駱明川，孫英俊,3，秦英楠,3，楊勇，吳冬，郭少軍,2,＊

1北京大學(xué)工學(xué)院材料科學(xué)與工程系，北京 100871

2北京大學(xué)工學(xué)院工程科學(xué)與新興技術(shù)高精尖中心，北京 100871

3青島科技大學(xué)化學(xué)與分子工程學(xué)院，山東 青島 266042

4北京大學(xué)前沿交叉學(xué)科研究院，北京 100871

1 Introduction

For several decades, the sluggish kinetic of oxygen reduction reaction (ORR) has hindered the wide-spread application of polymer electrolyte membrane fuel cells (PEMFCs), the ideal conversion device for clean and renewable energy, and is still not fully addressed1-5. This challenge has spurred tremendous efforts in searching and developing electrocatalytic materials towards the ORR with higher activities and better durability than the state-of-the-art carbon supported platinum nanoparticles (Pt/C), for the purpose of lowering the required Pt loading amount at cathode6-10. Among various reported strategies, the design and construction of novel Pt-based nanostructures with satisfied catalytic performance and sustainable Pt usage is considered as the most effective one11-17.

Prior to the nanostructure engineering, it is necessary to know the target structures that are expected to show promising catalytic performance. By combining fundamental understanding of structure-property relationships, theoretical calculations and single-crystal electrode techniques, several extended Pt-based surfaces have been demonstrated to hold impressive catalytic reactivity for ORR, including the monolayer-Pt surface18-27, Pt-skin (111) surface28-32, Pt-skeleton surface33-42, Pt-early transition metals alloy surface9,43-51and so on. In particular, the Pt3Ni-skin (111) surface developed by Markovic and co-workers showed nearly two orders of magnitude higher specific activity (SA, kinetic current density per unit area at 0.9 V versus reversible hydrogen electrode (RHE), proportional to the turnover frequency) than that of commercial Pt/C catalyst, representing the most active surface ever known32. The key challenge lies on how to extend the aforementioned well-defined surfaces onto nanocatalysts that can be practically fabricated into the cathode of PEMFCs. The capability to rationally design and deliberately engineer Pt-based nanomaterials is the key to address this challenge.

Benefiting from the rapid development in nanotechnologies, the past decade has witnessed significant progress in the development of novel Pt-based nanostructures that expose extremely active surfaces for the ORR. For example, to mimic the attractive Pt3Ni-skin (111) surface, octahedral Pt3Ni nanoparticles (NPs), synthesized via various wet-chemistry approaches, exhibited impressive ORR activities52. Several other nanostructures, with different compositions, sizes, shapes and architectures, were also designed and synthesized11. These important progresses can achieve ORR activity and stability far above the technical targets for automotive applications, albeit the translation of these spectacular gains in electrocatalytic performance to real fuel cells obviously lags behind53.

In this Review, we discuss the latest progress in boosting the electrocatalytic performance for ORR by means of tuning the dimensionality of multimetallic nanostructures. In particular, the design principles, construction methods, physicochemical properties as well as catalytic performance of these advanced electrocatalysts are discussed according to the dimensionalities of Pt-based nanocrystals, covering the 0-dimensional nanoparticles (0D NPs), 1-dimensional nanowires (1D NWs) or nanotubes (NTs), 2-dimensional nanoplates (2D NPLs) and 3-dimensional nanoframes (3D NFs) and nanodentrites (NDs). Finally, we present the remaining challenges and personal perspective on the development of promising ORR electrocatalysts by tuning the Pt-based nanostructures.

Shaojun Guo is currently a Professor of Materials Science and Engineering with a joint appointment at Department of Energy & Resources Engineering, at College of Engineering, Peking University. He received his BSc in chemistry from Jilin University (2005), Ph.D. from Chinese Academy of Sciences (2011) with Profs. Erkang Wang and Shaojun Dong, and joined Prof. Shouheng Sun's group as a postdoctoral research associate from Jan. 2011 to Jun. 2013 at Brown University. Then, he works as a very prestigious J. Robert Oppenheimer Distinguished Fellow at Los Alamos National Laboratory. His research interests are in engineering multimetallic nanocrystals and 2D materials for catalysis, renewable energy, optoelectronics and biosensors.

2 0-dimensional electrocatalysts

As is known, the majority of transition metals (TMs) tend to form close-packed face-centered cubic (fcc) crystal structure that is generally encased by three common low-index facets: [111], [100] and [110]54. According to Wolff′s theory, the truncated octahedron (Wulff polyhedron) is the equilibrium shape for a single crystal of an fcc metal, like Pt here. As a result, the 0D Pt-based nanostructure is relatively easy to be synthesized and stabilized. This is why the most common electrocatalyst for ORR is based on 0D Pt NPs. Previous mechanistic studies conclude that the key to accelerate the ORR rate is to optimize the binding strength between oxygen-based species (such as O, OH and OOH) and Pt surface via modification of the electronic structure of exposed Pt atoms. In practice, this can be achieved by rationally adding other elements into Pt7,55,56, manipulating the spatial distribution of different elements14,57, controlling the particle size58, tuning the exposed facets59and arrangement of atoms on the surface60, and so on. Due to limited space, it is not possible to discuss all of these achievements. We therefore selectively highlight recent progress in promoting the electrocatalytic performance of 0D Pt-based nanomaterials by alloying Pt with other TMs. The emphasis would be placed on the scientific breakthrough in fundamental understanding of such alloying effect on the ORR, rather than listing the detailed examples.

Due to higher activity, better durability and lower usage of Pt, alloy catalysts combining Pt with other TMs are among the most promising candidates to replace commercial Pt/C catalyst. As a matter of fact, PtCo alloy has been used as cathode (where the ORR happens) catalyst in the first commercial PEMFCs-powered vehicle Mirai lunched by Toyota company in 20148, confirming the practicability of such alloying strategy. The promoting effect of TMs to Pt has been studied for a long time, which was mainly ascribed to either ligand effect or strain effect in the field of electrocatalysis41. According to the position of periodic table, the chosen TMs can be further divided into two categories: early transition metals (ETMs) and late transition metals (LTMs), with the latter one being studied far more than the former one.

2.1 Pt-LTMs alloy catalysts

To date, numerous LTMs have been employed to alloy with Pt for the purpose of achieving higher catalytic performance over the pure Pt counterpart7. Meanwhile, efforts have also been devoted to identifying which element is the best partner for Pt. To date, the majority of literatures show that the 3d-transition metals, such as Co, Ni and Fe, in the composition of Pt3M are among the most active catalysts for the ORR. In order to rationalize this observation, the ORR activities of different Pt-based compositions have been compared on well-defined surfaces created by the ultrahigh-vacuum methodology10. Before further discussion, it is worth noting that the 3d-transition metals are not possible to survive the extremely acidic and high-potential environment during the operation of PEMFCs. And the PtM alloy nanoparticulate catalysts in the practical form normally hold a pure Pt shell that prevents further dissolution of M in the core. Therefore, the well-defined Pt3M surface has been designed to be terminated with either Pt-skeleton or Pt-skin structure31. By scanning the electrocatalytic performance of five compositions (Pt3Ni, Pt3Co, Pt3Fe, Pt3Ti, Pt3V), the highest activity was observed in the case of Pt3Co, followed by Pt3Ni and Pt3Fe (Fig.1a). Based on the observed activity trend, it was proposed that the ORR catalytic activity was governed by a balance between the binding strength of reactive intermediates and the surface coverage by spectator species (anions form the electrolyte).

