on the ORR pathways. As shown by the free energy diagram in Figure 4c, the O2 adsorption remains endothermic and the 2e pathway (red, solid line) is still favorable. These results are consistent with the experimental observation of only the 2e pathway over the full potential range and the lowest onset potential on Au {111}. The Au {310} surface with low-coordinated atoms at the edges of the {100} sub-facet (see Figure S6) can co-adsorb water at a wider potential region than that on Au {100}. Thus, we may explain the observed whole-potential-range 4e-ORR on Au {310} by finding a configuration, in which the co-adsorption of O2 and H2O promotes the O−O bond scission through enabling a surface proton transfer. Figure 4d shows that while the first protonation from solution yields stable O2H(ads), surface proton transfers do occur from co-adsorbed water molecules (marked by H1, H2, and H3 in Figure 4d1,4d2) upon second protonation from solution on the O2H(ads) (marked as H onto O1 and O3). Thus, O2H(ads) dissociation (the 4e pathway) is energetically more favorable than O2H(ads) desorption (the 2e pathway) on {310}. Compared to the case on the water-adsorbed Au {100} in details, the experimentally observed lower onset overpotential (higher onset potential vs RHE) and higher 4e-ORR selectivity on cubes than on TDPs can be explained. On cubes with {100} facets, the surface-proton-transfer-promoted O−O bond scission occurs upon O2(ads) getting a proton from solution, while on the TDP particles with {310} facets, the O−O bond scission occurs with the O2H(ads) getting another proton from solution. O2H(ads) on Au {310} is formed by protonating O2(ads) with the H from water in solution phase, which is a process driven by the overpotential. Therefore, the higher overpotential required for the O2H(ads) in comparison with the O2(ads), results in the higher ORR onset overpotential on {310} than on {100}. The high selectivity of the 4e-ORR on Au {100} facet at high potentials is explained by the instability of O2H(ads) on water-co-adsorbed {100} surface. As shown in

nanocrystals are 1 order of magnitude bigger than Pt particles and there is no carbon support, and (2) there is no ring current on cubes in the potential window above 0.7 V as on a Au(100) single crystal, suggesting a sufficient loading of Au cubes. While there are no reported ORR studies on Au(310) single crystals to compare, we used a comparable loading for the TDPs that have particle size similar to that of the cubes. Figure 3c shows the average number of transferred electrons calculated from the RRDE data. For the TDP, the value is about 3.8 over the whole potential region. On the cubes, complete 4e-ORR occurs in a narrow potential window with lowered 4e-ORR selectivity at low potentials. Several high-index Au surfaces with {100} terraces are known to exhibit behavior similar to that of the Au(100) single crystal,9,21 but a whole potential region of dominant 4e-ORR on Au has not been reported. The unique behavior of the {310} facet revealed on Au TDPs provides an additional check point for testing any mechanistic explanation of the facet- and potential-dependent 4e-ORR on Au, which we discuss below. DFT-Calculated Free Energy Diagrams and Surface Proton Transfer from Co-adsorbed Water. One major function of catalysts is to promote bond breaking in reactants.39 For the ORR, the selectivity toward the 4e or 2e pathway (Figure 3c inset) is determined by the catalyst’s propensity to break the dioxygen bond.40 When O2 dissociates, the reaction proceeds via a 4-electron-per-O2 reduction pathway with OH− as the final product in alkaline electrolyte; otherwise, the reaction ends by 2e reduction to O2H−. While O2 dissociation is facile on many transition metals, the high barriers for dissociating O2 on Au surfaces are known41−43 because O2 adsorbs on Au significantly weaker than on other transition metals.5−7 To understand why the 4e pathway is active on certain Au facets, we carried out DFT calculations to construct the free energy diagrams for the ORR on {100}, {111}, and {310} Au surfaces at the hydrogen reversible potential in alkaline solutions, with the focus on the cleavage of the O−O bond. In alkaline media, water is the source of proton for the ORR, and can adsorb on Au surfaces without dissociation in the ORR potential region, differing from the OH adsorption on more reactive Pt surfaces. The co-adsorbed water can act as the proton donor for the ORR44,45 and other reactions catalyzed by Au.46−48 Since the water adsorption generally decreases with lowering potential and strengthens on low-coordinated surface sites, the effect of co-adsorbed water on the ORR kinetics is dependent on potential and facet. In our DFT study, we considered the ORR without and with co-adsorbed water on the Au surfaces. Figure 4a,b illustrates the effect of co-adsorbed water on the reaction free energies for the ORR intermediates on Au {100}. Without adsorbed water, O2 adsorption on Au {100} is endothermic, and it becomes exothermic with co-adsorbed water because a network of hydrogen bonding (dashed blue lines in Figure 4b1) lowers the free energy of adsorbed oxygen, O2(ads). Upon gaining an H atom from solution, the O2H(ads) is stable without co-adsorbed water and the 2e pathway is energetically favorable (Figure 4a). In contrast, a protonation from solution leads to an O−O bond cleavage in the presence of co-adsorbed water (Figure 4b), which is assisted by a surface proton transfer from an adsorbed water (marked by H1 in Figure 4b1,b2). The configuration is such that the four hydrogen bonds act with a force to pull apart the O1−O2 (Figure 4b1); once the O2 is protonated, the O1−O2 bond break coincides 7315

