Ruthenium-cobalt nanoalloys encapsulated in nitrogen-doped graphene as active electrocatalysts for producing hydrogen in alkaline media

Abstract

The scalable production of hydrogen could conveniently be realized by alkaline water electrolysis. Currently, the major challenge confronting hydrogen evolution reaction (HER) is lacking inexpensive alternatives to platinum-based electrocatalysts. Here we report a high-efficient and stable electrocatalyst composed of ruthenium and cobalt bimetallic nanoalloy encapsulated in nitrogen-doped graphene layers. The catalysts display remarkable performance with low overpotentials of only 28 and 218 mV at 10 and 100 mA cm⁻², respectively, and excellent stability of 10,000 cycles. Ruthenium is the cheapest platinum-group metal and its amount in the catalyst is only 3.58 wt.%, showing the catalyst high activity at a very competitive price. Density functional theory calculations reveal that the introduction of ruthenium atoms into cobalt core can improve the efficiency of electron transfer from alloy core to graphene shell, beneficial for enhancing carbon–hydrogen bond, thereby lowing ΔG_H* of HER.

Introduction

Hydrogen energy, a renewable energy resource, has been intensely investigated as an ideal alternative to the conservative fossil fuels for its high gravimetric energy density, zero-emission and earth-abundance^1,2,3. Despite its promising application prospect, the current industrial route of hydrogen production not only consumes the fossil fuels but also brings on the greenhouse gas CO₂ emission^4,5. Thus, exploring a cleaner, more sustainable and efficient strategy is urgently needed. Currently, electrochemical water splitting, a non-fossil fuel-based technology, is evoking increasing interests and stimulating intense investigations to produce low-costing and high-pure hydrogen^6,7. As the heart of the scalable hydrogen production, it is of great importance to develop highly efficient electrocatalysts to drive the hydrogen evolution reaction (HER)⁸.

It is well known, electrochemical water splitting can be carried out in acidic or alkaline media^2,4. Unfortunately, the acid electrolyzers are technologically and commercially hindered by the lacking of efficient and low-cost counter electrode catalysts in acidic electrolytes^2,9,10. Consequently, tremendous efforts have been devoted to developing HER catalysts with high activity and stability in basic media on the basis of the available alkaline oxygen evolution electrocatalysts, aiming at accelerating commercialization of the basic electrolyzers for H₂-production.

Platinum, as ‘the Holy Grail’ of HER electrocatalysts, remains the best HER catalysts with nearly zero overpotential and excellent long-term durability^3,11,12,13. Unfortunately, the widespread commercialization of Pt-based electrocatalysts are hindered by their scarcity and expensive price^14,15. Therefore, it is of great significance to explore inexpensive alternatives for Pt electrocatalysts. To date, plenty of robust and efficient alternative catalysts have been reported aiming at replacing Pt-based electrocatalysts in alkaline media. Among these materials, transition metal (TM) -based catalysts, including Mo-based catalysts^2,4,9,16,17, Ni-based catalysts^6,10,18,19 and Co-based catalysts^14,20,21,22, have been proven to be competitive eletrocatalysts as a result of their high efficiency and low cost. In particular, the biphasic nanocrystalline Ni–Mo–N catalyst recently reported by Li’s group exhibited amazing HER activity with low overpotentials of 43 mV in 1 M KOH and 53 mV in 0.5 M H₂SO₄, which is close to commercial Pt/C catalyst at the same mass loading of 1 mg cm⁻² (ref. 2). Although, some of the TM-based catalysts show high HER performance, but still inferior to the Pt-based catalysts in overpotential and durability, which would increase energy consumption and hence decreasing economic competitiveness^23,24. Currently, the pure TM-based catalysts are unable to meet the requirement of replacing Pt-based electrocatalysts.

For decades, other cheaper platinum-group metals have been investigated for HER in view of their high similarity to Pt in chemical inertness^13,25,26,27. In particular, ruthenium (42 $ per oz) is more economically advantageous in price than the rest of Pt-group metals, such as Pt (992 $ per oz), Pd (551 $ per oz), Ir (500 $ per oz) and so on²⁸. As a matter of fact, Ru has evoked special attention as a top oxygen evolution electrocatalytic material²⁹. However, it is not very active for HER in basic media in previous works^30,31,32,33. It is shown that alloying noble metals with other TMs, with the amount of noble metal even decreased by up to an order of magnitude, is a major route to prepare highly efficient catalysts with balance of good cost-competitiveness. Moreover, previous studies have proven that the chemical properties of bimetallic surfaces could be modified by the combined changes in the average energy of the surface d-band and in the width of the d-band due to the cumulative strain and ligand effects originated from the formation of heteroatom bonds and the alteration of the bond length^8,34. Therefore, in addition to lowering the material cost, the noble-transition bimetallic alloys could remarkably boost the electrocatalytic activities contributed by the shift of charge distributions and the resulting modification of surface properties during the formation of alloys^26,35,36.