This was also in agreement with the Sabatier principle, that is, the best ORR catalyst should exhibit neither too strong nor too weak oxygen binding energy61. In their widely-cited paper, Markovic and co-workers further confirmed the above proposal, and presented the most active single-crystalline surface ever-known, that is, Pt3Ni (111) surface (Fig.1b). The unique Pt skin structure rendered the Pt3Ni (111) surface two orders of magnitude higher specific activity (at 0.9 V vs RHE) over commercial Pt/C catalyst32. Thereafter, numerous research groups tried to design and synthesize practical nanocatalyst with a similar Pt skin on Pt3Ni (111) surface, such as the Pt3Ni octahedrons (Fig.1c)62and Pt3Ni icosahedrons (Fig.1d)63, which continuously break the activity record in terms of Pt mass in the past few years.

Fig.1 (a) Catalytic activity as a function of d-band center position on various Pt3M catalysts. (b) Influence of the crystalline facets and electronic properties on the kinetics of ORR. HRTEM image of Pt3Ni (c) octahedrons and (d) icosahedrons. (e) Catalytic activities of polycrystalline Pt5M electrocatalysts as a function of corresponding lattice parameters. (a) Reprinted with permission from Ref.31.(b) Reprinted with permission from Ref.32.(c, d)Reprinted with permission from Ref.62.(e) Reprinted with permission from Ref.45.

2.2 Pt-LTMs alloy catalysts

Not like the LTMs, scientist started to notice the significance of Pt-ETMs alloys for catalyzing the ORR only several years ago50,51probably due to the synthetic difficulties in obtaining the Pt-ETMs alloy in nanoparticular form (as the ETMs are usually in extreme electronegativity, no effective reducing agents have been reported to obtain ETMs in metallic value). In 2009, based on the results of density functional theory calculations, a series of Pt-ETMs alloys were identified as promising ORR catalysts with both high activity and good stability64. The composition of Pt3Y was determined to be 6-10 times higher than that of pure Pt. Another important finding from this work is the outstanding stability of these Pt-ETMs alloy catalysts derived from the unique state of d bonds. For the Pt-ETMs alloys, Pt was supposed to denote 9 electrons from its d bond, while ETMs contribute 1 electron, which give rise to a fully filled bonding states and empty anti-bonding states9.

The Chorkendorff group has done a lot of work both theoretically and experimentally to push the unique Pt-ETMs alloys surface into practical ORR electrocatalysts46,49,50. In rationalizing the active sites of Pt-EMTs, they proposed that a compressive strain formed in the Pt shell (EMTs would rapidly dissolute into the acidic electrolyte, thus resulting in a pure Pt shell), which helped decrease the oxygen adsorption strength is the most important factor47,48. This proposal obviously contradicts the well-established trends: assuming a core-shell structure with a pure Pt shell, a smaller core is supposed to induce a compressive surface strain, while a larger core is supposed to induce a tensile surface strain. Since the atomic radius of EMTs elements are normally larger than that of Pt, a tensile strain is expected on the surface of a Pt-EMTs alloy NP41. The strain issues will not be discussed further in this review, and interesting readers can refer to these references44,47,65. In spite of the attractive advantages, it is very challenge to obtain the Pt-ETMs nanocatalysts due to the absence of a proper reducing agent, because the reduction potentials for EMTs are too negative. To fill this gap, Chorkendorff and co-workers synthesized PtY and PtGd NPs via a mass-selected gas-aggregation technique49. In agreement with the polycrystalline electrode experiments, the nanoparticulate Pt-ETMs catalysts showed much enhanced activity and stability over the benchmark Pt/C catalysts. More recently, they studied ORR on eight Pt-ETMs catalysts in the composition of Pt5M (Fig.1e), showing activity enhancement by a factor of 3 to 6 in relative to pure Pt, which further confirms the key role of compressive strain in accelerating the rates of ORR on Pt45.

3 1-dimensional electrocatalysts

The unique structure of 1-dimensional nanomaterials enables many distinct properties in comparison with the traditional 0-dimensional NPs for electrocatalysis. In literatures, the nanowires (NWs) and nanotubes (NTs) represent two typical 1-dimensional nanocatalyts, and thus are selected to be the focus of this section. The main structural characters of NWs or NTs are fewer lattice boundaries, lower density of defects, and anisotropy. These characters enable high electron and mass transport, low vulnerability to be poisoned by spectators, favorable reaction rates and stabilities during the operation of fuel cells66. In the following, we discuss recent progress on Pt-based NWs and NTs that are applied as ORR electrocatalysts.

3.1 Nanowire electrocatalysts

Numerous approaches have been developed for the synthesis of Pt-based nanowires, among which the wet-chemistry is the most powerful one. Intensive research in nanotechnology during the past few decades has enabled deliberate manipulation on the NW′s composition, size and structure that are closely related to the electrocatalytic performance67. To yield an anisotropic structure, surfactants are usually involved to guide the growth of desired directions and the suppression of undesired directions. We suggest here that some of the polymer surfactants, such as polyvinyl pyrrolidone (PVP), had better not included in the synthesis system, because they were reported to be detrimental to the electrocatalysis and notoriously difficult to be removed. The following highlights two types of nanowires developed in our group: hierarchical nanowires and ultrathin nanowires.

3.1.1 Hierarchical nanowires

Previous studies on Pt single-crystal electrodes in weak-adsorption acid (such as HClO4) have concluded the following activity trends: (100) < (111) < high index planes, which inspired tremendous efforts to transplant the high-active extended surface onto the practical nanomaterials4. To combine the exceptional activity of high-index planes and high stability of 1-dimensional structure, hierarchical NWs electrocatalysts have been developed recently. Through a facile wet-chemical approach, hierarchical bimetallic/trimetallic PtM (M = Ni, Co, Fe, Rh et al.) enclosed by high-index facets were obtained (Fig.2(a, b)). These unique structures showed extremely high catalytic activity towards the ORR68. For instance, in the cased of PtCo hierarchical nanowires, the mass activity was found to be 33.7 times higher than that of commercial Pt/C catalyst (Fig.2c)69. High stabilities were also found on this type of 1D catalysts, as revealed by the negligible degradation of activity before and after accelerating durability tests (Fig.2d). The impressive stability was mainly due to the larger aspect ratio and more anchoring sites with carbon support than nanoparticulate catalysts. Besides, the hierarchical NWs also showed excellent anti-aggregation properties upon the high-temperature annealing. The 1D hierarchical structures were well maintained even after being subjected to 500 °C. Considering the impressive catalytic properties, we believe these 1D NW nanocatalysts represent promising candidate alternative to the state-of-the-art Pt/C catalyst. However, a shortage of these NWs concerning the ORR electrocatalysis is the relative large diameter that leads to lower ECSAs and utilization efficiency of Pt. This problem could be addressed by creating hierarchical structures on thinner Pt-based NWs, which is still very challenging in nanotechnology.

Fig.2 Representative (a) STEM and (b) TEM images of the hierarchical Pt3Co NWs. (c) Comparison of specific and mass activities on various catalysts. (d) The changes on specific and mass activities of the hierarchical Pt3Co NWs/C catalyst before and after 10000, 15000 and 20000 potential cycles. (a, b) Reprinted with permission from Ref.68. (c, d) Reprinted with permission from Ref.69.

3.1.2 Ultrathin nanowires

As discussed above, NWs with smaller diameters are highly desirable for electrocatalysis due to their well balance in the durability/intrinsic activity and ECSAs. Moreover, the latest research based on practical fuel cell tests demonstrated that the ECSA of Pt was the dominant factor in determining the high-power performance for the low-Pt electrode due to the existence of the so-called local O2resistance53, which makes the development of ultrathin NWs electrocatalysts even more significant for future applications. In this section, we review recent advances in constructing ultrathin Pt-based NWs as efficient catalysts towards the ORR. Before further discussion, we here define “ultrathin NWs” to be the nanowires with a diameter of less than 3 nm.

