Charge localized atomic Moδ+ site boosting hydrogen evolution in alkaline solution

Electrochemical water splitting has drawn tremendous interest for the scalable and sustainable conversion of renewable electricity to clear hydrogen fuel and chemicals. However, the sluggishly kinetics of water dissociation step in alkaline solutions restrict severely the application of hydrogen evolution reaction (HER). Here, we designed and prepared cobalt layers with nitrogen modied atomically dispersed Mo sites (N-Mo/Co SAA) to boost the activity of HER. Density functional theory (DFT) calculations demonstrated that the N can induce the asymmetry charge localization of Mo δ+ to facilitate the water dissociation. The energy barriers of water dissociation reduced from 0.48 to 0.35 eV by the charge localized Mo δ+ site. High resolution transmission electron microscope (HRTEM) and synchrotron X-ray absorption spectroscopy (XAS) measurements conrmed the structure of N modied atomically dispersed Mo δ+ . Ambient pressure X-ray photoelectron spectroscopy (AP-XPS) measurements assessed the atomically dispersed Mo δ+ site is the active site for water dissociation. Thus, the obtained N-Mo/Co catalyst exhibits record activity with 12 mV overpotential to achieve the current density of 10 mA cm − 2 and Tafel slope of 31 mV dec − 1 in alkaline media, which are superior to 32 mV overpotential for 10 mA cm − 2 and 38 mV dec − 1 Tafel slope on best commercial 20 wt% Pt/C sample in the same condition. This design strategy provided a new pathway to boost the activity of single atom alloy (SAA) by regulating the charge localization of the active site precisely at the atomic-level.


Introduction
Hydrogen (H 2 ), as a carbon-free and high density energy source, is of signi cant importance in the sustainable energy systems and chemical industry. [1][2][3] Among various H 2 production methods, electrochemical water splitting in alkaline media has been considered as the most feasible approach for high-purity H 2 generation in industry. 4,5 However, the sluggish cathodic H 2 evolution reaction (HER) in alkaline media, 6,7 whose reaction rate is often two or three orders of magnitude lower than that in acidic electrolyte, severely limits the e ciency of water electrolysis. 8,9 Although Pt and Pt-based materials remain the benchmark of HER electrocatalysts, [10][11][12][13][14] their high cost and scarcity greatly hinder their practical applications. 15 Therefore, developing highly e cient and inexpensively alkaline HER catalyst is eagerly. [16][17][18][19] In alkaline solution, hydrogen originate from the water dissociation, 4,20,21 which usually consume tremendously energy and restrict the application of HER. 18,22,23 It has been proved that single atom alloys (SAA) containing atomic water activation component can effectively accelerate the kinetics of alkaline HER. 20,24−26 For example, the Co/Ru single atom alloy (SAA) was reported that the single Co atom reduced the energy barrier of water dissociation and produced hydrogen intermediates that adsorbed on the neighbor Ru sites to generate molecular hydrogen. 20 As a result, the obtained SAA showed a highly improved alkaline HER performance. Similarly, the Ni/Pt SAA exhibited enhanced alkaline HER activity through enhanced OH adsorption on single Ni atom and optimized H adsorption on Pt, respectively. 25 Although this method has enhanced the activity of Ru and Pt based materials in alkaline media, their activity cannot be continuously adjusted by simply increase the amount of single atoms because the single atom generally show its intrinsic property in noble metal host. 25,27−29 The weak interaction between single atom and noble metal host cannot obviously regulate the electronic state of the single atom site, which restrict the further improvement on performance and practical applications.
In this work, we developed a facile route to controllably modify the electronic structure of Mo single sites to boost the alkaline HER activity through nitrogen doping on non-noble metal Co host. Density functional theory (DFT) calculations shown that the asymmetry charge localization of Mo δ+ at N-Mo/Co was induced by doping N. The energy barriers of water dissociation reduced from 0.48 to 0.35 eV at the charge localized Mo δ+ site. The HRTEM, SAC-HAADF-STEM, and XAS measurement con rmed the nitrogen modi ed atomically dispersed Mo in Co layers to form the charge localized Mo δ+ site. The AP-XPS was employed to con rm the N lead the asymmetric charge localization of Mo δ+ active site in Mo/Co SAA accelerated water dissociation. Compared with bare Mo/Co SAA, the energy barriers of water dissociation reduced from 0.48 to 0.35 eV. Thus, the N-Mo/Co catalyst exhibited a high activity with only 12 mV overpotential to achieve the current density of 10 mA cm − 2 and Tafel slope of 31 mV dec − 1 in alkaline solution. The design strategy greatly boost water dissociation process and decrease the overpotential of HER in alkaline by precisely regulating the charge localization of the active site at the atomic-level.

