Currently, the major drawbacks to efficiently drive kinetic processes for electrochemical energy conversion reactions are poor catalytic performance and low utilization rates of electrocatalytic materials. There is an imminent need to improve catalytic activity and stability by using low dosages of catalysts, and adhering to design simplicity and green technology. To achieve these objectives, various electrocatalytic materials with different structures and chemical compositions including noble metals [1, 2], low-cost earth-abundant transition metals [3-5], and even metal-free [6-8] materials have been broadly explored as alternative electrocatalysts for various electrochemical applications. Platinum-based materials have played a particularly significant role in this area due to their high activities toward the oxygen reduction reaction (ORR) in fuel cells [9, 10] and the hydrogen evolution reaction (HER) in electrolyzers [11, 12]. However, for efficient electrochemical metal-air batteries and water splitting, the oxygen evolution reaction (OER) also requires a highly active electrocatalyst to lower the electrochemical overpotential. Pt-based electrocatalysts are very efficient for both ORR and HER, and are capable of achieving significant currents close to the thermodynamic potential. However, their relatively low OER activities render them ineffective in practical applications, restricting their use in metal-air batteries and overall water splitting by electrolysis.

Therefore, other expensive alternatives, such as IrO₂ and RuO₂ with moderate activities, are typically employed as OER catalysts. Even then, the durability of IrO₂ and RuO₂ is still insufficient to endure beyond many time-dependent electrochemical cycles. Thus, various methods have been explored to improve the OER performance of Pt-based catalysts, including the dispersion of Pt-crystals on pure carbon supports, high Pt-loading or by alloying with co-active transition metals [13, 14]. Despite all these efforts, the inability to shield the Pt atoms from direct contact with the electrolyte, or electrochemically induced oxidization often leads to the formation of inactive oxide layers, resulting in significant decline in the OER performance [14]. Despite these challenges, designing a highly active Pt-based OER catalyst with low Pt-loading and enhanced stability via a facile method is essential, if overall water splitting (simultaneous HER/OER) is to be achieved with only Pt-based materials at both, the cathode and the anode, in electrolyzers. Typically, the poor performance of Pt toward OER is largely attributed to the inevitable rapid formation of platinum(Ⅱ) oxide and platinum(Ⅳ) oxide (PtO and PtO₂, respectively), which usually occur at the exposed outermost surface layer covering the inter Pt-layers [2, 14, 15]. Such surface Pt-oxide species, which are in direct contact with the electrolyte, are inactive toward OER thereby hindering maximum Pt utilization and lowering the overall water splitting efficiency. Consequently, the formation of such oxides limits the potential to enhance the OER activity and electrochemical stability of Pt-based catalysts, and must be circumvented. Yet, this remains a very challenging task due to the high reactivity of Pt in alkaline media. To avoid the quick oxidization process and achieve high stability, ultrathin carbon layers have been applied to effectively anchor Pt-metal nanoparticles onto a support matrix or in a core-shell system. Such a strategy has proven to be very efficient for stabilizing and shielding Pt particles against quick oxidization, particle migration and aggregation, thereby leading to improved electrocatalytic performance towards ORR and HER [2, 14, 16, 17]. Additionally, both the encapsulating porous carbon layer and support matrix also increase the electrical conductivity and enhance the diffusion of ionic species. Nonetheless, such strategies are still relatively unexploited for enhancing the OER activity and stability of Pt catalysts. In order to further enhance the electrocatalytic OER properties, the use of a highly porous, functionalized carbon matrix, such as a metal organic framework (MOF) is preferable.

MOFs have been recognized as suitable platforms for the preparation of structure-tailored, high surface area nanocarbon frameworks owing to their high flexibility, abundant pores, versatile surface geometry, and high specific surface area [18-23]. Indeed, MOFs are currently being investigated for various emerging applications, including ORR, HER and OER for energy conversion, as well as for supercapacitors and metal-ion/metal-air batteries [18-21]. The direct pyrolysis of MOFs can yield both amorphous and graphitic carbon frameworks and/or an integrated quasi-amorphous carbon structure with graphitic edges. Thus, MOFs are suitable templates for the fabrication of nanoporous carbon materials with high specific surface area including carbon nanotubes, carbon nanorods, graphene, and metal-modified nanocarbon-based materials with immense potential for multifunctional electrochemical applications [18, 19, 21, 22]. Among the sub-classes of MOFs, MOF-253 exhibits a high specific surface area and high porosity, and can be effectively utilized as a suitable template for an even distribution and stabilization of metal species. For such materials, the carbonization of the organic linkers can create porous carbon networks, leading to the formation of catalysts with high specific surface area, uniform distribution of active species and high stability.