Using a simple organic-phase decomposition of [Fe(CO)5] (or [Co2(CO)8]) and reduction of [Pt(acac)2] in a mixture of ODE and OAm containing sodium oleate, PtFe and PtCo NWs in diameters of 2-3 nm were obtained70,71. The mass activity of the PtFe NWs catalyst was 3 times higher than commercial Pt/C catalyst (Fig.3a). Significantly, the ORR polarization curve before and after stability tests almost overlapped each other, indicating high stability of NWs catalysts during the electrochemical operations (Fig.3b). More recently, even thinner Pt-based alloy NWs with a diameter of only 4-5 atomic layer thickness were synthesized (Fig.3(c, e))72. As we known, decreasing NPs to atomic size would largely increase the proportion of under-coordinated surface atoms, on which the oxygen binding energy was significant larger than the atoms on smooth facets, thus resulting in poor specific activity and stability. However, the atomic thick NWs mainly exposed smooth (111) facets with low amounts of defects (Fig.3(d, f)). It was thus not surprise that these NWs exhibited an exceptional mass activity of 4.2 A·mg-1at 0.9 V vs RHE. Considering the exposed atoms were well-coordinated, the atomic thick NWs showed negligible activity decay over the course of 30000 cycles.

Fig.3 (a) Comparison of specific activities of different catalysts at 0.9 and 0.95 V vs RHE. (b) Polarization curves of the FePt NWs before and after 4000 potential cycles. Representative STEM (c and e) and HRTEM (d and f) images of subnanometer PtCo and PtNiCo NWs. HRTEM images (g-i) of the Pt/NiO core/shell nanowires, the PtNi alloy nanowires, and the J-PtNWs supported on carbon, respectively. (a, b) Reprinted with permission from Ref.70. (c-f) Reprinted with permission from Ref.72. (g-i) Reprinted with permission from Ref.73.

The ultimate solution to maximize Pt mass activity would be nanostructures composed of ultra-small dimensionality (to maximize the dispersion of Pt atoms) and high-active surface structure (to maximize the specific activity). To simultaneously address these two aspects, Huang and co-workers prepared jagged Pt NWs with diameter of around 2 nm, on which the world-record mass activity of 13.6 A·mg-1was achieved73. First, Pt/NiO core/shell NWs were synthesized via a wet- chemical approach (Fig.3g). Then, the core/shell NWs were converted into PtNi alloy NWs through thermal annealing (Fig.3h). The jagged Pt NWs was finally obtained via electrochemical dealloying (Fig.3i). Impressively, the ECSA of the jagged NWs catalyst reached 118 m2·g-1, almost 2 times higher than the commonly used carbon supported Pt NPs in diameter of around 3 nm. Moreover, the ultrahigh specific activity of around 11.5 mA·cm-2was ascribed to the highly stressed, undercoordinated rhombus-rich surface configurations74. The mass activity could be well preserved after 6000 accelerating cycles.

3.2 Nanotube electrocatalysts

The first attempt to use Pt-based NTs as unsupported electrocatalysts for the ORR was reported by Yan and co-workers75. With the Ag NWs as the sacrificial template, they synthesized polycrystalline Pt NTs through galvanic replacement and subsequently Kirkendall effect. The diameter, length and wall thickness of resultant Pt NTs are 50 nm, 5-20 mm and 4-7 nm, respectively (Fig.4(a, b)). Such large dimension granted the Pt NTs with outstanding stability. After 1000 electrochemical cycles between 0 and 1.3 V vs RHE in Ar-saturated 0.5 mol·L-1H2SO4electrolyte, the electrochemical active surface area (ECSAs) of Pt NTs decreased by only 20%, much better than those of commercial Pt black (51%) and commercial Pt/C (90%) catalysts (Fig.4c). Even though the ECSA of NTs catalysts was much lower than the Pt/C, the mass activity (activity in terms of Pt mass) of the former was still higher than the latter, mainly due to their 4 times enhancement in specific activity (Fig.4d). To further promote the catalytic performance of Pt-based NTs, Pd was subsequently added to form bimetallic NTs, resulting in even higher activity than the pure Pt NTs. In spite of the promising performance of NTs catalysts, there is still a large space to further improve the mass activity that is more meaningful to the practical applications. One of the most interesting researches would be how to expose the Pt atoms that are buried under the thick wall without breaking the exceptional active surface.

4 2-dimensional electrocatalysts

Even since the breakthrough on graphene by Novoselov and Geim76, 2-dimensional nanomaterials have been studied intensively in the past decade, with applications covering physics, chemistry, biology as well as electronics and optics77. The majority of electrocatalysts are based on pure metals. To date, it is still challenge to synthesize 2-dimensional nanostructures of pure metals, because metal atoms tend to form fcc/hcp crystal structures78. That is why only a few articles reported on 2-dimensional Pt-based nanomaterials for the ORR electrocatalysis.

Fig.4 (a) SEM and (b) TEM images of PtPd NTs (inset shows the electron diffraction pattern). (c) Loss of electrochemical surface areas of Pt/C, platinum-black and Pt NTs with CV cycles. (d) ORR curves of Pt/C, platinum black, Pt NTs and PdPt NTs. Reprinted with permission from Ref.75.

Inspired by the Hummer′s method to prepare graphene, ultrathin Pt nanosheet with thickness of around 1 nm was prepared by chemically exfoliating layered platinum oxide79. Experimentally, layered Pt nanosheet precursor was firstly synthesized by mixing an aqueous solution of K2PtCl4, hexamethylenetetramine (HMT), and sodium dodecyl sulfate (SDS). After centrifugation and rinsing with distilled water and ethanol, the layered Pt precursor was exfoliated in a mixture of ethanol and tetrabutylammonium hydroxide (TBAOH). Finally, the Pt nanosheet was obtained by chemically/electrochemically reducing. Impressively, when catalyzing ORR in alkaline electrolyte, the mass activity of Pt nanosheet was more than two orders of magnitude higher than the commercial Pt/C catalyst. In spite of the high activity, several questions concerning the practical applications in fuel cells remain. The primary question arises on how to assemble the highly anisotropic nanosheet into the catalytic layer with high mass and electron transport, and whether the ultrathin nanosheet is stable?

Recently, core/shell structured PtPb/Pt hexagonal nanoplates were presented as the ORR electrocatalysts with both superior activity and exceptional stability80. In a typical synthesis, Pt(acac)2and Pb(acac)2, oleylamine/octadecene mixture, and ascorbic acid were used as the metal precursors, solvents/surfactants, and reducing agent, respectively. The edge length and thickness of the resultant PtPb nanoplates were around 16 nm and 4.5 nm, respectively, and the atomic ratio of Pt/Pb was close to 1/1. More detailed characterization revealed the PtPb nanoplates were in core/shell structure composed of an PtPb intermetallic core and a pure Pt shell (4 to 6 atomic layers) (Fig.5(a, b)), which was responsive for an impressive stability, with a negligible activity decay after cycling for 50，000 cycles. Moreover, due to the unique assembly of atoms, the majority of exposed surface are Pt (110) facets, which is especially active for ORR in acidic electrolyte with weak anionic adsorption strength. Electrochemical tests showed the specific activity and mass activity reached 7.8 mA·cm-2and 4.3 A·mg-1at 0.9 V vs RHE, respectively (Fig.5(c, d)). The density functional theory calculations reveal the enhanced activity was ascribed to the Pt (110) facets with a tensile strain induced by the PtPb intermetallic core. This article represents the first claim on the beneficial role of tensile strain on Pt towards ORR, and also suggests that the influence of induced strain on the adsorption behavior of Pt is highly facet-dependent.

