Modulating electronic structure of metal-organic frameworks by introducing atomically dispersed Ru for efficient hydrogen evolution

Developing high-performance electrocatalysts toward hydrogen evolution reaction is important for clean and sustainable hydrogen energy, yet still challenging. Herein, we report a single-atom strategy to construct excellent metal-organic frameworks (MOFs) hydrogen evolution reaction electrocatalyst (NiRu0.13-BDC) by introducing atomically dispersed Ru. Significantly, the obtained NiRu0.13-BDC exhibits outstanding hydrogen evolution activity in all pH, especially with a low overpotential of 36 mV at a current density of 10 mA cm−2 in 1 M phosphate buffered saline solution, which is comparable to commercial Pt/C. X-ray absorption fine structures and the density functional theory calculations reveal that introducing Ru single-atom can modulate electronic structure of metal center in the MOF, leading to the optimization of binding strength for H2O and H*, and the enhancement of HER performance. This work establishes single-atom strategy as an efficient approach to modulate electronic structure of MOFs for catalyst design.

T he increasing consumption of fossil fuels and deterioration of living environment drive people to explore environmental friendly and sustainable energy sources as alternatives for the traditional fossil fuels [1][2][3][4] . Among them, hydrogen is considered as the most promising substitute because of its high gravimetric energy density as well as zero CO 2 emission [5][6][7][8] . Recently, electrochemical water splitting, generally including two half-reactions, hydrogen evolution reaction (HER) and oxygen evolution reaction (OER), has aroused increasing interests [9][10][11][12] . HER, producing low-cost and high purity hydrogen gas, is the hot spot in energy-conversion technologies and arouses increasing interests. Up to now, Pt is recognized as the high-performance electrocatalyst due to its fast dynamics and low overpotential [13][14][15][16] . Despite of the high efficiency, the high cost and scarcity impede its large-scale application and drive people to pursue more cheap and efficient electrocatalysts [17][18][19][20][21] .
Metal-organic frameworks (MOFs) are a class of emerging porous crystalline materials [22][23][24][25] composed of variety organic ligands and metal centers with various applications, such as water splitting [26][27][28][29] , gas storage [30][31][32] and metal-air batteries [33][34][35][36][37] . Benefitting from flexible tunability and well-defined structure of MOFs, their performance can be optimized using fundamental molecular chemistry principles 38 . This makes MOFs as promising model catalysts for investigating the design of catalysts at the molecular level. Though, some approaches including metal node engineering 39 , missing-linker MOF 40 and lattice-strained MOF 41 have been reported to design advanced OER electrocatalysts. Developing efficient strategies to regulate electrocatalytic performance of MOFs for HER is challenging.
Currently, single-atom catalysts (SACs) have intrigued new interests in heterogenous electrocatalysis because of their outstanding activity and maximum atom utilization efficiency 42,43 . The impressive catalytic activity of SACs greatly results from the distinctive electronic structure of single metal atoms and the interaction between single metal atoms and supports [44][45][46][47] . So, beyond serving as active sites, incorporating single-atom metals into catalysts can also lead to the local electronic structure modulation of initial catalysts 48 , owning to the electronic interaction between them. Moreover, the incorporation of atomically dispersed single-atom can preserve the structural feature of original materials. All of these aspects provide promising opportunity to introduce single-atom to enhance the electrocatalytic performance of MOFs.
Herein, we propose a single-atom strategy to tailor HER performance of the MOF Ni-BDC (Ni 2 (OH) 2 (C 8 H 4 O 4 ) 49,50 , H 2 BDC: terephthalic acid) by introducing atomically dispersed Ru (named as NiRu 0.13 -BDC). Remarkably, the as-synthesized catalyst performs enhanced activity toward HER in all pH. The optimized NiRu 0.13 -BDC catalyst exhibits high HER performance with a low overpotential of 36 mV at 10 mA cm −2 and a Tafel slope of 32 mV dec −1 , which are much improved compared with pure Ni-BDC in 1 M phosphate buffered saline (PBS) solution. More importantly, combining with calculation results, the electronic structure of Ni can be regulated by the construction of Ru singleatom into Ni-BDC, thus optimizing the adsorption strength for H 2 O and H* and contributing to the enhanced performance.