In this work, a Ru-based electrochemical catalyst for HER was developed by alloy Ru with TM Co. Pervious work of our group and Bao’s group have shown that a metal core coated with a carbon shell, especially N-doped graphene, can simultaneously promote the HER activity through the synergism and enhance the stability due to the protection from carbon cage^8,37. Inspired by this, a metal-organic frameworks (MOFs) -assisted strategy was adopted for the preparation of RuCo nanoalloys encapsulated in nitrogen-doped graphene layers (RuCo@NC). We adopted an in situ method that consists of the one-step annealing of Ru-doped Prussian blue analogues, which are ideal nitrogen-rich precursors composed of metals as nodes and CN^- groups as linkers, for the fabrication of alloy materials wrapped in N-doped carbon^{8,26,38,39,40,41}. The RuCo@NC hybrid material shows an unprecedented high electrocatalytic performance towards HER in basic conditions, even superior to the commercial Pt/C catalysts.

Results

Synthesis and characterization of RuCo@NC catalyst

Obviously, Fig. 1 illustrates the synthetic route and model of the RuCo@NC hybrids. The Co₃[Co(CN)₆]₂ MOF precursor particles, designated as S-0-MOF, were synthesized according to our previous studies^41,42. As revealed by the field-emission scanning electron microscopy (FESEM) and transmission electron microscopy (TEM) images (Supplementary Fig. 1a,b), the morphologies of the S-0-MOF particles were truncated nanocubes which have very narrow diameter distributions with a mean diameter of ∼100 nm. In Fig. 1b, the Ru-doped Co₃[Co(CN)₆]₂ precursor were obtained via an ion-exchange reaction in the liquid phase. In brief, various RuCl₃ solution was added into the Co₃[Co(CN)₆]₂ solution, and the corresponding products were hereinafter designated as S-1-MOF, S-2-MOF, S-3-MOF, S-4-MOF, S-5-MOF and S-6-MOF, respectively. After adding Ru source, Ru³⁺ ion diffused into the open framework of Co₃[Co(CN)₆]₂, which triggered an ion-exchange reaction between Ru³⁺ and Co³⁺ while maintaining intrinsic framework structure^43,44. The FESEM and TEM images of the obtained Ru-doped precursor particles were illustrated respectively in Fig. 2a,b and Supplementary Fig. 2–6a,b, exhibiting that the as-prepared noble metal-doped Co₃[Co(CN)₆]₂ inherited the nanocubic morphology with an average diameter of ∼100 nm. The corresponding X-ray diffraction patterns are shown in Fig. 2c. The patterns of the samples all showed no additional reflections except a series of Bragg reflections corresponding to the diffractions from the Co₃[Co(CN)₆]₂ (JCPDS No. 77−1161). The unchanged X-ray diffraction patterns of Ru-doped MOFs suggest that Co₃[Co(CN)₆]₂ could maintain the intrinsic framework structure during the Ru doping process, which is in good agreement with the unchanged morphologies of the precursors discussed above. The X-ray photoelectron spectroscopy (XPS) spectra of S-4-MOF were revealed in Fig. 2d and Supplementary Fig. 9. Obviously, the Ru 3p peaks at 463.3 and 485.8 eV were, respectively, assigned to Ru 3p 3/2 and Ru 3p 1/2 for Ru(III) species, conforming that Ru³⁺ have taken the place of some Co³⁺ ions. Essentially, the Ru doping process was carried out in the MOF framework structure via an ion-exchange reaction⁴³.

Figure 1: Schematic illustration of the synthetic route and model of the RuCo nanoalloys encapsulated in nitrogen-doped graphene layers.