Considering the advantages of the above-mentioned 2-dimensional nanocatalysts, we here suggest the following research directions: (1) mixing other transition metals with Pt to achieve the surface strain tuning and thus to optimize oxygen binding strength on exposed facets; (2) decreasing the thickness of nanoplate to maximize the atomic utilization of Pt; (3) further surface modifications, such as creating interfaces by adding other elements or phases.

Fig.5 (a) TEM image of PtPb nanoplates. (b) A model of one single hexagonal nanoplate and HAADF-STEM image from in-plate view. (c) ORR polarization curves and (d) specific and mass activities of different catalysts. Inset in (c) is the CVs of different catalysts. Reprinted with permission from Ref.80.

5 3-dimensional electrocatalysts

Intensive research in nanotechnologies over the past few decades has enabled delicate manipulation of very complicated nanostructures with desired properties. The complicated 3D nanomaterials are attracting growing interest in the field of electrocatalysis due to their unique geometry that provides a pathway for tuning physical and chemical properties. In this section, we discuss the following three typical 3D nanocatalysts: nanocages/nanoframes, porous nanostructures and nanodentrites.

5.1 Nanocages and nanoframes

The concept of nanocage catalysts was initially proposed to maximize the mass activity of precious metal-based electrocatalytic materials81, which was the major obstacle in hindering the wide-spread application of several energy conversion/storage technologies. Taking the electrocatalysis of ORR as an example, the mass activity of a given catalytic metal is the sum of its specific activity and electrochemical active surface area (ECSA). Currently, the most widely used catalyst for ORR is carbon supported Pt NPs with diameter of only 3-5 nm, in order to achieve a considerable high ECSAs and thus high Pt utilization efficiency. However, previous studies have shown that the technical target of mass activity (0.44 A·mg-1, proposed by the U.S. Department of Energy) for automobile applications cannot be achieved by solely increasing the ECSAs via decreasing the particles size58,82. A substantial enhancement in the specific activity on Pt is also highly demanded. Additionally, for conventional NPs system, the proportion of under-coordinated atoms located at vertices or edges detrimental to both the specific activity and long-term durability, increases with decreasing particle size12. Therefore, the most desired nanostructure should not only integrate specific activity and ECSAs, but also minimize the proportion of vertices/edges. Nanocages with wall thickness as thin as a few atomic layers are considered as one promising choice.

The thin-wall Pt nanocages were initially studied as the ORR electrocatalysts by Xia′s group. To fabricate nanocages, they first deposited a few atomic layers of platinum as conformal shells on shape-controlled palladium nanocrystals83. The main reason for choosing Pd as the substrate material is due to the small lattice mismatch of only 0.77% that favors the epitaxial growth of Pt on Pd84. In spite of the small lattice mismatch, it was still not easy to achieve layer-by-layer deposition of Pt atoms, because the intrinsic high surface energy and strong interatomic bond energy of Pt favor an island growth mode thermodynamically. Xia and co-workers overcame this challenge by controlling ratio between the deposition rate and surface diffusion rate in order to avoid the generation of nano-island on Pd surface85. On the basis of the core/shell Pd/Pt structure, the nanocages with well-defined facets were obtained by etching away the Pd substrate via a chemical approach. It was found that the resultant wall was made of PdPt alloy with a thickness of only several atomic layers (Fig.6(a-d)). By selecting different Pd templates, nanocages in different shape were readily prepared, such as cubes, octahedra and icosahedra86. When using as the ORR electrocatalysts, the nanocages in an octahedral shape exhibited an impressive specific activity of 1.98 mA·cm-2at 0.9 V vs RHE, around 8 times higher than that of benchmark Pt/C catalyst. The enhanced activities were mainly attributed to the well-defined facets as well as induced compressive strain. However, the ECSAs of these nanocages catalysts are even less than that of the commercial Pt/C (50-60 m2·g-1), which is far below one′s initial expectation, considering the ultrathin wall and the access of the inner surface, thus resulting in mass activities below 1 A·mg-1(Fig.6e). Nevertheless, these nanocages catalysts still showed fairly good durability (Fig.6f). As a result, further efforts can be devoted to enlarging the ECSAs by creating nanocages with even thinner walls.

Nanoframes represent another type of nanostructures combining high specific activity and high ECSAs for ORR electrocatalysis. Chen et al. transferred the spectacular ORR activity of Pt-skin overlaying Pt3Ni (111) single crystal to Pt3Ni nanoframes87, which exhibited extremely high specific and mass activities, exceeding those of benchmark Pt/C by 16 and 22 times, respectively. Similar to the synthesis of nanocages, solid crystalline PtNi3polyhedra were used as the starting materials to synthesize Pt3Ni nanoframes that were composed of 24 edges with thicknesses of around 2 nm (Fig.7a). Especially, after properly thermal treating, each edge was enclosed by Pt-skin surface, favoring the ORR rates31. To make use of the hollow structure, they further filled ionic liquid into the nanoframes. Since the used ionic liquid ([MTBD][NTf2]) had a higher O2solubility than the commonly used liquid electrolyte, a higher ORR activity was then expected on this composite catalyst. Indeed, the ionic liquid-encapsulated Pt3Ni nanoframes exhibited a factor of 36 enhancement in mass activity relative to benchmark Pt/C (Fig.7(b, c)), along with superior stability during the prolonged potential cycling. Recently, the PtNi nanoframes in tetrahexahedral and rhombic dodecahedral shape were obtained by combining a wet-chemical method and acidic leaching88. Such shape-controlled nanoframes catalysts also showed superior electrocatalytic performance towards ORR.

5.2 Porous nanostructures

Fig.6 (a) TEM and (b) HAADF-STEM images of nanocages. (c) High-resolution HAADF-STEM image taken from the region boxed in (b), showing a wall thickness of six atomic layers. (d) EDS elemental mapping of Pt and Pd for two nanocages. (e) Mass activities of the catalysts at 0.9 V vs RHE. (f) Mass activities of the catalysts before and after accelerated durability test. Reprinted with permission from Ref.83.

Fig.7 (a) Schematic illustrations and corresponding TEM images of the samples obtained at four representative stages during the evolution process from polyhedra to nanoframes. Specific activities (b) and mass activities (c) measured at 0.95 V vs RHE, and improvement factors vs Pt/C catalysts.Reprinted with permission from Ref.87.

In heterogeneous catalysis, nanoporous catalysts have gained special attentions because of their capability to offer more reactive sites than the conventional nanoparticulate catalysts89,90. Additionally, nanoconfinement effect in nanoporous structure is also beneficial to the catalytic process38. A general method to prepare Pt-based nanoporous catalysts is the so-called dealloying process, during which other transition metals are selectively removed from the initial PtM bulk via either chemical or electrochemical way34,42. The key to obtain nanoporous structure is controlling the ratio of the dissolution rate of transition metals and the surface diffusion rates of Pt. After the removal of transition metals on the surface, the mobility of exposed Pt atoms increases due to a decreased coordinated number. Then, if the surface diffusion rates of Pt atoms are larger than the dissolution rate of transition metals, these transition metals would be protected from further dissolution, resulting in a solid core/shell structure with PtM alloy as the core and pure Pt as the shell; conversely, nanoporous structure yields (Fig.8a). Experimentally, the two rates discussed above can be controlled by manipulating the structural parameters of nanocatalysts, such as the composition, the dimensional size, and distribution of elements40.

