Influence of Alkaline Medium in Molten Salt Reaction on the Local Structures of Iridium Oxide Catalysts for Oxygen Evolution Reaction.

A key challenge for the design of efficient photoelectrochemical devices that employ nanoscaled interfacial designs for solar energy-to-fuels conversion lies in resolving atomic structures and dynamics at the active interfaces, and relating these to the complex cascade of events which includes excited-state charge separation, charge-accumulation, and multi-step energy conversion and catalysis. Our research programs have been developing in-situ, time-resolved X-ray techniques aimed at the resolution of atomic scale structures that underlie interfacial charge transfer and photo-driven water-splitting catalysis under conditions relevant to photoelectrochemical function for both interfacial thin-films and molecular photosensitizers and catalysts bound to semiconductor surfaces. This presentation will discuss our development of X-ray techniques aimed at the resolution of atomic scale structures that underlie interfacial charge transfer and photo-driven water-splitting catalysis under conditions relevant to photoelectrochemical function. We have developed high surface area TiO2 semiconductor and ITO and IZO conducting oxide porous assemblies that duplicate oxide-supported dye and catalyst architectures widely used in dye sensitized solar cells (DSSC) and photoelectrochemial (PEC) electrodes, but have been tailored to allow for interfacial X-ray structure characterization. Thin-Film Metal-Oxide Water Oxidation Catalysts. Non-noble-metal, thin film oxides are widely investigated as promising catalysts for oxygen evolution reactions (OER). Amorphous cobalt oxide films electrochemically formed in the presence of borate (CoBi) and phosphate (CoPi) share a common cobaltate domain building block, but differ significantly in OER performance that derives from different electron-proton charge transport properties. Here, we use a combination of L-edge synchrotron X-ray absorption (XAS), resonant X-ray emission (RXES), resonant inelastic X-ray scattering (RIXS), resonant Raman (RR) scattering, and high-energy X-ray pair distribution function (PDF) analyses that identify electronic and structural factors correlated to the charge transport differences for CoPi and CoBi. The analyses show that CoBi is composed primarily of cobalt in octahedral coordination, while CoPi contains approximately 17% tetrahedral Co(II), with the remainder in octahedral coordination. Oxygen-mediated 4p-3d hybridization through Co-O-Co bonding was detected by RXES and the inter-site dd excitation was observed by RIXS in CoBi, but not in CoPi. RR shows that CoBi resembles a disordered layered LiCoO2-like structure while CoPi is amorphous. Distinct domain models in the nanometer range for CoBi and CoPi have been proposed on the basis of the PDF analysis coupled to XAS data. The observed differences provide information on electronic and structural factors that enhance OEC performance.The combined electronic and structural analyses demonstrate that hole transfer to catalytic sites in amorphous cobalt oxide thin films is the rate-limiting step for electrochemical water-splitting rather than the multi-step catalytic events themselves. The macroscopic catalytic properties of the thin films were found to be correlated to the electronic structures measured at the atomic scale for the metal-oxo cluster domains. These results show the interplay between intrinsic catalytic activity and charge transport properties of semiconductor thin-film catalysts. In-situ Structure-Function Analysis of Molecular Water Oxidation Catalysts. Building from the microporous electrodes for in-situ X-ray analyses of amorphous oxide thin-film OECs, we have extended this approach by the development of nano-porous electrode architectures which enable the use of combined PDF and X-ray spectroscopy analyses of coordination structures for molecular photosensitizers and catalysts in solution and when bound to semiconductor oxide surfaces. PDF measurements require PEC oxide supports with sufficiently low background scattering to permit detection of the surface bound complexes. We have found that atomic layer deposition of semiconductor oxides on anodic aluminium oxide templates, AAO, provides a suitable layered, high-surface area architecture. PDF patterns measured for the N3 dye, cis-bis(isothiocyanato) bis(2,2’-bipyridyl-4,4’-dicarboxylato) ruthenium(II), bound to an amorphous TiO2 surface reveal details on outer sphere ligand structures. Each of the PDF peaks can be assigned to the atom pair distances calculated from crystal and DFT structures. Measurements on ruthenium photosensitizer complexes bound to the TiO2 surface resolve distortions in the surface-coordinating ligand structures. On-going work is investigating the correlation between ligand structural distortion and the metal-to-ligand charge-transfer (MLCT) states with interfacial charge injection function.