Results
Synthesis and characterization of NiRu 0.13 -BDC. First, a pristine MOF nanosheet array supported on Ni foam was prepared by a hydrothermal method. Furthermore, the structure of initial MOF material was introduced as "Ni-BDC" with layered-pillared structure constructed by the coordinated octahedrally divalent Ni and terephthalic acid (H 2 BDC), where the terephthalates are coordinated and pillared directly to the Ni hydroxide layers and form a three-dimensional framework ( Supplementary Fig. 1) 50 . Ru single-atom catalyst supported on Ni-BDC growing on nickel foam was synthesized by replacing part of Ni atoms through an ion-exchange method (Fig. 1a, Supplementary Fig. 1b and details in the Method section). Through varying the amount of RuCl 3 , a series of catalysts with different loading amount of Ru were synthesized. The content of Ru was determined by inductively coupled plasma mass spectrometry (ICP-MS) (Supplementary Table 1). The structures of these materials were firstly studied by powder X-ray diffraction (XRD). As revealed in Supplementary  Fig. 4, NiRu 0.09 -BDC and NiRu 0.13 -BDC have the similar diffraction patterns as Ni-BDC, the diffraction peak appeared at 8.9 was identified to the characteristic (2,0,0) facet of the Ni-BDC 49,50 , indicating the similar crystal structure. When the loading amount is higher than 13%, the structure of the MOF was destroyed, as shown in Supplementary Figs. 2 and 4. To further certify the structure of NiRu 0.13 -BDC, NiRu 0.13 -BDC powder sample was also synthesized via the same method excepting for the addition of nickel foam. As shown in Supplementary Fig. 5, the NiRu 0.13 -BDC powder exhibited the similar diffraction peaks as Ni-BDC.
The morphologies and microstructures of these catalysts were deeply investigated by scanning electron microscope (SEM) and transmission electron microscope (TEM). As disclosed in To further investigate the chemical composition and electronic structure of the NiRu 0.13 -BDC, X-ray photoelectron spectroscopy (XPS) was conducted. From Fig. 2a and Supplementary Fig. 6, the XPS spectra of the NiRu 0.13 -BDC demonstrated the coexistence of Ni, Ru, O, and C elements. Compared with Ni-BDC, after introducing Ru atoms, the Ru 3p spectra (Fig. 2b) can be clearly detected in NiRu 0.13 -BDC, further proving the successful doping of Ru. In addition, the Ru 3p peaks located at 462.9 and 485.5 eV were assigned to Ru 3+ 3p 3/2 and Ru 3+ 3p 1/2 , conforming the oxided state of Ru rather than metallic [51][52][53][54][55][56] . Meanwhile, there is a pair of peaks with an energy of 281.9 and 286.4 eV, which can be assigned to Ru 3d 5/2 and Ru 3d 3/2 in the C 1s and Ru 3d spectra  Fig. 6b), the Ru 3p peaks with binding energy of 463.4 and 485.8 eV were also result from the oxidized Ru. While, from Ru 3p of NiRu 0.21 -BDC, the peaks located at 461.7 and 484.3 eV for Ru 3p 3/2 and 3p 1/2 are assigned to metallic Ru, indicating the formation of Ru nanoparticles [51][52][53][54][55][56] . The Ni 2p spectra (Fig. 2c) of Ni-BDC show two characteristic peaks at 855.9 and 873.2 eV, identified as Ni 2p 3/2 and Ni 2p 1/2 severally, which were the characteristic peaks of the Ni 2+57,58 . The Ni 2p binding energy of NiRu 0.13 -BDC is higher than that of Ni-BDC, suggesting strong electron interaction between Ni and Ru atoms and electron depletion on Ni. The O 1s spectra in Fig. 2d, can be deconvoluted into three peaks with their binding energies at 531.3, 532.0 and 532.9 eV, assigned to the Ni (Ru)-O, O-C=O, and absorbed water species respectively 57 .