(a) a nanocube of Co₃[Co(CN)₆]₂ precursor, (b) a nanocube of Ru-doped Co₃[Co(CN)₆]₂, (c) an aggregate of RuCo alloys encapsulated in graphene layers and (d) enlarged model of RuCo nanoalloy encapsulated in nitrogen-doped graphene layers as an electrocatalyst towards hydrogen evolution reaction in alkaline media.

Figure 2: Ru-doped MOF Co₃[Co(CN)₆]₂ MOF precursor particles.

(a,b) FESEM and TEM images of the as-prepared S-4-MOF. Scale bars, 100 nm. (c) The X-ray diffraction patterns of S-0-MOF, S-1-MOF, S-2-MOF, S-3-MOF, S-4-MOF, S-5-MOF and S-6-MOF. (d) The XPS result of the Ru 3p spectrum enlarged from Supplementary Fig. 9a.

In this work, the electrocatalysts were synthesized via an ‘in situ’ formation process without adding any other carbon sources. Co₃[Co(CN)₆]₂ is composed of CN⁻ groups as linkers and TM Co as metallic nodes, which is an ideal precursor for the preparation of composite materials of TM-based material with highly N-doped carbon³⁹. Specifically, the corresponding RuCo@NC hybrids were synthesized via one-step annealing of the obtained Ru-doped MOFs, hereinafter marked as S-0, S-1, S-2, S-3, S-4, S-5 and S-6, respectively. As illustrated in Fig. 1c,d, the nanocubic MOF precursors were directly carbonized at 600 °C under a nitrogen flow without adding any other carbon sources. The resulting product is composed of the bimetallic RuCo nanoalloys encapsulated in nitrogen-doped graphene layers. During the annealing process, Ru and Co atoms from the precursor would form the bimetallic RuCo nanocrystals, meanwhile some remaining CN⁻ group linkers would transfer into nitrogen-doped graphene layers. As can be seen in Fig. 1d, the formation of RuCo nanoalloys will be coated speedily by the in situ formed N-doped graphene layers, which is able to efficiently avoid the agglomeration of the inside alloy particles to provide more eletrocatalytic active sites which are beneficial to HER activity and long-term corrosion protection to enhance the catalytic durability.

All the obtained samples embraced an irregular particle-like morphology according to the FESEM results (Fig. 3a and Supplementary Fig. 2–6d). As revealed in the TEM images (Fig. 3b, Supplementary Figs 2–6c,e,f and 7 and Supplementary Table 1), the irregular products were composed of small alloy particles with a mean diameter of ∼30 nm encapsulated carbon layers. The high-resolution TEM (HRTEM) images (Fig. 3c,d) showed that the small alloy particles were coated with N-doped graphene layers. Most of the graphene layers were ∼6–15 layers thick. However, as shown in Supplementary Fig. 8, a large proportion of graphene shells consisting of 1–5 layers still existed, which are good active sites for HER. The thin graphene shells were beneficial in electron transfer from the alloy core to the shell and hence improved the catalytic activity on the graphene surface^8,37. Besides, Fig. 3e clearly revealed that lattice fringe spaces of 0.251 and 0.204 nm are, respectively, consistent with the (111) and (220) planes of the cubic Co phase or the (002) and (220) planes of the hexagonal Co phase. While, no lattice fringe spaces of metallic Ru were found, implying the formation of RuCo alloy which kept the crystal structure of metallic Co. As shown in Fig. 3f–j, the images of elemental mapping from energy filtered TEM showed that the Co and Ru elements (5.69 wt.% Ru and 94.31 wt.% Co) were uniformly distributed in the inner particles and surrounded by the C and N elements, further confirming the formation of RuCo nanoalloys encapsulated in nitrogen-doped graphene layers.

Figure 3: SEM and TEM characterization of RuCo@NC hybrid.

(a,b) The FESEM and TEM images of S-4. Scale bars, 100 nm. (c–e) HRTEM images of S-4. Scale bars, 10 nm. (f–j) HAADF-STEM image and corresponding EDX maps of S-4 for Co (g), Ru (h), C (i) and N (j), respectively. Scale bars, 20 nm.

The corresponding X-ray diffraction patterns of RuCo@NC were shown in Fig. 4a. Similar diffraction features of a broad and weak peak of C (002) were detected, confirming the existence of carbon layers. Besides, the patterns of all the samples showed no additional reflections except a series of Bragg reflections corresponding to the diffractions from the hexagonal structure Co (JCPDS card no. 05−0727) phase and face centred cubic (FCC) Co (JCPDS card no. 15-0806) (refs 7, 26), suggesting that Ru atoms were in situ dissolved in metallic Co to form RuCo alloy, which is in good agreement with the value determined by the HRTEM observations discussed above.