With the increase in fossil energy consumption and the attendant pronounced climate changes, the call for sustainable energy forms has become urgent and has attracted more and more attention.1 Hydrogen (H2), generated by the electricity-derived water electrolysis, is now considered the promising energy carrier for conversion between electricity and chemical energy.2 The most developed water electrolysis technology up to date is the proton-exchange membrane (PEM) water electrolysis, where commercial Pt/C works as the cathodic catalyst for hydrogen evolution reaction (HER) and noble metal-based materials serve as the anodic catalyst for oxygen evolution reaction (OER).3 Considering the sluggish four-electron reaction kinetics of OER and the harsh acidic environment of PEM water electrolysis, the selectivity of high-efficient OER catalysts is narrowed down to Ir-based catalysts because of their superior activity, long-term stability and corrosion resistance against acidic environment.4 However, the scarcity of Ir on earth has impeded the wide application of PEM water electrolysis and has driven the research of cost-efficient OER catalysts. In this study, we thoroughly investigated several amorphous IrOx catalysts with different degree of crystallinity (SA103, SA58, SA14.6 and SA58) and revealed the relation between their structures and OER activities. Besides, BET surface areas, X-ray diffraction (XRD), Transmission electron microscopy (TEM), scanning TEM (STEM), PDF, and XAS were measured to thoroughly understand the catalytic properties.XRD patterns in Figure 1(a) showed that compared with the well-crystallized IrO2, SA58 sample showed slightly broad peak, while the diffraction peaks were much broad of other samples, indicating their amorphous structures. The amorphous structure of IrOx was characterized through atomic pair distribution function (PDF) analyses as shown in Figure 1(b). Different from the typical tetragonal symmetry of crystal IrO2, the combination of orthorhombic symmetry and monoclinic symmetry was found in the amorphous IrOx. Further, OER activities of all samples were investigated in 0.1 M HClO4 and the current density was calculated against the BET surface areas to examine the intrinsic catalytic activity (Figure 1(c)). The crystallized SA58 sample showed the lowest activity, while SA3.5 sample outperformed all other amorphous oxides. From PDF, it was found that the highest OER active SA3.5 sample exhibited the highest monoclinic phase ratio content. The catalytic structure properties were further evidenced by the XAS results. EXAFS results showed different electronic environment of monoclinic phase samples and further fitting results were in good agreement with the PDF analyses. Moreover, O K-edge results showed the formation of µ2-O (O-O) bond at around 528.7 eV, which was previously observed by Nong, H. N. et. al. through operando O-K edge measurements during OER process.5 It was concluded that that the low symmetry of monoclinic phase in amorphous samples resulted in the structure defection, and thus lead to large amount of active electrophilic OI - species, which boosting the OER activity. This finding provided fundamental understanding of the amorphous iridium oxides and could promisingly shed light on the future OER catalyst design. Acknowledgements This work is based on results obtained from a project (JPNP14021) commissioned by the New Energy and Industrial Technology Development Organization (NEDO) of Japan. References Götz, M.; Lefebvre, J.; Mörs, F.; McDaniel Koch, A.; Graf, F.; Bajohr, S.; Reimert, R.; Kolb, T., Renewable Power-to-Gas: A technological and economic review. Renew. Energy 2016, 85, 1371-1390.Dincer, I.; Acar, C., Review and evaluation of hydrogen production methods for better sustainability. Int. J. Hydrogen Energy 2015, 40 (34), 11094-11111.Carmo, M.; Fritz, D. L.; Merge, J.; Stolten, D., A comprehensive review on PEM water electrolysis. Int. J. Hydrogen Energy 2013, 38 (12), 4901-4934.An, L.; Wei, C.; Lu, M.; Liu, H.; Chen, Y.; Scherer, G. G.; Fisher, A. C.; Xi, P.; Xu, Z. J.; Yan, C.-H., Recent Development of Oxygen Evolution Electrocatalysts in Acidic Environment. Adv. Mater. 2021, 33 (20), 2006328.Nong, H. N.; Falling, L. J.; Bergmann, A.; Klingenhof, M.; Tran, H. P.; Spöri, C.; Mom, R.; Timoshenko, J.; Zichittella, G.; Knop-Gericke, A.; Piccinin, S.; Pérez-Ramírez, J.; Cuenya, B. R.; Schlögl, R.; Strasser, P.; Teschner, D.; Jones, T. E., Key role of chemistry versus bias in electrocatalytic oxygen evolution. Nature 2020, 587 (7834), 408-413. Figure 1