DOI: 10.1021/jacs.7b01735 J. Am. Chem. Soc. 2017, 139, 7310−7317

Article

Journal of the American Chemical Society ORCID

Figure 4b, the two stable adsorbates after protonation of O2(ads) on {100} are 2OH(ads) and 2O(ads), i.e., the O−O bond scission is 100%. In contrast, as shown in Figure 4d, O2H(ads) intermediate is stable on {310}, which opens the possibility for the ORR to end with 2e transfer upon O2H(ads) desorption, albeit this is the less favorable pathway. Therefore, the dominant 4e-ORR (3.8e) was observed on the TDPs with {310} facets. In summary, the DFT calculated free energy diagrams well explained the experimental observation of the mechanistic variation for the ORR on the three Au surfaces. Based on the atomistic structures for the reaction intermediates, we elucidated the key role of co-adsorbed H2O molecules for the facet- and potential-dependent 4e-ORR on Au in alkaline solutions. When the reactant H2O is co-adsorbed with O2 in a geometrically favorable configuration that enables a surface proton transfer, the desirable 4e-ORR pathway is activated. Besides promoting the O−O bond scission, surface proton transfer opens a lower-barrier path for the ORR than that relying on getting proton from water in liquid phase. Thus, the onset potential is the highest for {100} and the lowest for {111} because their lattice geometries are, respectively, the most and least favorable for surface proton transfer. These new insights provide guidance for searching favorable lattice structures for modulating the selectivity and activity of electrocatalysts.

Fang Lu: 0000-0003-3765-7615 Dong Su: 0000-0002-1921-6683 Ping Liu: 0000-0001-8363-070X Author Contributions §

F.L. and Y.Z. contributed equally.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The research was carried out at the Center for Functional Nanomaterials and the Chemistry Division of Brookhaven National Laboratory, supported by the U.S. Department of Energy, Office of Basic Energy Sciences, under Contract No. DE-SC0012704. We acknowledge the computational resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.

■

■

CONCLUSIONS We generated several types of Au shaped nanocrystals, including octahedra, cubes and TDPs, which are correspondingly enclosed by atomically well-defined nanocrystal surfaces, {111}, {100}, and {310} respectively. Using these high-quality single-crystalline Au nanocrystals with differently expressed atomic arrangements of their surfaces, we observed the 2e-ORR on Au octahedra and the 4e-ORR in a narrow potential region on Au cubes in alkaline solutions, similar to those on corresponding single-crystal surfaces. Intriguingly, Au {310} surface catalyzes the dominant 4e-ORR over the full potential range never seen before. Combined with DFT calculations, we illustrated that the lattice structures supporting surface proton transfers from co-adsorbed reactant H2O to the dioxygen intermediates promote the O−O bond break and, thus, activate the 4e-ORR on Au. This new insight can help mechanism studies of other facet-sensitive reactions, e.g., ammonia oxidation reaction on Pt-based catalysts, and development of facet-specific nanocatalysts for various applications.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b01735. Experimental details, method for DFT calculations, additional characterizations, calculation for the ORR diffusion-limiting current density, and lattice structures of studied Au surfaces, including Figures S1−S6 and Tables S1 and S2 (PDF)