In this work, we present the design and facile synthesis of monodispersed single nanocrystal Pt confined by an ultrathin layer of a N-doped carbon framework (Pt@N/C) that addresses both the OER activity and durability issues. Thus, by using MOF-253 as a carbon template, a well-dispersed and highly stabilized Pt single nanocrystal encapsulated in a N-doped carbon framework is synthesized. The encapsulating carbon layer serves as an effective protective sheath for the Pt nanocrystals against electrolyte/electrochemically induced oxidation. This enhances the stability of the Pt core against particle migration and aggregation, as well as improves the conductivity. The carbon layer also provides contact and porous channels for effective diffusion and transport of ionic species. The resulting Pt-N-C framework with highly stabilized and dispersed Pt single nanocrystals and low Pt-loading gives rise to superior OER activity and stability over the commercial counterpart (Pt/C, 20 wt%), as well as the state-of-the-art IrO₂ catalyst. The synthesis route to the monocrystalline Pt-embedded N-doped C (Pt@N/C) is depicted in Fig. 1. The structure of MOF-253 is composed of one-dimensional chains of hydroxide-bridged octahedrally coordinated Al-cations (Al³⁺) interconnected by 2, 2′-bipyridine-5, 5'-dicarboxylic acid (bpydc^2‒) ligands, resulting in the formation of three-dimensional frameworks with rhombic channels. Typically, MOF-253 is first synthesized and then used as a template for the support carbon framework.

In a typical synthesis, 400 mg of the as-prepared MOF-253 (details provided in Supporting Information) was dispersed with a certain amount of H₂PtCl₆·6H₂O (0.0211, 0.0444, and 0.1 mg) in acetonitrile (15 mL). The obtained solution was then heated at 85 ℃ in a Teflon-capped autoclave for 72 h. The formed solids were collected by centrifugation and then immersed in acetonitrile (15 mL) for 3 days. The solvent was replaced with fresh acetonitrile every 24 h. The products were then collected by filtration and dried at 150 ℃ for 12 h in vacuum. The obtained Pt@MOF-253 solids were then heat-treated at 900 ℃ at a ramp rate of 1 ℃ min^-1 for 4 h in Ar atmosphere. Finally, the samples were pickled with NaOH solution at 70 ℃ for 6 h to remove inactive Al-metal species and residues, followed by thorough washing in deionized water until the resultant solution had a neutral pH, before finally being dried under vacuum at 80-110 ℃. For simplicity, the as-obtained catalysts were labelled as Pt@N/C-x, where x represents 2%, 5%, and 10% of Pt-loading in the catalyst.

A JEM-2010FEF high-resolution electron microscope operated at 20 kV was used to obtain the TEM images. X-ray diffraction (XRD) data were recorded on a Rigaku D/MAX-RB diffractometer with monochromatized Cu-K_α radiation, operated at 50 mA and 40 kV. Surface features and morphologies were examined using a JSM-7100F field-emission scanning electron microscope (FE-SEM) installed with an energy-dispersive analyzer and operated at 10 kV. Raman spectra were recorded with a LabRAM Aramis Raman equipment using Ar-ion laser at λ = 632.8 nm. A VG Multilab 2000 instrument was used to acquire the X-ray photoelectron spectroscopy (XPS) data. Elemental composition was performed using both a German elemental analysis instrument (GmbH EL Cube Vario Elemental Analyzer) and an inductively coupled plasma atomic emission spectrometer (ICP-AES, Optima 4300DV).