A surfactant free hydrolysis approach was used to synthesize colloidal IrOx nanoparticles, which were loaded on a high surface area carbon support to obtain IrOx (IrOx/C), having a mass specific oxygen evolution reaction (OER) activity ∼ 5 times higher than that of unsupported IrO2, a benchmark commercial OER electrocatalyst. The synthesized IrOx/C and the unsupported IrO2 were then treated under microwave (MW) irradiation and their electrocatalytic activity and stability towards OER in acidic media was investigated. For both electrocatalysts, the microwave treatment improved the mass specific OER activity and enhanced the durability, with no apparent increase in the IrOx nanoparticle size, demonstrated in a potentiodynamic accelerated stress test (AST). Surface chemical state studies of the as-synthesized and MW-treated IrOx/C samples suggest increased Ir-O-Ir bonding on expense of the Ir-OH bonding through MWI. Further, using X-ray pair distribution function (PDF) analysis it was suggested that the MW treated IrOx/C leads to a gradual change in the local coordination from a rutile-like structure to a highly OER active hollandite-like structure. The presented OER catalyst synthesis route and microwave irradiation approach may be put to use as a scalable method for the stability enhancement of the IrOx electrocatalysts for acidic water electrolysis.

Zirconium dioxide (ZrO2) and hafnium dioxide (HfO2) have emerged as promising alternatives to conventional ferroelectric materials. Understanding the crystal phases of these oxides under different conditions is crucial for optimizing their properties. There are several theories for the (anti)ferroelectric properties; however, comprehensive analysis, particularly at the local structure level, is lacking. In this study, we investigate the local structure of ZrO2 nanocrystals using X-ray pair distribution function (PDF) analysis, revealing an unexpected local orthorhombic distortion irrespective of crystallite size. This finding suggests the potential existence of an intermediate orthorhombic phase during the microscopic switching pathway observed in previous studies. Additionally, we explore the influence of crystallite size and surface effects on the PDF. These results contribute to a deeper understanding of the structural dynamics in ZrO2 and offer insights for the design of next-generation ferroelectric materials.

Non-noble-metal, thin-film oxides are widely investigated as promising catalysts for oxygen evolution reactions (OER). Amorphous cobalt oxide films electrochemically formed in the presence of borate (CoBi) and phosphate (CoPi) share a common cobaltate domain building block, but differ significantly in OER performance that derives from different electron-proton charge transport properties. Here, we use a combination of L edge synchrotron X-ray absorption (XAS), resonant X-ray emission (RXES), resonant inelastic X-ray scattering (RIXS), resonant Raman (RR) scattering, and high-energy X-ray pair distribution function (PDF) analyses that identify electronic and structural factors correlated to the charge transport differences for CoPi and CoBi. The analyses show that CoBi is composed primarily of cobalt in octahedral coordination, whereas CoPi contains approximately 17% tetrahedral Co(II), with the remainder in octahedral coordination. Oxygen-mediated 4 p-3 d hybridization through Co-O-Co bonding was detected by RXES and the intersite dd excitation was observed by RIXS in CoBi, but not in CoPi. RR shows that CoBi resembles a disordered layered LiCoO2-like structure, whereas CoPi is amorphous. Distinct domain models in the nanometer range for CoBi and CoPi have been proposed on the basis of the PDF analysis coupled to XAS data. The observed differences provide information on electronic and structural factors that enhance oxygen evolving catalysis performance.

Atomic-scale structure of RCo2 (R[dbnd]Tb,Dy) compounds studied by X-ray diffraction and pair distribution function analysis