The local electronic structure and coordination environment of Ru in NiRu 0.13 -BDC were further investigated by X-ray absorption measurements. Figure 3a and Supplementary Fig. 7 show the X-ray absorption near-edge structure (XANES) of RuO 2 , Ru foil and NiRu 0.13 -BDC, it can be observed that the threshold value (E 0 ) of the NiRu 0.13 -BDC (22129.17 eV) is between RuO 2 (22130.25 eV) and Ru foil (22127.02 eV), indicating the valence state of Ru is between 0 and +4. In order to further estimate the valance state of the Ru atoms in NiRu 0.13 -BDC, we performed the ΔE 0 as a function of oxidation state of Ru atoms in these materials 59 . Hence, the average valance state of Ru atoms in NiRu 0.13 -BDC is +2.67, as shown in Fig. 3b. The extended X-ray absorption fine structure (EXAFS) of NiRu 0.13 -BDC shows only a primary peak at 1.5 Å, which is assigned to Ru-O bond (Fig. 3c). Compared with Ru foil and RuO 2 , there are no obvious characteristic peaks can be identified to Ru-Ru metallic bond at 2.50 Å in Ru foil and Ru-Ru bond at 3.25 Å in RuO 2 53 , revealing the Ru single-atom successfully dispersed in the Ni-BDC. In order to further confirm the structure of Ru in NiRu 0.13 -BDC, we fitted the Fourier transform XAFS in R-space of Ru k-edge using the structure model of replacing Ni atoms in Ni-BDC by Ru atoms. As can be seen from Supplementary Fig. 7c and Table 2, the fitting spectrum is well consistent with as measured, further confirming Ru single-atom replaced part of Ni in Ni-BDC. Taking account into XPS spectra and XRD pattern, Ru was atomically anchored in the MOF Ni-BDC. Furthermore, as shown in Fig. 3d and Supplementary Fig. 8, the energy of Ni positively shifted to higher in comparison with Ni-BDC, verifying the electron interaction between Ni and Ru atoms, which is in agreement with XPS analysis.
Electrocatalytic performance toward HER. The electrocatalytic performances of these catalysts were firstly measured in 1 M PBS solution with a three-electrode cell system under room temperature. And the loading mass of Ni-BDC, NiRu 0.09 -BDC, NiRu 0.13 -BDC, and NiRu 0.21 -BDC was about 2.5 mg cm −2 . To evaluate the HER activity of these materials, commercial Pt/C and Ru/C were utilized as benchmarks with the same loading. As exhibited in Fig. 4a, Ni-BDC performed poor electrocatalytic activity with an overpotential of 389 mV to reach a current density of 10 mA cm −2 . In contrast, after introducing Ru single-atom into Ni-BDC, the catalysts exhibited enhanced electrocatalytic performance toward HER. Remarkably, the NiRu 0.13 -BDC displayed high HER activity with a low overpotential of 36 mV at 10 mV cm −2 in 1 M PBS solution, which is much lower than that of Ru/C (115 mV) and even comparable to the commercial Pt/C (22 mV). Furthermore, NiRu 0.13 -BDC only needed an overpotential of 132 mV to reach a high current density of 100 mA cm −2 , which was lower than that of Pt/C (139 mV) ( Fig. 4a and Supplementary Fig. 9). The remarkable electroctatalytic performance of NiRu 0.13 -BDC also outperformed other previously reported HER electrocatalysts (Supplementary  Table 3). Moreover, NiRu 0.21 -BDC with higher Ru content, displayed worse activity than NiRu 0.13 -BDC, further suggesting the superiority of single-atom NiRu 0.13 -BDC electrocatalyst. The Tafel slope (Fig. 4b) of NiRu 0.13 -BDC is 32 mV dec −1 , lower than that of Ni-BDC (219 mV dec −1 ), NiRu 0.09 -BDC (54 mV dec −1 ) and NiRu 0.21 -BDC (60 mV dec −1 ), which is comparable to Pt/C, indicating that the NiRu 0.13 -BDC performed as the best catalyst of the series. The electrocatalytic performances were also tested in 1 M KOH and 1 M HCl. NiRu 0.13 -BDC also exhibited the best HER activity among these catalysts with an overpotential of 34 and 13 mV at a current density of 10 mA cm −2 in 1 M KOH and 1 M HCl, respectively, in Fig. 4c and Supplementary Fig. 9. The Tafel  slope curves (Fig. 4d) also revealed NiRu 0.13 -BDC possessing smaller Tafel slope (32 mV dec −1 ) with accelerated HER kinetics in alkaline solution. The TOFs (Supplementary Fig. 10) of Ni-BDC, NiRu 0.09 -BDC, NiRu 0.13 -BDC, and NiRu 0.21 -BDC are 0.000048, 0.0032, 0.0091, and 0.0035 s −1 , respectively. As can be seen from Supplementary Fig. 10, NiRu 0.13 -BDC has higher TOF value, indicating higher intrinsic activity. In addition, from the Supplementary  Fig. 11, the faradic efficiency of NiRu 0.13 -BDC catalyst for HER is estimated to be close to 100%, indicating that almost all electrons are utilized for producing hydrogen.