A catalyst-ink was prepared for each sample by ultrasonically dispersing 5.0 mg of catalyst in a solution of 20 µL Nafion (5 wt%, DuPont)/UHP water/isopropanol (v/v ≈ 3/7) and optimized for homogeneous dispersion. The working electrode was prepared by loading the catalyst ink (10 µL) onto a glassy carbon electrode (0.196 cm²) and dried under constant rotation of 600 rpm under ambient conditions. Commercial Pt/C (20 wt%, 20 µg Pt cm^-2) and IrO₂ were used as benchmark catalysts. A Hg/HgO and graphite plate were used as reference and counter electrodes, respectively. All the experimentally measured potentials were referenced to the standard reversible hydrogen electrode (RHE) according to E_RHE = E_Hg/HgO + 0.059pH + E⁰. All electrochemical properties were measured using a three-electrode electrochemical system (CHI660E) at room temperature. The oxygen evolution reaction (OER) properties were characterized in 1.0 M KOH at a scan rate of 5 mVs^-1, and the electrode potential for water oxidation was evaluated at current density of 10 mA cm^-2 (E_j = 10).

SEM images of the as-synthesized MOF-253, before and after pyrolysis, are shown in Fig. S1. It can be observed that the rod-like morphology of the pristine MOF-253 appears to be retained even after carbonization. Fig. 2(a) and (b) show the TEM images of Pt@N/C-10, revealing monodispersed Pt single nanocrystals confined in a N-doped-C framework. The corresponding low-magnification TEM images are presented in Fig. S2. No fingerprint of metallic Al or its oxides/alloys is observed, which is also confirmed by both XPS and ICP-AES measurements, as shown in Fig. 3(c) and Table S1, respectively. This indicates the complete removal of the inactive Al species. Fast Fourier-transform (FFT) images taken from the points marked as 1-4 further show that the monodispersed Pt-nanocrystals retained their periodic lattice structures and crystallinity.

The HR-TEM image (Fig. 2(b)) further reveals that such Pt nano-single crystals are well encapsulated in a N-doped carbon framework, ~0.51 nm in thickness, with a lattice spacing of 0.22 nm for the Pt (111) facet. Fig. 2(c) shows that the average size of the monodispersed Pt single nanocrystals is 6.7 nm, which is suitable for effective carbon layer encapsulation during pyrolysis. The formation of monodispersed, small, single nanocrystals attests to the existence and effectiveness of confinement in a thin carbon layer. This is also supported by the fact that untethered Pt particles on a carbon support (e.g. commercial Pt/C) have a high tendency to freely aggregate into larger particles during thermal treatments, which are less active catalysts [2, 16, 17]. The high degree of clarity seen in the crystal lattices is also an indication of an ultrathin carbon layer encapsulation, because thicker layers or many carbon layers would normally mask the transparency of the Pt-lattices and limit accessibility of the electrolyte to the active Pt underneath [2, 16, 17]. In a series of control experiments, catalyst samples including Pt-free N-doped carbon (N/C), N-free Pt/C and similar Pt@N/C-x catalysts with different Pt and N loadings were also evaluated, to understand the structure-property correlation and its effect on catalytic performance. As shown in Figs. S3 and S4, the average particle size of Pt@N/C-10 (~6.7 nm) is smaller than that of Pt@N/C-2 (~9.6 nm) and Pt@N/C-5 (~8.5 nm), based on the corresponding TEM analyses.

XRD patterns of the Pt@N/C-based samples (Fig. 3(a)) appear to be similar, exhibiting increased peak intensities as the Pt content increases, suggesting that they have similar crystalline properties (JCPDS: 01-087-0640). Meanwhile, the XRD pattern of the carbonized MOF-253 without Pt (Pt-free N/C) matches well with that of a typical N-doped carbon framework [8, 18, 19, 21], confirming the complete transformation of the pristine MOF-253 into an N-doped carbon (N/C) structure. Typically, the diffraction peaks at 39.8°, 46.4°, 67.7°, 81.7°, and 86.0° are attributed to the Pt (111), (200), (220), (311) and (222) crystal planes, respectively. This indicates a typical crystalline Pt phase with a face-centered-cubic (fcc) structure. It is worth noting that the lattice indices of Pt@N/C-10 also match well with that of Pt/C (Fig. S5), suggesting that the N-doping and/or carbon encapsulation have no effect on the lattice structure of the active Pt core.