Since high dissolution rate of transition metals is required to form nanoporous structure, PtM alloy with higher percentage of M is more favorable in the synthesis of nanoporous catalysts, making the composition of PtM3the most typical precursor91. The size of PtM alloy precursor is another key structural parameter that determines the corresponding dealloyed structure92. The surface diffusion of Pt atoms on larger NPs is slower than that of Pt atoms on smaller NPs due to higher average coordinated number in the former. Therefore, during the dealloying process, nanoporous structure tends to form in bulk alloy materials; while solid core-shell structure tends to form in small NPs. Efforts have been devoted to identifying such critical particle size, beyond which the nanoporous structure generates. For example, using PtNi3as the precursor alloy, Strasser and co-workers found the dealloyed NPs larger than 15 nm could form the nanopores while the NPs smaller than 15 nm could result in a solid core/shell structure (Fig.8b)92. Similarly, Snyder et al. also showed the critical diameter of NPs for porosity evolution required around 15 nm, below which the surface would be quickly passivated by Pt during the stages of transition metals dissolution, thus preventing further etching to the interior of the NPs38. In addition to the structural parameters of initial alloys, it was also found that the dealloying protocol, such as the types of applied acid and dealloying atmosphere have the determinant roles in the final structure93. Normally, oxidative acid and oxygen atmosphere would favor the generation of nanoporosity, because transition metals are more oxophilic than Pt (Fig.8c)40.

The ORR activity of nanoporous PtNi nanocatalyst was several times higher than commercial Pt/C catalyst, and even slightly higher than the solid core/shell counterpart, mainly due to a higher surface area/volume ratio and a possible nanoconfinement of oxygen molecules that enables increased attempt frequencies38,94. After the porous NPs were encapsulated with high-oxygen-solubility protic ionic liquid, the activity can be further enhanced, with half-wave potential reaching nearly 1.0 V (Fig.8d)95. The nanoporous structure has also been transplanted into 1D nanotubes96and 2D nanofilms97, on which the ORR activity improved significantly. In spite of the impressive activity enhancement, Gan et al. demonstrated these nanoporous catalysts were highly unstable during the electrochemical cycling, further resulting in a severe activity degradation40. In comparison, the solid core-shell catalyst was more stable during the long-term operation and able to maintain the initial high activity. Therefore, the durability issues of nanoporous catalysts should be well addressed before their further implement into practical fuel cells.

5.3 Nanodendrites

Nanodendrites are of particular interest for catalysis due to their unique highly branched structures. First, nanodendrites composed of ultrathin branches enable high surface area/mass ratio and thus a better utilization efficiency of precious metals. Secondly, the branched structures typically have high proportion of edges, corners, and stepped atoms, which are more beneficial to the catalytic rates than the facets atoms89. To date, various synthetic approaches have been developed to prepare nanodendrites, including kinetically controlled overgrowth, aggregation-based growth, heterogeneous seed-mediated growth, selective etching and template-directed methods98. These approaches have been well discussed in previous comprehensive review papers, and thus were not specified here. In the following, we selectively discuss recent progress in promoting ORR electrocatalysis via engineering the dendritic structures.

Using polyhedral Pd NPs as seeds, Xia and co-workers deposited multiple Pt branches on the seeds and subsequently prepared electrocatalysts by mixing these PdPt nanodendrites with carbon (Fig.9(a, b))99. The ECSA of these dendritic catalysts was 3 times higher than that of the Pt black catalysts, and only slightly lower than the Pt/C catalyst, suggesting reasonably high utilization efficiency of Pt. Moreover, the dendritic catalysts showed over 3 times enhancement in specific activity than that of Pt/C (Fig.9(c, d)) due to the preferential exposure of particularly active facets, such as [111], [110] and [311] on the Pt branches. Together, the mass activity of PtPb nanodendrites were around 2.5 and 5 times higher than those of the commercial Pt/C and Pt black catalysts, respectively. Even though the seed-mediated growth had certain advantages in controlling the size and morphology, the

Fig.9 (a) TEM and (b) HRTEM images recorded from Pt-Pb nanodentrites. (c) ORR polarization curves for the Pd-Pt nanodendrites, Pt/C catalyst and Pt black. (d) Mass activity at 0.9 V vs RHE for these three catalysts. Reprinted with permission from Ref.99.

complicated and multi-step procedure greatly hindered its scalable applications. Later, Huang et al. developed a one-pot synthetic approach for the Pt3Ni nanodendrites, which served as highly efficient catalysts towards the ORR100. In comparison with the commercial Pt/C and Pt black catalysts, the Pt3Ni nanodendrites showed 6.4 and 16.7 enhancement in terms of mass activity at 0.9 V vs RHE, respectively. For the nanoparticulate catalysts, the use of carbon support is necessary to avoid aggregation of NPs during the operation and also to improve the conductivity. As revealed previously, the carbon corrosion contributed significantly to the total activity degradation of carbon supported catalysts. To address this issue, self-supported Pt nanodendrites catalyst was used to further improve the durability101. It was found that the ECSA of self-supported Pt nanodendrites could be well maintained after the accelerated durability tests, while that of Pt/C and carbon supported Pt nanodendrites decreased significantly.

6 Conclusions and perspective

The past decade has witnessed tremendous progress in promoting the efficiency of ORR electrocatalysis, which is the key technical barrier for fuel cells, including the fundamental studies on reaction mechanism, researches in structure- performance relation and design/construction of promising nanostructure. Specifically, significant enhancement in ORR activity has been achieved by tuning the Pt-based nanostructures with different dimensionality, which brings the fuel cell technologies closer to wide-spread applications. It is thus reasonable to expect that, via rationally structural design/tuning, the ultimate catalytic performance target on both activity and durability will be realized.

In this review, we discussed recent advances in Pt-based electrocatalysts in terms of dimensionality, including 0D NPs, 1D NWs/NTs, 2D NPs and 3D NCs/NFs/NDs. Rather than listing a large number of examples, we here stress the mechanistic studies on the structure-performance relationship of Pt-based nanocatalysts for ORR, and selectively highlight classic examples on using these fundamental guidelines to design and construct nanostructures with promising catalytic activity and stability. We also provide possible research directions for further promoting the catalytic performance.

Despite the incredible progress in theoretically understanding the ORR mechanism and experimentally boosting the catalytic performance in the past few years, a number of challenges still exist to finally eliminate the ORR electrocatalysis problem in fuel cells. Here, we offer some challenges and associated research directions in this field: (a) The catalytic performances of most of currently high-active nanocatalysts discussed in this paper are evaluated in ideal half-cell settings. It remains to be seen if these promising activity and stability can be transferred to the real fuel cells. Therefore, more works should be done to bring the progress achieved in half-cells to practical fuel cells. (b) Although some nanostructures show the impressive catalytic performance, the associated synthetic processes are too complicated or expensive to be scaled, which makes them difficulty to be practical applied. Therefore, great efforts should also be placed on developing more efficient synthetic approaches. (c) Little knowledge has been gained in the structural evolution of nanocatalysts during the working conditions, which makes the in-situ characterization techniques urgently be needed.