■

REFERENCES

(1) Yamada, Y.; Tsung, C. K.; Huang, W.; Huo, Z. Y.; Habas, S. E.; Soejima, T.; Aliaga, C. E.; Somorjai, G. A.; Yang, P. D. Nat. Chem. 2011, 3, 372−376. (2) Ruditskiy, A.; Choi, S.-I.; Peng, H.-C.; Xia, Y. MRS Bull. 2014, 39, 727−737. (3) Spendelow, J. S.; Wieckowski, A. Phys. Chem. Chem. Phys. 2007, 9, 2654−2675. (4) Katsounaros, I.; Schneider, W. B.; Meier, J. C.; Benedikt, U.; Biedermann, P. U.; Cuesta, A.; Auer, A. A.; Mayrhofer, K. J. J. Phys. Chem. Chem. Phys. 2013, 15, 8058−8068. (5) Haruta, M. Catal. Today 1997, 36, 153−166. (6) Hashmi, A. S. K.; Hutchings, G. J. Angew. Chem., Int. Ed. 2006, 45, 7896−7936. (7) Turner, M.; Golovko, V. B.; Vaughan, O. P.; Abdulkin, P.; Berenguer-Murcia, A.; Tikhov, M. S.; Johnson, B. F.; Lambert, R. M. Nature 2008, 454, 981−983. (8) Adzic, R. R.; Markovic, N. M.; Vesovic, V. B. J. Electroanal. Chem. Interfacial Electrochem. 1984, 165, 105−120. (9) Strbac, S.; Adzic, R. R. J. Electroanal. Chem. 1996, 403, 169−181. (10) Markovic, N. M.; Adzic, R. R.; Vesovic, V. B. J. Electroanal. Chem. Interfacial Electrochem. 1984, 165, 121−133. (11) Krylova, G.; Giovanetti, L. J.; Requejo, F. G.; Dimitrijevic, N. M.; Prakapenka, A.; Shevchenko, E. V. J. Am. Chem. Soc. 2012, 134, 4384−4392. (12) Xia, Y.; Xia, X.; Wang, Y.; Xie, S. MRS Bull. 2013, 38, 335−344. (13) Niu, W.; Zhang, W.; Firdoz, S.; Lu, X. J. Am. Chem. Soc. 2014, 136, 3010−3012. (14) Tao, A.; Sinsermsuksakul, P.; Yang, P. Nat. Nanotechnol. 2007, 2, 435−440. (15) Solla-Gullon, J.; Vidal-Iglesias, F. J.; Feliu, J. M. Annu. Rep. Prog. Chem., Sect. C: Phys. Chem. 2011, 107, 263−297. (16) Shao, M.; Odell, J.; Humbert, M.; Yu, T.; Xia, Y. J. Phys. Chem. C 2013, 117, 4172−4180. (17) Zhang, Y.; Lu, F.; Yager, K. G.; van der Lelie, D.; Gang, O. Nat. Nanotechnol. 2013, 8, 865−872. (18) Crespo-Quesada, M.; Andanson, J.-M.; Yarulin, A.; Lim, B.; Xia, Y.; Kiwi-Minsker, L. Langmuir 2011, 27, 7909−7916. (19) Kleijn, S. E. F.; Lai, S. C. S.; Koper, M. T. M.; Unwin, P. R. Angew. Chem., Int. Ed. 2014, 53, 3558−3586. (20) Zhang, G.-R.; Xu, B.-Q. Nanoscale 2010, 2, 2798−2804. (21) Kim, J.; Gewirth, A. A. J. Phys. Chem. B 2006, 110, 2565−2571. (22) Koper, M. T. M. Chem. Sci. 2013, 4, 2710−2723. (23) Mei, D.; He, Z. D.; Zheng, Y. L.; Jiang, D. C.; Chen, Y. X. Phys. Chem. Chem. Phys. 2014, 16, 13762−13773.