Figure 4: Structural analysis of RuCo@NC hybrid.

(a) The X-ray diffraction patterns of S-0, S-1, S-2, S-3, S-4, S-5 and S-6, respectively. (b) Raman spectrum of S-4, (c) The XPS result of the N1s spectrum enlarged from Supplementary Fig. 10a, (d) N₂ adsorption–desorption isotherm and pore size distribution plot (inset) of S-4.

The Raman spectra of S-4 were shown in Fig. 4b. The product displayed three Raman peaks locating at ∼1,349, 1,583 and 2,703 cm⁻¹, which corresponded to the D, G and 2D bands, respectively. The high I_D/I_G band intensity ratio of S-4 indicated the generation of large amounts of defects, suggesting that a large amount of N atoms were doped in the graphitic carbon layers. Moreover, the second-order band is broad and weak, implying that the coated carbon is thin graphene with several layers⁴⁵.

The XPS spectra of S-4 were illustrated in Supplementary Fig. 10a. The nitrogen content of S-4 was ∼3.51 atom%. The high-resolution N1s spectrum of S-650 (as shown in Fig. 4c) can be deconvoluted into four individual peaks assigned to pyridinic-N (398.5 eV), pyrrolic-N (399.6, 400.9 eV) and quaternary-N (401.6 eV), respectively^8,46. Importantly, two pyrrolic-N-binding energies are observed here. This phenomenon might result from the energy shift which was induced by the interaction between some pyrrolic-N and metal atoms⁴⁷. Therefore, the percentages of pyridinic-N, metal-N, pyrrolic-N and quaternary-N were 40%, 17%, 27% and 16%, respectively. Besides, XPS investigation of the Co 2p spectrum revealed the presence of two distinct chemical species: Co⁰ and Co^II species peaks, indicating the existence of Co^II originated from surface oxidation of metallic Co. The existence of metallic Ru was also confirmed by the weak peak at 462.2 eV (Supplementary Fig. 10c), suggesting that the Ru content was very tiny in the RuCo alloys with good balance of its cost.

The specific surface area and pore size distribution were obtained by N₂ adsorption/desorption isotherms. As can be seen in Fig. 4d, a type-II isotherm with a H3-type hysteresis loop is obtained, which is characteristic of mesoporous non-rigid aggregates⁴⁸. Obviously, S-4 have a specific surface area of 41.365 m² g⁻¹. The pore size distribution (insert image) revealed a strong and narrow distribution centred at ∼3 nm. The other samples showed the similar adsorption/desorption isotherm curves and pore size distribution in Supplementary Fig. 11. The BET surface area were listed in Supplementary Table 1. These results were in good agreement with the value determined by the morphology observations discussed above.

Electrochemical characterization for HER catalysis

The electrochemical catalytic activities of the annealed samples for HER were recorded by a typical three-electrode electrochemical cell in N₂ saturated 1 M KOH electrolyte. The HER performance was evaluated by observing the overpotential versus the reversible hydrogen electrode (RHE) at 10 mA cm⁻², which is the current density expected for a 12.3% efficient solar water-splitting device¹⁵. As revealed in Fig. 5a, the undoped S-0 catalyst exhibited an excellent HER activity but far inferior to the doped counterparts of RuCo alloy encapsulated in N-doped graphene layers, proving that alloying TMs with less noble metals played a key role in boosting the electrocatalysis activity with good balance of the cost. The polarization curves also showed that the S-4 catalyst had the highest HER activity among seven catalysts with an overpotential of only 28 mV without IR-correction. As illustrated in Table 1, the trend in the overpotentials was found to be S-4 (28 mV)<S-3 (67 mV)<S-2 (83 mV)<S-1 (91 mV)<S-0 (300 mV), suggesting that the activity was increased with the increasing amount of Ru. Interestingly, the activity would decrease with the further increase of Ru amount. Besides, the similar trend of overpotentials could also be observed reaching a higher current density of 100 mA cm⁻² (Table 1 and Supplementary Fig. 12b). To gain further insight into the activity of S-4, the inner metal of S-4 was etched via 1 M HCl solution. Obviously, S-4 retains the same morphology, but a part of its metallic cores are removed after the acid leaching (Supplementary Fig. 13a). Besides, the HCl solution became pink in Supplementary Fig. 13c, suggesting that just a part of metallic Co was corroded and dissolved into the solution. It was also testified by the neglectable Ru concentration (0.46 wt.% Ru in Co and Ru) of ICP. It was shown in Fig. 5d and Supplementary Fig. 13d that both of the overpotentials reaching current densities of 10 and 100 mA cm⁻² increased after etching, revealing that the metallic Co in S-4 was indispensable and had a key role in HER activity. It was also found that S-4 with lower Ru content (3.58 wt.% Ru in RuCo alloy, obtained from ICP) showed better activity than the etched counterpart with higher Ru content (17.7 wt.% Ru in RuCo alloy, obtained from ICP), which was in good agreement with the above- mentioned trend of overpotentials. All these findings revealed that S-4 (Ru, 3.58 wt. %, obtained from ICP) hold the best alloy structure for HER among RuCo alloys.