Here we describe the topological transformation of the pores of a new framework in the bio-MOF-100 family (dia-c) into the known isomer (lcs) by doubling the pore volume, which occurs during postsynthesis modifications. During this transformation, reassembling of the metal-organic framework (MOF) building blocks into a completely different framework occurs, involving breaking/forming of metal-ligand bonds. MOF crystallinity and local structure are retained, as determined by powder X-ray diffraction (PXRD) and pair distribution function (PDF) analyses, respectively. We exploited the inherent dynamism of bio-MOF-100 by coupling chemical decorations of the framework using solvent-assisted ligand exchange to the topological change. Following this method and starting from the pristine dense dia-c phase, open lcs-bio-MOF-100 was prepared and functionalized in situ with an iridium complex (IrL). Alternatively, the dia-c MOF could be modified with wide-ranging amounts of IrL up to ca. 50 mol %, as determined by solution 1H NMR spectroscopy, by tuning the concentration of the solutions used and with no evidence for isomer transformation. The single-site nature of the iridium complexes within the MOFs was assessed by X-ray absorption spectroscopy (XAS) and PDF analyses. Ligand exchanges occurred quantitatively at room temperature, with no need of excess of the iridium metallolinker.

Despite the importance of oxygen evolution reaction (OER) catalysts in energy conversion applications, the time-resolved dynamics of their amorphous phases remain elusive. Understanding the structural evolution of OER catalysts is pivotal to improve their design and sustainability and increase the energy conversion efficiency. SrIrO3 is a promising OER catalyst for achieving high energy conversion efficiencies, but it has not been sufficiently characterized in terms of its phase dynamics during operation. Herein, in-situ total X-ray scattering measurements and pair distribution function analysis are used to probe time-resolved structural changes in an IrOx/SrIrO3 catalyst during OER, revealing the occurrence of persistent Ir-O bond shortening and dynamic structural interconversions between multiple crystalline phases and an amorphous component. The relative contents of the main crystalline phases-3C-SrIrO3 (Pnma) and 6H-SrIrO3 (C2/c)-exhibited synchronized changes, whereas those of an additional C2/c phase and the amorphous component showed opposite trends, suggesting active structural interplay among the phases during the reaction. Furthermore, the observed Ir-O bond shortening under applied potential was attributed to a structural change associated with increased oxidation state of Ir, which is closely related to the local environment at the rate-determining step of OER. Our findings underscore the importance of precisely controlling bond distances and structural dynamics at the atomic scale for enhancing the activity of OER catalysts and provide valuable insights for future catalyst design.

Application of of polymer electrolyte fuel cells (PEFCs) in automotive sector shows great promise in arresting environmental degradation. They use hydrogen gas as a fuel that electrochemically reacts with air to produce electrical energy and water as a by-product. In a fuel cell electric vehicle (FCEV), these zero tail pipe emission systems offer high efficiency and power density for medium-heavy duty and long range transportation. However, PEFC technology is currently challenged by its high cost due to expensive platinum and limited durability when subjected to harsh and adverse operating conditions that can arise during the normal course of vehicle operation. A severe cause of PEFC degradation is “cell reversal”, resulting from partial and complete fuel starvation. Hydrogen starvation at the anode can arise from blokage in the hydrogen supply system by foreign impurities, water flooding or ice formation during winter [1,2]. Hydrogen starvation could be aggravated when FCEVs are operated under transient conditions such as start-up and rapid load change especially at the cells downstream in a cell stack. When the anode of a particular cell in the stack is starved of hydrogen, the anode requires an additional source of electrons and protons to complete the load circuit. At such a condition, the anode starts generating electrons and protons through water electrolysis reaction which is quickly followed by carbon corrosion reaction (in the presence of water). It causes the anode to consume itself to sustain the load demand, leading to severe degradation [3]. This process increases the anode potential while the cathode potential remains unchanged, leading to cell potential being reversed [1,3]. The existing concept of material-based solution involves adding a water electrolysis catalyst, i.e. an oxygen evolution reaction (OER) catalyst into the PEFC anode, which helps prolong the water electrolysis during hydrogen starvation and in the process preventing the anode potential from increasing further (i.e. cell potential plummeting down). These anodes are termed as Reversal Tolerant Anodes (RTA) [4]. The OER catalyst thus keeps the driving potential for carbon corrosion minimal and protects the cell by preventing self-consumption of the anode while fuel-starved under load demand. However, studies [2,4] indicate that the current strategy of adding expensive OER catalyst to enhance the durability of automotive fuel cells is only a temporary solution. The cell potential still plummets down to -2 V and lower, unless the load is terminated intentionally. Although the durability is increased to certain extent by slowing down the cell performance degradation caused by fuel starvation, the protection is not guaranteed for an indefinite time. The reversal tolerance increases with increasing loading of precious metal-based OER catalysts, however it further increases the PEFC cost. The question that we want to address is the cause of the ultimate failure or deactivation of the OER catalyst in the anode. The complexity of the electro-thermo-chemical phenomena occurring in a PEFC makes it difficult to pin-point the exact cause of degradation. In order to delineate the electrochemical phenomena under cell reversal condition, we need to subject each material component of the RTA to reversal condition and study its behavior. We propose to design well controlled experiment to explore the limitations of the materials used in PEFC electrodes and their vulnerability through insitu electrochemical diagnostics (Cyclic voltammetery, Chronopotentiometry, Electrochemical Impedance Spectroscopy, etc.) and also correlate the observations to the changes the material chemical-electronic and physical properties obtained through exsitu material characterization of the anode electrocatalyst layer. Custom anodes will be fabricated using (1) OER catalyst IrO2 with Nafion® ionomer binder, to study any intrinsic changes in IrO2 phisico-chemical properties and activity towards water electrolysis reaction with time. (2) IrO2 and different loadings of Carbon black (Vulcan, XC 72R) with Nafion ®ionomer binder, to find out if carbon corrosion alongside OER causes deactivation of the catalyst or if the deactivation is caused by electronic isolation of the OER aggregates in the anode. By gaining insight of how each of these materials influence the overall degradation mechanism when put together in an RTA we can eventually design better RTA by engineering new microstructure and material in an economical way. [1] A. Taniguchi, T. Akita, K. Yasuda, and Y. Miyazaki, J. Pow. Sources 130 (2004), 42-49. [2] T.P. Ralph, M. P. Hogarth, Platinum Met. Rev. 46 (2002) 117-135. [3] P. Mandal, B-K Hong, J-G Oh, S. Litster, ECS Trans. 69 (2015), 443-457. [4] T. R. Ralph, S. Hudson, D. P. Wilkinson, ECS Trans. 1 (2006), 67-84.