(1) Gasteiger, H. A.; Markovic, N. M. Science 2009, 324, 48. doi: 10.1126/science.1172083

(2) Mayrhofer, K. J.; Arenz, M. Nat. Chem. 2009, 1, 518. doi: 10.1038/nchem.380

(4) Shao, M.; Chang, Q.; Dodelet, J. P.; Chenitz, R. Chem. Rev. 2016, 116, 3594. doi: 10.1021/acs.chemrev.5b00462

(5) Bezerra, C. W. B.; Zhang, L.; Liu, H.; Lee, K.; Marques, A. L. B.; Marques, E. P.; Wang, H.; Zhang, J. J. Power Sources 2007, 173, 891. doi: 10.1016/j.jpowsour.2007.08.028

(6) Stephens, I. E.; Rossmeisl, J.; Chorkendorff, I. Science 2016, 354, 1378. doi: 10.1126/science.aal3303

(7) He, T.; Kreidler, E.; Xiong, L.; Luo, J.; Zhong, C. J. J. Electrochem.Soc. 2006, 153, A1637. doi: 10.1149/1.2213387

(8) Yoshida, T.; Kojima, K. Electrochem. Soc. Interface 2015, 24, 45. doi: 10.1149/2.F03152if

(9) Stephens, I. E. L.; Bondarenko, A. S.; Gr?nbjerg, U.; Rossmeisl, J.; Chorkendorff, I. Energy Environ. Sci. 2012, 5, 6744. doi: 10.1039/C2EE03590A

(10) Stamenkovic, V. R.; Mun, B. S.; Arenz, M.; Mayrhofer, K. J.; Lucas, C. A.; Wang, G.; Ross, P. N.; Markovic, N. M. Nat.Mater. 2007, 6, 241. doi: 10.1038/nmat1840

(11) Seh, Z. W.; Kibsgaard, J.; Dickens, C. F.; Chorkendorff, I.; Norskov, J. K.; Jaramillo, T. F. Science 2017, 355. doi: 10.1126/science.aad4998

(12) Norskov, J. K.; Bligaard, T.; Rossmeisl, J.; Christensen, C. H. Nat. Chem. 2009, 1, 37. doi: 10.1038/nchem.121

(13) Zhou, Z. Y.; Tian, N.; Li, J. T.; Broadwell, I.; Sun, S. G. Chem. Soc. Rev. 2011, 40, 4167. doi: 10.1039/C0CS00176G

(14) Wang, Y. J.; Zhao, N.; Fang, B.; Li, H.; Bi, X. T.; Wang, H. Chem. Rev. 2015, 115, 3433. doi: 10.1021/cr500519c

(15) Guo, S.; Zhang, S.; Sun, S. Angew. Chem. Int. Ed. 2013, 52, 8526. doi: 10.1002/anie.201207186

(16) Yang, H. Angew. Chem. Int. Ed. 2011, 50, 2674. doi: 10.1002/anie.201005868

(17) Zhu, H.; Luo, M. C.; Cai, Y. Z.; Sun, Z. N. Acta Phys. -Chim.Sin. 2016, 32, 2462. [朱紅, 駱明川, 蔡業(yè)政, 孫照楠. 物理化學(xué)學(xué)報, 2016, 32, 2462.] doi: 10.3866/PKU.WHXB201606293

(18) Zhang, J.; Mo, Y.; Vukmirovic, M. B.; Klie, R.; Sasaki, K.; Adzic, R. R. J. Phys. Chem. B 2004, 108, 10955. doi: 10.1021/jp0379953

(19) Zhang, J.; Vukmirovic, M. B.; Xu, Y.; Mavrikakis, M.; Adzic, R. R. Angew. Chem. Int. Ed. 2005, 44, 2132. doi: 10.1002/anie.200462335

(20) Adzic, R. R.; Zhang, J.; Sasaki, K.; Vukmirovic, M. B.; Shao, M.; Wang, J. X.; Nilekar, A. U.; Mavrikakis, M.; Valerio, J. A.; Uribe, F. Top. Catal. 2007, 46, 249. doi: 10.1007/s11244-007-9003-x

(21) Shao, M.; Sasaki, K.; Marinkovic, N.; Zhang, L.; Adzic, R. Electrochem. Commun. 2007, 9, 2848. doi: 10.1016/j.elecom.2007.10.009

(22) Zhou, W. P.; Yang, X.; Vukmirovic, M. B.; Koel, B. E.; Jiao, J.; Peng, G.; Mavrikakis, M.; Adzic, R. R. J. Am. Chem. Soc. 2009, 131, 12755. doi: 10.1021/ja9039746

(23) Gong, K.; Su, D.; Adzic, R. R. J. Am. Chem. Soc. 2010, 132, 14364. doi: 10.1021/ja1063873

(24) Sasaki, K.; Naohara, H.; Cai, Y.; Choi, Y. M.; Liu, P.; Vukmirovic, M. B.; Wang, J. X.; Adzic, R. R. Angew. Chem.Int. Ed. 2010, 49, 8602. doi: 10.1002/anie.201004287

(25) Koenigsmann, C.; Santulli, A. C.; Gong, K.; Vukmirovic, M. B.; Zhou, W. P.; Sutter, E.; Wong, S. S.; Adzic, R. R. J. Am.Chem. Soc. 2011, 133, 9783. doi: 10.1021/ja111130t

(26) Adzic, R. R. Electrocatalysis-Us 2012, 3, 163. doi: 10.1007/s12678-012-0112-3

(27) Kuttiyiel, K. A.; Sasaki, K.; Choi, Y.; Su, D.; Liu, P.; Adzic, R. R. Energy Environ. Sci. 2012, 5, 5297. doi: 10.1039/C1EE02067F

(28) Chen, S.; Ferreira, P. J.; Sheng, W.; Yabuuchi, N.; Allard, L. F.; Shao-Horn, Y. J. Am. Chem. Soc. 2008, 130, 13818. doi: 10.1021/ja802513y

(29) Van der Vliet, D. F.; Wang, C.; Li, D.; Paulikas, A. P.; Greeley, J.; Rankin, R. B.; Strmcnik, D.; Tripkovic, D.; Markovic, N. M.; Stamenkovic, V. R. Angew. Chem. Int. Ed. 2012, 51, 3139. doi: 10.1002/ange.201107668

(30) Chi, M.; Wang, C.; Lei, Y.; Wang, G.; Li, D.; More, K. L.; Lupini, A.; Allard, L. F.; Markovic, N. M.; Stamenkovic, V. R. Nat. Commun. 2015, 6, 8925. doi: 10.1038/ncomms9925

(31) Stamenkovic, V. R.; Mun, B. S.; Mayrhofer, K. J.; Ross, P. N.; Markovic, N. M. J. Am. Chem. Soc. 2006, 128, 8813. doi: 10.1021/ja0600476

(32) Stamenkovic, V. R.; Fowler, B.; Mun, B. S.; Wang, G.; Ross, P. N.; Lucas, C. A.; Markovic, N. M. Science 2007, 315, 493. doi: 10.1126/science.1135941

(33) Zhu, H.; Luo, M.; Zhang, S.; Wei, L.; Wang, F.; Wang, Z.; Wei, Y.; Han, K. Int. J. Hydrogen Energy 2013, 38, 3323. doi: 10.1016/j.ijhydene.2012.12.127

(34) Srivastava, R.; Mani, P.; Hahn, N.; Strasser, P. Angew. Chem.Int. Ed. 2007, 46, 8988. doi: 10.1002/anie.200703331

(35) Hasché, F.; Oezaslan, M.; Strasser, P. Chemcatchem 2011, 3, 1805. doi: 10.1002/cctc.201100169

(36) Cui, C.; Ahmadi, M.; Behafarid, F.; Gan, L.; Neumann, M.; Heggen, M.; Cuenya, B. R.; Strasser, P. Faraday Discuss. 2013, 162, 91. doi: 10.1039/C3FD20159G

(37) Yu, Z.; Zhang, J.; Liu, Z.; Ziegelbauer, J. M.; Xin, H.; Dutta, I.; Muller, D. A.; Wagner, F. T. J. Phys. Chem. C 2012, 116, 19877. doi: 10.1021/jp306107t