Results And Discussion
To investigate the relationship between the electronic structure of active site and water dissociation, we built the Mo/Co and N-Mo/Co SAAs slab model, and calculated the charge density difference, the Bader charge and the energy barrier of water dissociation on catalyst surface ( Fig. 1, Supplementary Fig. 1). The charge density difference and Bader charge analyses results show that 0.64 electrons transfer from single atom Mo to substrate in N-Mo/Co, which suggests that an asymmetric charge localization is formed on Mo (Fig. 1a). 9,30,31 The energy barriers of water dissociation on the Co(001), Mo/Co SAA and N-Mo/Co were 0.62, 0.44 and 0.35 eV (Fig. 1b, Supplementary Figs. 1 and 2) respectively, indicating that the asymmetric charge localization of Mo in the N-Mo/Co SAA can improve the water dissociation activity. [32][33][34][35][36] The water transition state Bader charge analyses performed to prove charge localization Mo δ+ can effectively boost water dissociation (Fig. 1c). From the adsorbed water DOS on Mo/Co and N-Mo/Co, we found that the N-Mo/Co leads to the downshift of Mo DOS, which bene t to the activation of water molecule by delocalizing the orbital electron density of the water molecule (Fig. 1d). We also considered the HER activity on Co(101) surface and obtained the same results, proving the charge localization Mo δ+ in N-Mo/Co is the key center for water dissociation to promote the kinetics of HER (Supplementary

Figs. 3, 4 and 5).
Based on these results, we synthesized Mo/Co SAA and N-Mo/Co by electrochemical reduction precursor in alkaline (Supplementary Figs. 6,7,8,9 and 10). X-ray diffraction (XRD) patterns clearly showed that the Mo/Co and N-Mo/Co are the hexagonal close-packed cobalt structure (Fig. 2a, Supplementary Figs. 11 and 12). From the HRTEM image (Fig. 2b), the interplanar spacing of the main surface is 0.22 nm, which belongs to the Co(001) surface. Spherical aberration corrected high angle annular darkeld scanning TEM (SAC-HAADF-STEM) image (Fig. 2c) showed that the Mo, circled in red, is singleatomic dispersion in N-Mo/Co. In addition, the intensity pro les along X-Y in the HAADF-STEM con rmed the atomically dispersion of Mo in N-Mo/Co ( Fig. 2b and 2d). The energy dispersive spectrometer (EDS) mapping images (Fig. 2e) and the line-scan pro les ( Supplementary Fig. 13) reveal the Co, Mo and N elements are homogenously distributed in N-Mo/Co, reveal N has been uniformly doped into the Mo/Co SAA. The electron energy loss spectroscopy (EELS) displays that there is no obvious oxygen in N-Mo/Co ( Supplementary Fig. 14). Inductively coupled plasma optical emission spectrometer (ICP-OES) measurement reveals that the concentration of Mo is about 2% doped in N-Mo/Co (Table 1).
To study the electronic structure and local structure of N-Mo/Co, XAS and X-ray photoelectron spectroscopy (XPS) are performed. The K-edges energy of Mo in Mo/Co and N-Mo/Co located between Mo foil and MoO 3 (Fig. 3a), proving the charge localization states of Mo contained in N-Mo/Co. The local structures of Mo in Mo/Co and N-Mo/Co are obtained by extended X-ray absorption ne structure (EXAFS) spectrometer. In the Fig. 3b, Mo/Co SAA shows a coordination peak at 2.18 Å corresponding to the crystal structure of Co. After the N doped into the Mo/Co SAA, a new peak located at 1.56 Å arises, indicating the N-Mo bond formed. 37 Figure 3c shows the Co L-edge spectra of Co, Mo/Co and N-Mo/Co, the two peaks located at 778 eV and 795 eV corresponds to the electron transition from Co 2P 3/2 and 2P 1/2 to 3d, respectively. Compared with Co, the binding energy of Co in N-Mo/Co displays a positive shift of about 0.2 eV to higher energy region due to the N doping. 38 Meanwhile, similar shift of Co was also found in the Co K-edge of N-Mo/Co ( Supplementary Fig. 15). For the sample of N-Mo/Co, the peaks at 398.3 and 399.9 eV can be attribute to Mo 3p 3/2 and N ( Fig. 3d), 26 respectively, con rming that the nitrogen was doped in the Mo/Co SAA surface to form the N-Mo/Co ( Supplementary Fig. 16).
The ultraviolet photoelectron spectrometer (UPS) ( Supplementary Fig. 17) indicates that the d band center of N-Mo/Co shifts away from the Fermi level after N doped in Mo/Co, suggesting that the interaction of Mo and N makes a signi cant contribution to valence band structure of N-Mo/Co, which is consistent with DOS analysis (Fig. 1d). Overall, all these results demonstrate that the atomically dispersed charge localization site of Mo δ+ has been built in N-Mo/Co SAA.