Ni–Fe (oxy)hydroxides, Ni(1–z)FezOxHy, are among the fastest-known water oxidation catalysts in alkaline media on a per-cation basis. At current densities relevant for electrolysis (e.g., >0.5 A/cm–2), mass and electron transport through catalyst films with high mass loading are critical and depend substantially on the extended and intermediate catalyst architecture. Here we use X-ray pair distribution function (PDF) analysis to determine the intermediate nanostructures of electrodeposited Ni(1–z)FezOxHy films. We report the effects of electrodeposition technique (pulsed versus continuous), electrochemical cycling, and Fe content on the structure of the catalyst film. The PDF patterns for Ni(1–z)FezOxHy films are best simulated by model structures consisting of brucite-like β-Ni(OH)2 fragments 1 to 3 layers in thickness. Only the oxidation state of the film significantly affects the intralayer scattering behavior (i.e., metal–oxygen bond distance). The interlayer interactions, however, are affected by Fe ...

X-ray pair distribution function (PDF) analysis of Al0.57Sn0.43O1.71, a promising candidate as anode material for Li ion batteries, has been carried out using synchrotron radiation at 60 keV. The average and short-range ordering of nanocrystalline Al0.57Sn0.43O1.71was investigated both with the traditional Rietveld refinement method and with X-ray PDF analysis. The instrumental parameters of the high-resolution powder diffraction beamline P02.1, Petra III, have been characterized. Reverse Monte Carlo (RMC) refinements based on PDF data indicate that Al substitution induces local distortions around the Al/Sn atoms and that oxygen vacancies, induced by the Al substitution, are mostly located on an anion site around the Al atoms. Additionally, RMC refinements hint at a clustering of Al atoms.

Catalyst for oxygen evolution reaction (OER) and oxygen reduction reaction (ORR) is at the heart of key renewable energy technologies such as water splitting and rechargeable batteries. But developing a low-cost oxygen electrode catalyst with high activity at low overpotential remains a great challenge. Coconut shells can be utilized as suitable raw material to produce activated carbon for enhanced adsorption capacity, bulk density, and hardness to be used as regenerative fuel cells running ORR and OER. The present work is designed to obtain an alternative to noble metal-based catalysts by synthesizing electroactive N-doped porous carbon from coconut shells; the use of biodegradable raw material through a single-step activation followed by nitrogen doping provides a more economical and environmentally friendly route to produce green catalysts for fuel cell applications. In valorization of biomass for the development of novel catalytic materials, our aim is also to reduce the use of hazardous chemicals. N-doped activated carbon shows promising bifunctional catalyst for ORR and OER as low-cost noble-metal-free and carbon-based oxygen catalysts.