(38) Snyder, J.; McCue, I.; Livi, K.; Erlebacher, J. J. Am. Chem.Soc. 2012, 134, 8633. doi: 10.1021/ja3019498

(39) Strasser, P.; Kühl, S. Nano Energy 2016. 29, 362. doi: 10.1016/j.nanoen.2016.04.047

(40) Gan, L.; Heggen, M.; O’Malley, R.; Theobald, B.; Strasser, P. Nano Lett. 2013, 13, 1131. doi: 10.1021/nl304488q

(41) Strasser, P.; Koh, S.; Anniyev, T.; Greeley, J.; More, K.; Yu, C. F.; Liu, Z. C.; Kaya, S.; Nordlund, D.; Ogasawara, H.; Toney, M. F.; Nilsson, A. Nat. Chem. 2010, 2, 454. doi: 10.1038/nchem.623

(42) Wang, D.; Zhao, P.; Li, Y. Sci. Rep-Uk 2011, 1, 37. doi: 10.1038/srep00037

(43) Ulrikkeholm, E. T.; Pedersen, A. F.; Vej-Hansen, U. G.; Escudero-Escribano, M.; Stephens, I. E. L.; Friebel, D.; Mehta, A.; Schi?tz, J.; Feidenhansl', R. K.; Nilsson, A.; Chorkendorff, I. Surf. Sci. 2016, 652, 114. doi: 10.1016/j.susc.2016.02.009

(44) Pedersen, A. F.; Ulrikkeholm, E. T.; Escudero-Escribano, M.; Johansson, T. P.; Malacrida, P.; Pedersen, C. M.; Hansen, M. H.; Jensen, K. D.; Rossmeisl, J.; Friebel, D.; Nilsson, A.; Chorkendorff, I. Stephens, I. E. L. Nano Energy 2016, 29, 249. doi: 10.1016/j.nanoen.2016.05.026

(45) Escudero-Escribano, M.; Malacrida, P.; Hansen, M. H.; Vej-Hansen, U. G.; Velazquez-Palenzuela, A.; Tripkovic, V.; Schiotz, J.; Rossmeisl, J.; Stephens, I. E.; Chorkendorff, I. Science 2016, 352, 73. doi: 10.1126/science.aad8892

(46) Velázquez-Palenzuela, A.; Masini, F.; Pedersen, A. F.; Escudero-Escribano, M.; Deiana, D.; Malacrida, P.; Hansen, T. W.; Friebel, D.; Nilsson, A.; Stephens, I. E. L.; Chorkendorff, I. J. Catal. 2015, 328, 297. doi: 10.1016/j.jcat.2014.12.012

(47) Malacrida, P.; Casalongue, H. G.; Masini, F.; Kaya, S.; Hernandez-Fernandez, P.; Deiana, D.; Ogasawara, H.; Stephens, I. E.; Nilsson, A.; Chorkendorff, I. Phys. Chem.Chem. Phys. 2015, 17, 28121. doi: 10.1039/C5CP00283D

(48) Malacrida, P.; Escudero-Escribano, M.; Verdaguer-Casadevall, A.; Stephens, I. E. L.; Chorkendorff, I. J. Mater. Chem. A 2014, 2, 4234. doi: 10.1039/C3TA14574C

(49) Hernandez-Fernandez, P.; Masini, F.; McCarthy, D. N.; Strebel, C. E.; Friebel, D.; Deiana, D.; Malacrida, P.; Nierhoff, A.; Bodin, A.; Wise, A. M.; Nielsen, J. H.; Hansen, T. W.; Nilsson, A. Stephens, I. E. L.; Chorkendorff, I. Nat.Chem. 2014, 6, 732. doi: 10.1038/nchem.2001

(50) Stephens, I. E. L.; Bondarenko, A. S.; Bech, L.; Chorkendorff, I. ChemCatChem 2012, 4, 341. doi: 10.1002/cctc.201100343

(51) Escudero-Escribano, M.; Verdaguer-Casadevall, A.; Malacrida, P.; Gronbjerg, U.; Knudsen, B. P.; Jepsen, A. K.; Rossmeisl, J.; Stephens, I. E.; Chorkendorff, I. J. Am. Chem.Soc. 2012, 134, 16476. doi: 10.1021/ja306348d.

(53) Kongkanand, A.; Mathias, M. F. J. Phys. Chem. Lett. 2016, 7, 1127. doi: 10.1021/acs.jpclett.6b00216

(54) Xia, Y.; Xiong, Y.; Lim, B.; Skrabalak, S. E. Angew. Chem.Int. Ed. 2009, 48, 60. doi: 10.1002/anie.200802248

(55) Gilroy, K. D.; Ruditskiy, A.; Peng, H. C.; Qin, D.; Xia, Y. Chem. Rev 2016, 116, 10414. doi: 10.1021/acs.chemrev.6b00211

(56) Ferrando, R.; Jellinek, J.; Johnston, R. L. Chem. Rev. 2008, 108, 845. doi: 10.1021/cr040090g

(57) Wu, J.; Yang, H. Acc. Chem. Res. 2013, 46, 1848. doi: 10.1021/ar300359w

(58) Perez-Alonso, F. J.; McCarthy, D. N.; Nierhoff, A.; Hernandez-Fernandez, P.; Strebel, C.; Stephens, I. E.; Nielsen, J. H.; Chorkendorff, I. Angew. Chem. Int. Ed. 2012, 51, 4641. doi: 10.1002/anie.201200586

(59) Wang, C.; Daimon, H.; Onodera, T.; Koda, T.; Sun, S.Angew. Chem. Int. Ed. 2008, 47, 3588. doi: 10.1002/anie.200800073

(60) Huang, X. Q.; Zhao, Z. P.; Cao, L.; Chen, Y.; Zhu, E. B.; Lin, Z. Y.; Li, M. F.; Yan, A. M.; Zettl, A.; Wang, Y. M.; Duan, X. F.; Mueller, T.; Huang, Y. Science 2015, 348, 1230. doi: 10.1126/science.aaa8765

(61) N?rskov, J. K.; Rossmeisl, J.; Logadottir, A.; Lindqvist, L.; Kitchin, J. R.; Bligaard, T.; Jónsson, H. J. Phys. Chem. B 2004, 108, 17886. doi: 10.1021/jp047349j

(62) Zhang, J.; Yang, H.; Fang, J.; Zou, S. Nano Lett. 2010, 10, 638. doi: 10.1021/nl903717z

(63) Wu, J.; Gross, A.; Yang, H. Nano Lett. 2011, 11, 798. doi: 10.1021/nl104094p

(64) Greeley, J.; Stephens, I. E.; Bondarenko, A. S.; Johansson, T. P.; Hansen, H. A.; Jaramillo, T. F.; Rossmeisl, J.; Chorkendorff, I.; Norskov, J. K. Nat. Chem. 2009, 1, 552. doi: 10.1038/nchem.367

(65) Yang, S.; Liu, F.; Wu, C.; Yang, S. Small 2016, 12, 4028. doi: 10.1002/smll.201601203

(66) Wang, W.; Lv, F.; Lei, B.; Wan, S.; Luo, M.; Guo, S. Adv.Mater. 2016, 28, 10117. doi: 10.1002/adma.201601909

(67) Kwon, S. G.; Hyeon, T. Small 2011, 7, 2685. doi: 10.1002/smll.201002022

(68) Bu, L.; Ding, J.; Guo, S.; Zhang, X.; Su, D.; Zhu, X.; Yao, J.; Guo, J.; Lu, G.; Huang, X. Adv. Mater. 2015, 27, 7204. doi: 10.1002/adma.201502725