Raman spectra were recorded in order to study the configuration and structural defects of the synthesized catalysts. As shown in Fig. 3(b), all samples display two main peaks at ~1349.95 and ~1597.95 cm^-1, which are assigned to the D and G bands, respectively. The D-band is attributed to lattice defects, while the G-band is ascribed to sp²-hybridized carbon [7, 13, 16]. The ratio of the band intensities (I_D/I_G) is found to be ~0.979, 1.044, 1.050 and 1.075 for Pt@N/C-2, Pt@N/C-5, Pt@N/C-10 and N/C, respectively, signifying that nitrogen doping and integration of metallic Pt induced an increase in the number of carbon lattice defects [17, 20]. In comparison to Pt@N/C-2 and Pt@N/C-5, the higher N and Pt loading of Pt@N/C-10 is considered significant for enhancing overall catalytic activity.

XPS was used to analyze the surface chemical composition, as well as the binding energies associated with the various atomic species. The survey scan (Fig. 3(c)) shows the presence of all elemental components, including Pt, C, N and O. Metallic-Al or its alloys were not detected, further confirming the complete removal of Al residues from the samples by the pickling process. Removal of the Al species leads to the creation of more porous structures, which is known to improve diffusion properties and catalytic activity. The high resolution N1s peak is resolved into four subpeaks, assignable to pyridinic-N (398.34 eV), graphitic-N (400.89 V), Pyrrolic-N (399.33 eV) and oxidized-N (403.46 eV) species, respectively (Fig. 3(d)). Graphitic-N is considered to improve the diffusion-limited properties, while the pyridinic-N, as an active site for OER, raises the onset potential, electrical conductivity and surface wettability [7, 21, 25, 26, 27]. Several studies have also reported that the pyridinic-N site can accept electrons from neighboring C atoms, making possible the adsorption of water oxidation intermediates (OH^-, OOH^-) as the rate defining steps for OER in alkaline solutions [25, 26]. Metal-induced reactions during thermal treatments can avail high structural defects as suitable sites for N-doping. As displayed in Table S4, Pt@N/C-10 shows the highest proportion of combined pyridinic-N and graphitic-N species, which, coupled with the optimal Pt-loading, is expected to enhance catalytic activity over all the other catalyst samples.

The above Raman spectra also show the presence of defects in the carbon lattices for all samples, which is beneficial for further enhancement of catalytic properties. The deconvoluted Pt 4ƒ peak (Fig. 3(e)) shows four subpeaks constituting two pairs of doublets, with weak peak intensities. The two peaks at 71.28 and 74.58 eV are attributed to Pt 4ƒ_7/2 and Pt 4ƒ_5/2 excitations of metallic Pt, respectively, whereas the peaks at 72.18 and 75.58 eV are ascribed to oxidized Pt [2, 13]. The deconvoluted C 1s peak (Fig. 3(c)) displays subpeaks associated with sp²-hybridized C=C (~284.5 eV), C-N (~285.2 eV), C=O (~286.3 eV) and O-C=O (~288.9 eV) species [20, 21]. The corresponding deconvoluted C, N and Pt subpeaks for Pt@N/C-2 and Pt@N/C-5 are shown in Fig. S6 and S7, respectively.

Given the importance and urgent need to enhance the OER performance of Pt-based materials, the electrocatalytic activity of all catalyst samples was probed by using a standard three-electrode cell setup in 1.0 M KOH solutions at a scan rate of 5 mV s^-1. Linear sweep voltammetry (LSV) was employed to generate the polarization curves (Fig. 4(a)) for all catalyst samples after iR corrections. The catalyst sample with a Pt-loading of 6.1 wt% (Pt@N/C-10) is found to exhibit the best OER activity. To achieve a current density of 10 mA cm^-2, the catalysts require a potential as low as ~1.564, ~1.544, and ~1.528 V for Pt@N/C-2, Pt@N/C-5 and Pt@N/C-10, respectively. It is worth noting that the OER activity of the best sample (Pt@N/C-10, with an onset overpotential, η ≈ 298 mV) is about 55 mV lower than that of the benchmark catalyst (IrO₂, η ~ 353 mV), and is clearly superior to the N-free carbon commercial catalyst (Pt/C), as well as the Pt-free N-doped carbon (N/C) catalyst, as shown in Fig. S8.