To figure out the origin of the enhanced activity of the NiRu 0.13 -BDC electrocatalyst, the double-layer capacitances (C dl ) measurements were carried out to evaluate the electrochemical active surface areas ( Supplementary Fig. 12). NiRu 0.13 -BDC had higher C dl (1.39 F cm −2 ) 60 than that of Ni-BDC (0.0021 F cm −2 ), NiRu 0.09 -BDC (0.2797 F cm −2 ) and NiRu 0.21 -BDC (0.7572 F cm −2 ), indicating more electroactive surface exposed in NiRu 0.13 -BDC compared with pure Ni-BDC. The electrochemical impedance spectroscopy (EIS) was performed to deeply study the charge-transfer mechanism and the resulted Nyquist plots were shown in Supplementary Fig. 13 and Table 4. Apparently, the NiRu 0.13 -BDC presented a smaller R ct value than other electrocatalysts, implying faster charge transfer. The long endurable stability was measured by chronoamperometry with an overpotential of 50 mV (vs RHE) in 1 M PBS solution. NiRu 0.13 -BDC exhibited good stability with a negligible current decrease after 30 h test ( Supplementary Fig. 13). After stability test for 10 h, the nanosheets morphology maintained well ( Supplementary  Fig. 14). The main phases are NiRu 0.13 -BDC, indicating that the crystal structure of MOF catalyst showed limited changes ( Supplementary Fig. 15). The XPS spectra of NiRu 0.13 -BDC after electrocatalysis exhibited that the peak of Ru 3p also showed limited changes ( Supplementary Fig. 16), indicating the main component is still NiRu 0.13 -BDC MOF catalyst. The limited changes in crystal structure and chemical environment of NiRu 0.13 -BDC after 10 h HER stability test indicated the MOF had good stability.
Density functional theory (DFT) calculations. To understand the effect of Ru single-atom in HER performance, we conducted DFT calculations. Supplementary Fig. 17 shows structure models of Ni-BDC and NiRu 0.13 -BDC (details in "Computation method"). For investigating the electronic structure of NiRu 0.13 -BDC, the charge density difference was firstly simulated. From Fig. 5a, there is overt charge accumulation around Ru and charge depletion around Ni, revealing charge interaction between Ni and Ru, which is in agreement with the above-mentioned XPS and XANES results. It further confirmed after introducing of Ru single-atom, the electronic structure of metal center was modulated. In order to further elucidate the origin of the improved HER activity of NiRu 0.13 -BDC after introducing Ru single-atom, the density of states (DOS) was also calculated to deeply investigate the changes of electronic structure of the catalyst. According to the partial and total DOS calculations ( Fig. 5b and Supplementary Figs. 18, 19), with the formation of Ru singleatom in the MOF, the electronic structure of NiRu 0.13 -BDC changed. It should be noted that the d-band center of Ni shifted to lower energy (Fig. 5b), corresponding to a weaker H* adsorption on catalyst 61 . To elucidate the inherent relationship between the electronic structure and the enhanced electrocatalytic HER performance of NiRu 0.13 -BDC, the adsorption free energy of the HER intermediates was also calculated. Generally, there are two steps involved in neutral pH HER, including H 2 O adsorption/activation and H recombination on the surface of the catalyst 62 ; so, a strong bonding of H 2 O and neither too strong nor too weak bonding of H to the surface are desired 63,64 . As displayed in Fig. 5c, Ru in NiRu 0.13 -BDC shows much lower adsorption energy of H 2 O (ΔG H2O* ) compared with Ni in Ni-BDC and NiRu 0.13 -BDC, indicating the strongest water adsorption, which benefits for the following step to generate adsorbed H atoms 65,66 . In addition, the calculated ΔG H* for adsorbed H atom forming molecular H 2 is the key descriptor to predict and evaluate the activity for HER on catalyst surface 67 . While, Ru in NiRu 0.13 -BDC and Ni in Ni-BDC possess a more negative value of ΔG H* , indicating excessive strong adsorption of H and adverse for H desorption and H 2 release. By contrast, the ΔG H* of Ni in NiRu 0.13 -BDC (Fig. 5d) is closer to the optimal value of ΔG H* = 0 eV, supporting high HER activity. For the NiRu 0.13 -BDC catalyst, with both Ni and Ru atoms in the crystal structure, it showcases a preferential HER activity. From the above results, it can be found that constructing Ru single-atom into the MOF can regulate the electronic states of Ni and Ru, and d-band center of Ni, resulting in an enhanced ΔG H2O* and a more thermoneutral ΔG H* , thus devoting to improved HER performance.