(69) Bu, L.; Guo, S.; Zhang, X.; Shen, X.; Su, D.; Lu, G.; Zhu, X.; Yao, J.; Guo, J.; Huang, X. Nat. Commun. 2016, 7, 11850. doi: 10.1038/ncomms11850

(70) Guo, S.; Li, D.; Zhu, H.; Zhang, S.; Markovic, N. M.; Stamenkovic, V. R.; Sun, S. Angew. Chem. Int. Ed. 2013, 52, 3465. doi: 10.1002/anie.201209871

(71) Guo, S.; Zhang, S.; Su, D.; Sun, S. J. Am. Chem. Soc. 2013, 135, 13879. doi: 10.1021/ja406091p

(72) Jiang, K.; Zhao, D.; Guo, S.; Zhang, X.; Zhu, X.; Guo, J.; Lu, G.; Huang, X. Sci. Adv. 2017, 3, e1601705. doi: 10.1126/sciadv.1601705

(73) Li, M. F.; Zhao, Z. P.; Cheng, T.; Fortunelli, A.; Chen, C. Y.; Yu, R.; Zhang, Q. H.; Gu, L.; Merinov, B. V.; Lin, Z. Y.; Zhu, E. B.; Ted Yu, T.; Jia, Q. Y.; Guo, J. H.; Zhang, L.; Goddard III, W. A.; Huang, Y.; Duan, X. F. Science 2016, 354, 1414. doi: 10.1126/science.aaf9050

(74) Fortunelli, A.; Goddard Iii, W. A.; Sementa, L.; Barcaro, G.; Negreiros, F. R.; Jaramillo-Botero, A. Chem. Sci. 2015, 6, 3915 doi: 10.1039/C5SC00840A

(75) Chen, Z.; Waje, M.; Li, W.; Yan, Y. Angew. Chem. Int. Ed.2007, 119, 4138. doi: 10.1002/anie.200700894

(76) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A.Science 2004, 306, 666. doi: 10.1126/science.1102896

(78) Cheong, W. C.; Liu, C.; Jiang, M.; Duan, H.; Wang, D.; Chen, C.; Li, Y. Nano Res. 2016, 9, 2244. doi: 10.1007/s12274-016-1111-0

(79) Funatsu, A.; Tateishi, H.; Hatakeyama, K.; Fukunaga, Y.; Taniguchi, T.; Koinuma, M.; Matsuura, H.; Matsumoto, Y. Chem. Commun. 2014, 50, 8503. doi: 10.1039/C4CC02527J

(80) Bu, L. Z.; Zhang, N.; Guo, S. J.; Zhang, X.; Li,J.; Yao, J. L.; Wu, T.; Lu, G.; Ma, J. Y.; Su, D.; Huang, X. Q. Science 2016, 354, 1410. doi: 10.1126/science.aah6133

(81) Xia, Y.; Yang, X. Acc. Chem. Res. 2017, 50, 450. doi: 10.1021/acs.accounts.6b00469

(82) Nesselberger, M.; Ashton, S.; Meier, J. C.; Katsounaros, I.; Mayrhofer, K. J.; Arenz, M. J. Am. Chem. Soc. 2011, 133, 17428. doi: 10.1021/ja207016u

(83) Zhang, L.; Roling, L. T.; Wang, X.; Vara, M.; Chi M. F.; Liu, J. Y.; Choi, S.; Park, J. B.; Herron, J. A.; Xie, Z. X.; Mavrikakis, M.; Xia, Y. N. Science 2015, 349, 412. doi: 10.1126/science.aab0801

(84) Skrabalak, S. E.; Au, L.; Li, X.; Xia, Y. Nat. Protoc. 2007, 2, 2182. doi: 10.1038/nprot.2007.326

(85) Xia, X.; Xie, S.; Liu, M.; Peng, H. C.; Lu, N.; Wang, J.; Kim, M. J.; Xia, Y. Proc. Natl. Acad. Sci. USA 2013, 110, 6669. doi: 10.1073/pnas.1222109110

(86) Wang, X.; Figueroa-Cosme, L.; Yang, X.; Luo, M.; Liu, J.; Xie, Z.; Xia, Y. Nano Lett. 2016, 16, 1467. doi: 10.1021/acs.nanolett.5b05140

(87) Chen, C.; Kang, Y. J.; Huo, Z. Y.; Zhu, Z. W.; Huang, W. Y.; Xin, H. L.; Snyder, J. D.; Li, D. G.; Herron, J. A.; Mavrikakis, M.; Chi, M. F.; More, K. L.; Li,Y. D.; Markovic, N. M.; Somorjai, G. A.; Yang, P. D.; Stamenkovic, V. R. Science 2014, 343, 1339. doi: 10.1126/science.1249061

(88) Ding, J.; Bu, L.; Guo, S.; Zhao, Z.; Zhu, E.; Huang, Y.; Huang, X. Nano Lett. 2016, 16, 2762. doi: 10.1021/acs.nanolett.6b00471

(89) Zhu, C.; Du, D.; Eychmuller, A.; Lin, Y. Chem. Rev. 2015, 115, 8896. doi: 10.1021/acs.chemrev.5b00255

(90) Erlebacher, J.; Aziz, M. J.; Karma, A.; Dimitrov, N.; Sieradzki, K. Nature 2001, 410, 450. doi: 10.1038/35068529

(91) Oezaslan, M.; Hasché, F.; Strasser, P. J. Phys. Chem. Lett.2013, 4, 3273. doi: 10.1021/jz4014135

(92) Oezaslan, M.; Heggen, M.; Strasser, P. J. Am. Chem. Soc.2012, 134, 514. doi: 10.1021/ja2088162

(93) Han, B.; Carlton, C. E.; Kongkanand, A.; Kukreja, R. S.; Theobald, B. R.; Gan, L.; O'Malley, R.; Strasser, P.; Wagner, F. T.; Shao-Horn, Y. Energy Environ. Sci. 2015, 8, 258. doi: 10.1039/C4EE02144D

(94) Snyder, J.; Livi, K.; Erlebacher, J. Adv. Funct. Mater. 2013, 23, 5494. doi: 10.1038/nmat2878

(95) Snyder, J.; Fujita, T.; Chen, M. W.; Erlebacher, J. Nat. Mater. 2010, 9, 904. doi: 10.1038/nmat2878

(96) Alia, S. M.; Zhang, G.; Kisailus, D.; Li, D.; Gu, S.; Jensen, K.; Yan, Y. Adv. Funct. Mater. 2010, 20, 3742. doi: 10.1002/adfm.201001035

(97) Todoroki, N.; Kato, T.; Hayashi, T.; Takahashi, S.; Wadayama, T. ACS Catal. 2015, 5, doi: 2209-2212.10.1021/acscatal.5b00065

(98) Lim, B.; Xia, Y. Angew. Chem. Int. Ed. 2011, 50, 76. doi: 10.1002/anie.201002024

(99) Lim, B.; Jiang, M.; Camargo, P. H.; Cho, E. C.; Tao, J.; Lu, X.; Zhu, Y.; Xia, Y. Science 2009, 324, 1302. doi: 10.1126/science.1170377

(100) Huang, X.; Zhu, E.; Chen, Y.; Li, Y.; Chiu, C. Y.; Xu, Y.; Lin, Z.; Duan, X.; Huang, Y. Adv. Mater. 2013, 25, 2974. doi: 10.1002/adma.201205315

(101) Sun, S.; Zhang, G.; Geng, D.; Chen, Y.; Li, R.; Cai, M.; Sun, X. Angew. Chem. Int. Ed. 2011, 50, 422. doi: 10.1002/ange.201004631