The OER kinetic activity was also probed by analyzing the Tafel slopes. As shown in Fig. 4(b), the Tafel slopes of Pt@N/C-2, Pt@N/C-5, and Pt@N/C-10 are 59.8, 59.4 and 55.1 mV dec^-1, respectively, which are also smaller than that of the benchmark catalysts (IrO₂ ~64.2 mV dec^-1, and commercial Pt/C ~124.3 mV dec^-1), further demonstrating the high-performance electrocatalytic ability of Pt@N/C toward OER. Interestingly, as shown in Fig. 4(c), the catalytic activity of Pt@N/C is superior to the noble metal-based catalysts, as well as transition metal-based OER catalysts (Table S5).

An important consideration for the application of such Pt-based materials is their time-dependent electrochemical durability. As shown in the inset of Fig. 4(d), Pt@N/C-10 demonstrates good stability even after more than 20 h of continuous electrochemical scanning. However, a slight gradual drop in current is observed for the initial 10 h, which can be attributed to the inevitable surface oxidation due to exposure to the strongly oxidizing environment of the electrolyte and also, due to the electrochemical current during the scan process, consistent with the LSV scan results. However, the catalyst gradually stabilizes and then remains fairly stable thereafter, indicating that the unavoidable surface oxidation process is not pervasive over the catalyst structure to significantly affect its electrochemical activity. Interestingly, the catalyst is also found to maintain its high activity after being subjected to an accelerated degradation test (ADT) for 2000 cycles (Fig. 4(d)) and is, additionally, found to be superior to the fresh benchmark catalyst (Pt/C), further indicating that Pt@N/C-10 can sustain a current density of 10 mA cm^-2 with only a small potential degradation of ~19.4 mV. However, the benchmark IrO₂ deteriorates by ~23.3 mV at the same rate at 10 mA cm^-2 (Fig. S9).

To further evaluate the stability of the catalyst itself, we conducted both TEM (Fig. S10) and XPS (Fig. S11) investigations of Pt@N/C-10 after an accelerated durability test (ADT). The TEM observations of the catalyst structure indicate that the Pt nanocrystals continue to be well-dispersed in the catalyst, without any significant migration, or aggregation, owing to the robust thin-carbon entrapment. However, a slight fading of the Pt-lattices is observed due to the inevitable oxidization. Also, there is no significant change in position of the Pt 4f peaks or appearance of any new phase peaks, according to the XPS survey scan, which suggests that the chemical structure is very stable. However, the intensity of the Pt 4f peaks are reduced considerably (almost to the point of disappearance), with a corresponding slight increase in the intensity of the oxygen and carbon peaks. This can be attributed to the following reasons. (1) Long-term exposure of the catalyst to the concentrated electrolyte inevitably induces oxidation of the Pt particles, thereby increasing the content of oxidized-Pt species, and (2) the long exposure to concentrated electrolyte could also minimally leach out Pt species from the catalyst, leading to reduced content of metallic-Pt (Pt active site). The electrochemically induced oxidation and reduced content of metallic-Pt can be established from the XPS near-surface scan (Fig. S9), which displays a slight increase in the intensity of the oxygen peak after electrochemical ADT. Regardless of any of these unavoidable phenomena, the catalyst clearly demonstrates high chemical stability, without the formation of any new species that could significantly affect its structure or overall catalytic performance.