AUTHOR INFORMATION

Corresponding Authors

*[email protected] *[email protected] 7316

DOI: 10.1021/jacs.7b01735 J. Am. Chem. Soc. 2017, 139, 7310−7317

Article

Journal of the American Chemical Society (24) Carrasco, J.; Hodgson, A.; Michaelides, A. Nat. Mater. 2012, 11, 667−674. (25) Staszak-Jirkovsky, J.; Subbaraman, R.; Strmcnik, D.; Harrison, K. L.; Diesendruck, C. E.; Assary, R.; Frank, O.; Kobr, L.; Wiberg, G. K. H.; Genorio, B.; Connell, J. G.; Lopes, P. P.; Stamenkovic, V. R.; Curtiss, L.; Moore, J. S.; Zavadil, K. R.; Markovic, N. M. ACS Catal. 2015, 5, 6600−6607. (26) Lu, F.; Tian, Y.; Liu, M.; Su, D.; Zhang, H.; Govorov, A. O.; Gang, O. Nano Lett. 2013, 13, 3145−3151. (27) Lu, F.; Zhang, Y.; Zhang, L.; Zhang, Y.; Wang, J. X.; Adzic, R. R.; Stach, E. A.; Gang, O. J. Am. Chem. Soc. 2011, 133, 18074−7. (28) Niu, W. X.; Zheng, S. L.; Wang, D. W.; Liu, X. Q.; Li, H. J.; Han, S. A.; Chen, J.; Tang, Z. Y.; Xu, G. B. J. Am. Chem. Soc. 2009, 131, 697−703. (29) Tao, A. R.; Habas, S.; Yang, P. Small 2008, 4, 310−325. (30) Johnson, C. J.; Dujardin, E.; Davis, S. A.; Murphy, C. J.; Mann, S. J. Mater. Chem. 2002, 12, 1765−1770. (31) Henzie, J.; Grünwald, M.; Widmer-Cooper, A.; Geissler, P. L.; Yang, P. Nat. Mater. 2012, 11, 131−137. (32) Hamelin, A.; Sottomayor, M. J.; Silva, F.; Chang, S.-C.; Weaver, M. J. J. Electroanal. Chem. Interfacial Electrochem. 1990, 295, 291−300. (33) Wang, J. X.; Adzic, R. R.; Ocko, B. M. J. Phys. Chem. 1994, 98, 7182−7190. (34) Adzic, R. R.; Tripkovic, A. V.; Markovic, N. M. J. Electroanal. Chem. Interfacial Electrochem. 1983, 150, 79−88. (35) Hernández, J.; Solla-Gullón, J.; Herrero, E.; Aldaz, A.; Feliu, J. M. J. Phys. Chem. C 2007, 111, 14078−14083. (36) Min, X.; Chen, Y.; Kanan, M. Phys. Chem. Chem. Phys. 2014, 16, 13601−13604. (37) Inaba, M.; Yamada, H.; Tokunaga, J.; Tasaka, A. Electrochem. Solid-State Lett. 2004, 7, A474−A476. (38) Seidel, Y. E.; Schneider, A.; Jusys, Z.; Wickman, B.; Kasemo, B.; Behm, R. J. Faraday Discuss. 2009, 140, 167−184. (39) Somorjai, G. A. Chem. Soc. Rev. 1984, 13, 321−349. (40) Yeager, E. J. Mol. Catal. 1986, 38, 5−25. (41) Nørskov, J. K.; Rossmeisl, J.; Logadottir, A.; Lindqvist, L.; Kitchin, J. R.; Bligaard, T.; Jónsson, H. J. Phys. Chem. B 2004, 108, 17886−17892. (42) Fernandez, J. L.; Walsh, D. A.; Bard, A. J. J. Am. Chem. Soc. 2005, 127, 357−365. (43) Xu, Y.; Mavrikakis, M. J. Phys. Chem. B 2003, 107, 9298−9307. (44) Vassilev, P.; Koper, M. T. M. J. Phys. Chem. C 2007, 111, 2607− 2613. (45) Ramaswamy, N.; Mukerjee, S. J. Phys. Chem. C 2011, 115, 18015−18026. (46) Saavedra, J.; Doan, H. A.; Pursell, C. J.; Grabow, L. C.; Chandler, B. D. Science 2014, 345, 1599−1602. (47) Fujitani, T.; Nakamura, I.; Haruta, M. Catal. Lett. 2014, 144, 1475−1486. (48) Mullen, G. M.; Gong, J.; Yan, T.; Pan, M.; Mullins, C. B. Top. Catal. 2013, 56, 1499−1511. (49) Adzic, R. R., Reaction Kinetics and Mechanism on Metal SingleCrystal Electrode Surfaces. In Modern Aspects of Electrochemistry No. 21, Bockris, J. O. M., Conway, B. E., White, R. E., Eds.; Springer US: Boston, MA, 1990; pp 163−236. (50) Ibach, H. Surf. Sci. 2010, 604, 377−385. (51) Stacchiola, D.; Park, J.; Liu, P.; Ma, S.; Yang, F.; Starr, D.; Muller, E.; Sutter, P.; Hrbek, J. J. Phys. Chem. C 2009, 113, 15102− 15105.

7317

DOI: 10.1021/jacs.7b01735 J. Am. Chem. Soc. 2017, 139, 7310−7317