The AP-XPS was performed to con rm the active sites of water dissociation on N-Mo/Co. 2  These results further con rm that atomically dispersed asymmetry charge localization site of Mo δ+ is the active site of water dissociation.
To verify the catalytic activity of Mo sites in N-Mo/Co, the electrocatalytic performance for HER is evaluated by using a three-electrode system in 1.0 M KOH solution, with Hg/HgO as the reference electrode and graphite rod as the counter electrode, respectively. The linear sweep voltammetry (LSV) curves of Co, Mo/Co, N-Mo/Co and commercial Pt/C clearly shows that N-Mo/Co exhibits the best performance among all samples (Fig. 5a). The overpotential to achieve the current density of 10 mA cm − 2 is only 12 mV, signi cantly better than those of Co (209 mV), Mo/Co (82 mV) and even the commercial 20 wt% Pt/C (32 mV). In Fig. 5b, the derived Tafel slopes of Co and Mo/Co are 129 and 89 mV dec − 1 , indicating that the water dissociation is the rate-determining step on Co and Mo/Co. However, the Tafel slope of N-Mo/Co decreases to 31 mV dec − 1 , which is even better than Pt/C (38 mV dec − 1 ), suggesting that the atomically dispersed charge localization of Mo δ+ site can improve the intrinsic water dissociation activity of N-Mo/Co for alkaline HER. The N-Mo/Co catalyst exhibits record activity in alkaline media (Fig. 5c). The turnover frequency (TOF) values of N-Mo/Co is also obviously larger than Mo/Co (Fig. 5d, Supplementary Fig. 18), revealing the superior intrinsic HER activity of N-Mo/Co. The electrochemical impedance spectroscopy (EIS) results shows that the charge-transfer resistance (R ct ) of N-Mo/Co is only 1.7 Ω, much lower than those of Co (95.6 Ω) and Mo/Co (6.5 Ω), suggesting that the N doped in Mo/Co SAA can promote charge transport kinetics ( Supplementary Fig. 19).
To assess the energy barriers of HER, we studied the effect of temperature on the performance of the catalysts and found the rate constants to follow the Arrhenius relationship. The Arrhenius plots for the Co, Mo/Co and N-Mo/Co catalysts allowed us to extract electrochemical activation energies for hydrogen production (Fig. 5e, Supplementary Fig. 20). It shows that N-Mo/Co exhibits an apparent barrier value of 10.8 kJ mol − 1 , which is signi cantly lower than those of Co (30.0 kJ mol − 1 ) and Mo/Co SAA (18.2 kJ mol − 1 ) catalyst, indicating the obviously accelerated kinetic process of HER on N-Mo/Co catalyst. Furthermore, the durability test of N-Mo/Co electrode at potential of 12 mV was performed to evaluate the stability of the N-Mo/Co (Fig. 5f). The catalyst exhibits an initial current density of 10 mA cm − 2 , and there is no obvious decrease after 200 h test, indicating its excellent electrochemical stability for long-term operation. The Faradaic e ciency remained about 100% in the HER process.
Inspired by such excellent HER activity, we also assemble a water-alkali electrolyzer with N-Mo/Co as cathode and anode simultaneously to evaluate its potential for overall water splitting. The cell voltage needed only 1.5 V to give a current density of 10 mA cm − 2 and it presented a negligible loss after a long time test of about 45 h (Supplementary Fig. 21). These results fully demonstrated that the N-induced atomically dispersed charge localization site of Mo δ+ on N-Mo/Co has high water dissociation activity and stability. Characterization. The X-ray diffraction (XRD) patterns were obtained on a Bruker D8 Discover X-ray diffractometer using monochromatized Cu Kα (λ = 1.5406 Å) as radiation source with a scanning step of 0.02º per second from 5º to 80º. The morphology and structure of the obtained samples were characterized by eld-emission SEM (NOVA NANOSEM 450), TEM, HAADF-STEM and EELS (Talos F200X and JEMARM 200F). The XPS measurements were performed on an ESCALAB250Xi spectrometer (Thermo Scienti c) using Al Kα (1486.68 eV) as X-ray source.
XAFS and AP-XPS measurements and analysis. The EXAFS and XANES of Co K-edge and Mo K-edge were performed at BL01C1 station of National Synchrotron Radiation Research Center, Taiwan. The transmission and uorescence modes were used for collecting data of Co and Mo K-edge respectively at room temperature. The raw data of EXAFS were processed following the standard procedures with the software packages of ATHENA. The nal EXAFS spectra were obtained after removing the post-edge background and normalizing the edge-jump step. The soft XANES of Co L-edge and N K-edge were carried out at beamline BL10B of National Synchrotron Radiation Laboratory (NSRL) in Hefei, China. The total electron yield mode was used for collecting the data of Co L-edge under an ultrahigh vacuum of 5 × 10 − 10 mbar. The photon ux was about 10 10 and the energy resolution is about 1000 (E/ΔE). The energy range between 760 and 805 eV was used to obtain the Co L-edge absorption spectrum. The soft XANES original data were processed using a common strategy with several steps. The energy of absorption spectrum was calibrated rstly with the Au 4f peak as a standard and then the pre-edge of substrate was set zero. At last, the edge-jump of the obtained spectra was normalized to one. The ultraviolet photoemission spectroscopy spectra were performed at BL10B of NSRL with the excitation energy of 170 eV.