Discussion
In conclusion, we have synthesized the HER catalyst with significantly improved performance by introducing Ru single-atom into the MOF Ni-BDC. Impressingly, the as-sythesized NiRu 0.13 -BDC exhibits enhanced activity toward HER in all pH, comparable with commercial Pt/C, especially delivering a low overpotential of 36 mV at 10 mA cm −2 with a Tafel slope of 32 mV dec −1 in 1 M PBS solution. More improtantly, after introducing Ru, there is strong electron interaction between Ni and Ru atoms and electron depletion on Ni in the MOF from XPS results, also confirmed by EXAFS and XANES. DFT calculations disclose that the introduction of atomically dispersed Ru can promote the adsorption of H 2 O and optimize good thermoneutral ΔG H* to facilitate H 2 release on Ni site of the MOF by regulating electronic structure, thus effciently enhancing the HER activity. This work paves a way of designing efficient MOFs electrocatalysts via single-atom strategy to modulate electronic structure and intermediates asorption. Characterization. Powder X-ray diffraction was conducted on Rigaku SmartLab diffractometer equipped with Cu Kα X-ray source (λ = 1.540598 Å). SEM measurements were operated on a Hitachi SU8010 system. TEM images were carried out on a JEM-1400Plus apparatus. Scanning transmission electron microscopy (STEM) and corresponding EDS mapping images were obtained from a JEOL JEM-ARM 200F equipped with energy dispersive X-ray spectrometer. X-ray photoelectron spectra were obtained from a Thermo fisher Scientific K-Alpha + instrument. Inductively coupled plasma mass spectrometry (ICP-MS) was measured on Thermo Scientific iCAP RQ.
Electrochemical measurements. The electrochemical measurements were evaluated in a standard three-electrode cell system by using a CHI 760D (Shanghai, China) instrument. The catalysts were cut into 0.5 × 1 cm 2 pieces utilizing as the working electrodes. An Ag/AgCl (3 M KCl) electrode and carbon rod were used as the reference and counter electrode, respectively. The measured potential was converted relative to RHE according to the following equation: E (vs RHE) = E Ag/AgCl + 0.21 + 0.059 × pH. The Linear sweep voltammetry (LSV) was conducted at a scan rate of 2 mV s −1 with the potential corrected for iR loss. The CV tests were studied in the potential range from 0.386 to 0.486 V (vs RHE) at different scan rates. By plotting the difference of current density (j) at 0.436 V (vs. RHE) against the scan rate, we gained a line where the slope is equal to the geometric doublelayer capacitance (C dl ). EIS was evaluated with 5 mV amplitude in a frequency range from 0.01 to 10,000 Hz at −1.036 V (vs. Ag/AgCl). The turnover frequency (s −1 ) can be estimated with the following equation: TOF = I/2nF; where I is the current (A) during LSV, F is the Faraday constant (96485.3 C mol −1 ), n is the number of active sites (mol). The factor 1/2 is based on the assumption that two electrons are necessary to form a hydrogen molecule. Faradic efficiency was evaluated in a H-type cell with an anion exchange membrane as the separator and a gas chromatography (SHIMADZU GC-2014) for the hydrogen gas detection.
Computation method. All the calculations are performed in the framework of the density functional theory with the projector augmented plane-wave method, as implemented in the Vienna ab initio simulation package 68 . The generalized gradient approximation proposed by Perdew, Burke, and Ernzerhof is selected for the exchange-correlation potential 69 . The Ni-BDC crystal structure has been modeled using a single periodic slab with a (4 × 2) supercell based on the previously reported MOF 50 , and the (200) facet was investigated which was dominant facet according to the XRD patterns. The benzene ring was passivated with hydrogen. And the NiRu 0.13 -BDC structure was derived by replacing one Ni atom with Ru atom. The reasonable vacuum layers were set to 20 Å, in order to avoid the interaction between periodic structures. Furthermore, when a 30 Å vacuum layer was set, the energy deviation yields to be 0.1 meV/atom compared with that of 20 Å, for further calculations we chose the 20 Å vacuum layer. The cut-off energy for plane wave is set to 400 eV. The energy criterion is set to 10 −5 eV in iterative solution of the Kohn-Sham equation. The Brillouin zone integration is performed using a 1 × 2 × 2 k-mesh, for the DOS calculations the 1 × 5 × 5 k-mesh was applied. The slab contains four layers with two layers are fixed. All the structures are relaxed until the residual forces on the atoms have declined to less than 0.05 eV/Å. The free energies are calculated by the formula: ΔG = ΔE tot + ΔE ZPE − TΔS, where ΔE tot , ΔE ZPE , and ΔS are the changes in total energy of the system, vibrational zero-point energy, entropy during the reaction, respectively; while T represents temperature 70 .

Data availability
Full data supporting the findings of this study are available within the article and its Supplementary Information, as well as from the corresponding author upon reasonable request.