Generally, the poor OER performance of Pt/C is mainly attributed to the inevitable and rapid oxidization of the Pt nanoparticles on the carbon support. Surface-dispersed Pt nanoparticles are highly susceptible to oxidation, particle migration and coalescence. Such quickly oxidized Pt species are inactive toward OER, while the migration and agglomeration of Pt nanoparticles result in Pt-carbon detachment which, in turn, promotes the formation of larger particles, leading to a decrease in surface area and number of active sites. This explains why commercial Pt/C is a poor OER catalyst. However, in the protective structure of Pt@N/C-10, the activity originates in activated Pt sites that selectively bind with OH^- species. According to previous research [14, 28, 29], for oxygen evolution on Pt in alkaline media, the formation of Pt-OH_ads phase is promoted by the nucleophilic attack of hydroxide ions on the Pt-metal surface, followed by a fast electron transfer to the metastable configuration where OH is chemisorbed on the 'activated' Pt-site. The hydroxyl group is then selectively adsorbed on the Pt sublayer to form Pt-OH_ads as an active phase. Here, the N-doped carbon, which is often employed as a modified carbon support [30], robustly confines the Pt single nanocrystal growth, migration and dissolution. It also provides auxiliary defect-based active sites to impart high activity to the integrated Pt-N-C catalyst framework without blocking access to the interior active sites on Pt due to its interconnected pore structures, which promote efficient diffusion. The significant oxidation of the carbon support at high OER potentials can also be alleviated due to the protection offered by the graphitic outer shell. Thus, the thin carbon layer encapsulation is sufficient to deter the Pt nanocrystal migration and agglomeration, and also reduces the rate of nanocrystal oxidization and helps to maintain a high surface area.

The N-doping in the carbon framework also confers catalytic advantages to Pt@N/C-10 over commercial Pt/C. The defects presented in the N/C framework aid in the catalytic activity while the high porosity, from the pristine MOF-253 and subsequent removal of Al, allows for efficient diffusion and transport of ionic species. In addition, the higher electronegativity of oxygen, as compared to carbon, and the strong electron affinity of doped-N atoms lead to the creation of positively charged neighboring C-atoms as active sites [16, 18, 21]. The improved OER activity of Pt@N/C could also be attributed to the enhanced adsorption of OH^- and efficient transport of intermediate oxygen-containing products such as peroxides (O₂^2-) and superoxides (O^2-). Furthermore, previous reports show that the Pt/OH^- surface binding energy is lowered by the high positive potential at the electrode surface, while the high pH (14) of the electrolyte lowers the electrochemical potential [15, 16]. This can favor the adsorption of OH^-, as well as the activation of the Pt surface, followed by transfer of electrons and subsequent formation of surface Pt-OH species [15, 28, 29]. The surface coupling and deprotonation of the adsorbed hydroxyl groups result in the efficient evolution of oxygen.

The conductivity of the catalysts, which is important to their application, was also verified. As shown in the Nyquist plots (Fig. S12), Pt@N/C-10 exhibits a lower charge transfer resistance (~140.95 Ω) than Pt/C (~302.68 Ω). This indicates that the catalytic activity of Pt@N/C-10 is further improved by the optimal N-doping [31‒33] and the thin carbon layer confinement of the Pt core, enhancing the charge transfer and electrical conductivity properties in comparison to those of Pt/C.

To summarize, our study suggests that Pt@N/C can be facilely synthesized and demonstrates that it can be used as a highly active electrocatalyst with enhanced kinetic activity and stability toward the oxygen evolution reaction (OER). In contrast to commercial catalysts such as IrO₂ and Pt/C, it is clearly established that monodispersed Pt single nanocrystals (with sizes ~6.7 nm) confined in a thin carbon layer (~0.51 nm thick) constitute a very efficient catalyst which exhibits excellent OER performance (~1.528 V@10 mA cm^-2), with minimal degradation of the potential even after 2000 cycles and good potential retention. This simple confinement technique allows for the preparation of electrocatalysts with greatly reduced Pt loading, that can potentially exhibit superior OER performance in comparison to expensive and relatively unstable commercial counterparts (Pt/C 20 wt% and IrO₂). This work is expected to inspire the redesign of Pt-based materials to be used as efficient electrocatalysts for OER, while lowering the cost of using different electrode materials for hydrogen production via water splitting.

We gratefully acknowledge the National Natural Science Foundation of China (51672204), and the National Key Research and Development Program of China (2016YFA0202603) for the funding support.