Ambient-pressure XPS (AP-XPS) studies were performed at the National Synchrotron Radiation Research Center in Taiwan, at beam-line 24A1. The calibration of the binding energy (BE) scale was carried out using the Au 4f photoelectron peak as reference (4f7/2 BE = 84.0 eV), from a clean gold polycrystalline surface. The XPS peak shapes were tted with XPS Peak software using a Shirley background.
Electrochemical measurements. All electrochemical measurements were performed on an electrochemical workstation of Princeton 4000A using a three-electrode system with the as-prepared samples as working electrode, the graphite rod and Hg/HgO as the counter and reference electrode respectively. The sample of Pt/C was obtained by dropping the uniform ink of Pt/C (20%) on carbon cloth (1.0 mg cm − 2 ) where the Pt/C ink was prepared by ultrasonic dispersion of 10 mg Pt/C (20%) and 150 µL 5 wt% Na on solution in 2.5 ml water/isopropanol (v/v 4:1) solution. All potentials used in this work were converted to the values versus reversible hydrogen electrode (RHE) with the equation E RHE = E SCE + 0.0591 × pH + 0.241 V. The scan rate of 2 mv/s was used in the measurements of LSV. The overpotential of 50 mV with perturbation amplitude of 10 mV was performed in the measurements of electrochemical limpedance spectroscopy (EIS) in the frequency from 100 kHz to 0.001 Hz. The cyclic voltammograms (CVs) in the range between 0.01 and 0.11 V was used to estimate the electrochemical surface areas (ECSA) at the scan rates of 20, 40, 60, 80 and 100 mV s − 1 , respectively. The difference values of current density per square centimeter (cm 2 ) at 0.06 V vs RHE were plotted with the scan rates in which the slope was used to measure the ECSA. The quanti cation of the produced hydrogen was performed with a thermal conductivity detector (TCD) of gas chromatography (GC 2014 C SHIMADZU).
For temperature-dependent measurements, the electrochemical reaction cell with electrodes was put in a water bath. The temperature was rst improved to the value above 60 ℃, then the current values of catalyst at temperature of 60, 55, 50, 45, 40, 35, 30 and 25 ℃ were recorded with the temperature reducing in environment in order to decrease error. The performance of different catalysts at various temperatures is shown in table 1. As we all know, the kinetics of chemical reactions increases with the temperature rising which is the same with HER. The relation of chemical rate constant with the temperature is about proportional to exp(-ΔH*/kT), where ΔH* and k is the activation energy and the Boltzmann constant respectively. Specially, the approximate activation energy (E a ) for HER can be de ned by the Arrhenius relationship Where T is the test temperature, i k is the current at η= − 200 mV and R is the gas equilibrium constant.
The linear dependence on temperature of Arrhenius plots for four various catalysts are displayed in Fig. 3d. The E a for different HER catalysts can be obtained by the slopes of various Arrhenius plots (-R × slope).
Details of DFT calculations. Our density functional theory (DFT) simulations were performed using the Vienna ab initio simulation package (VASP). [39][40][41] Core − valence interactions was described by the projector augmented wave (PAW) method. 42 The Perdew-Burke-Ernzerh (PBE) functional within the generalized gradient approximation was used to describe the exchange and correlation functional. 43 The cutoff energy of 400 eV was set in our calculations, and 2 × 2 × 1 Monkhorst-Pack k grids are applied during the optimization. We built four layer Co(001) and Co(101) slab with vacuum layer 15 Å. During the geometry optimizations, the bottom two-layers were xed while the remaining layers were allowed to relax. Transition states (TS) were located by a combination of the Nudged Elastic Band (CI-NEB) and the Improved Dimer Method (IDM). [44][45][46] All TS structures have a single imaginary frequency. The iterative process considered was convergences, when the force on the atom was less than 0.02 eV Å −1 and the energy change was less than 10 − 5 eV per atom. We used the VESTA package to visualize the structures and charge density differences. 47 The Gibbs free energy of H* step was calculated as ΔG = ΔE + ΔZPE -T·ΔS Where ΔE is reaction energy calculated from DFT. ΔZPE and ΔS are the changes in zero-point energy and entropy during the reaction, respectively. [48][49][50] Data availability The data that support the ndings of this study are available from the corresponding author on reasonable request.

Con icts of interest
There are no con icts to declare.