It is crucial but challenging to reduce the required noble‐metal loading without compromising the catalytic performance of oxygen evolution reaction (OER) catalysts. This study presents a highly active OER catalyst composed of IrO2 with Ir rich surface (IrO2@Ir) nanoparticles supported over nano TiN coated with TiOxNy (IrO2@Ir/TiN). The present approach demonstrates superior OER catalysts with high activity through small, uniformly dispersed IrO2@Ir nanoparticles, along with high durability owing to robust catalyst support and strong catalyst‐support interaction. The synthesized IrO2@Ir/TiN with an Ir loading of 40 wt % exhibits a mass‐normalized OER activity of 637 AgIr−1, which is 2.4 times that of the unsupported commercial benchmark IrO2 OER electrocatalyst. The fine nanoparticles and high activity enable significant (∼60 %) reduction in the Ir metal loading required to obtain equivalent OER performance. In addition, when evaluated through an accelerated stress test using potential cycling, the catalyst exhibits outstanding durability (79 % retention) compared to that of the commercial equivalent (66 % retention). The OER activity loss was attributed to the catalyst dissolution (30 % loss) and the catalyst particle growth (70 %), with no measurable loss due to the TiN support corrosion. The development of ultra‐fine IrO2@Ir nanoparticles and robust ceramic catalyst support significantly improved the Ir utilization and open a new perspective for supported OER catalyst.

Surface self-reconstruction of catalysts in electrocatalytic oxygen evolution reaction

A bifunctional oxygen reduction reaction (ORR)/ oxygen evolution reaction (OER) catalyst is essential for rechargeable metal-air batteries and regenerative fuel cells. Platinum (Pt) and iridium oxide (IrO2) are the state-of-the-art ORR and OER catalysts, respectively. However the high price and scarcity of these platinum group metals (PGMs) has been an obstacle for wide spread application of these catalysts. Recently, in alkaline media, carbon based ORR and perovskite OER catalysts have demonstrated similar or even better catalytic activities compared to the counterpart PGM catalysts [1, 2]. Therefore, if we combine these two non-PGM catalysts, a non-PGM bifunctional ORR/OER catalyst can be obtained. A hindrance in this approach is the vulnerability of carbon-based ORR catalysts to oxidation in the OER potential range, i.e., potentials > 1.5 V vs. RHE. Thus development of robust ORR catalysts under practical OER conditions is a key to realize this kind of bifunctional catalysts. The carbon support used in our ORR catalysts was black pearl (BP) 2000 [1]. In preliminary tests, however, we found that BP 2000 undergoes oxidization at potentials around ca. 1.2 V vs. RHE and above (data not shown). In this work, we used reduced graphene oxide (rGO) as an alternative support to synthesize oxidation resistant ORR catalysts. The OER catalyst we chose was a perovskite (La1-xSrx)CoO3-δ (LSC). Pre-synthesized LSC was added into the initial solution of the rGO based ORR catalyst synthesis process, and after drying and heat-treatment, bifunctional (LSC + rGO) catalysts were obtained. In measuring the OER activity of the LSC catalyst, acetylene black (AB) carbon was added to the LSC (LSC + AB) to increase the electrical conductivity. Fig. 1 shows the comparison of ORR/OER activities between (LSC + AB) and (LSC + rGO). As expected, the ORR activity of (LSC + rGO) is greatly improved by ca. 200 mV in terms of E½ , in comparison to that of (LSC + AB). Interestingly even the OER activity of (LSC + rGO) becomes higher than that of (LSC + AB). Thanks to the enhancement of both ORR and OER activities with (LSC + rGO), highly active bifunctional catalysts are obtained. In this talk, material analysis results and diverse electrochemical performances of the (LSC + rGO) catalysts will be presented. Acknowledgements Support from the Directed Research of the Los Alamos National Laboratory’s Laboratory Directed Research & Development (LDRD-DR) is greatly acknowledged. References Chung et al., Nat. Commun. 4, 1922 (2013).Suntivich et al., Science, 334, 1383 (2011). Figure 1