Figure 5: Electrocatalytic HER performance test of catalysts in N₂ saturated 1 M KOH solution.

(a) HER polarization curves of RuCo@NC samples with the same mass loading, (b) HER polarization curves of S-4, Ru and Pt/C with the same mass loading and durability test after 10,000th cycles, (c) the Tafel plots of S-4, Ru and Pt/C, (d) HER polarization curve of etched S-4 by 1 M HCl.

Table 1 Comparison of HER activity of different catalysts.

The electrochemically active surface area of samples was estimated using a simple cyclic voltammetry (CV) method^9,49. Due to the unknown capacitive behaviour (Cs) of the RuCo alloy electrode especially with N-doped graphene shell, we can safely estimate relative surface areas of seven samples, since the double-layer capacitance (Cdl) is expected to be linearly proportional to effective active surface area for samples with similar composition and this method was also employed in previous study⁸. The results in Supplementary Fig. 14 suggested that with the increasing density of catalytically active sites of samples also follow the similar trend of the catalytic performance. A series of activity normalization of special activity (SA) and active site activity (ASA), taking into account of the surface area and active site concentration respectively at overpotential of 100 mV, were shown in Supplementary Fig. 16 (ref. 50). Especially, the S-4 catalyst exhibited a SA of 0.707 mA cm⁻², which are much higher than other catalysts. Besides, the result also suggests that the SA of samples increased with the increasing catalytic active in general, suggesting the effect of surface area is a negligible factor on activity. Due to the unknown capacitive behaviour (C_s) of the RuCo alloy electrode especially with N-doped graphene shell, we carried out activity normalization of ASA*C_S. Especially, the S-4 catalyst exhibited an ASA*C_S of 1.242 mA cm⁻², which are much higher than other catalysts with the similar value of ASA*C_S. Besides, the electrochemical impedance spectroscopy images (Supplementary Fig. 15) showed that the Co@NC has a biggest semicircle radius, indicative of a slightly higher charge-transfer impedance than of S-1 (8.53 Ω), of S-2 (8.55 Ω), of S-3 (8.58 Ω), of S-4, of S-5 (8.53 Ω) and of S-6 (8.47 Ω) (refs 51, 52). Therefore, alloying Co with small amount of Ru could lead to a higher charge-transfer rate and more facile catalytic kinetics toward HER.

The commercial Ru powders catalyst (Ru: 99.9 wt.%) and the commercial Pt/C catalyst (Pt: 20 wt.%) were also measured as references. Surprisingly, the obtained S-4 even exhibited better activity than Ru catalyst (58 mV) and Pt/C catalyst (40 mV) (Fig. 5b). Moreover, the fast HER kinetics also enabled the S-4 catalyst to reach a high current density of 100 mA cm⁻² at η as low as 0.218 V, outperforming the other HER catalysts including Pt/C (Supplementary Fig. 12a and Table 1). It is also the best one among some catalysts in the recent reports towards HER in basic media (Table 2)^{1,2,4,6,9,17,18,19,21,22,25,53,54,55,56}. The Tafel slope (Fig. 5c) of the S-4 sample was 31 mV dec⁻¹, which was as same as the value of Pt/C, implying a rapid HER rate and a Tafel–Volmer mechanism with electrochemical desorption of H₂ as the rate-determining step in the HER process⁵⁷. While, the Tafel slope was 55 mV for the Ru powder catalyst, which was much larger and indicative of a typical Volmer−Heyrovsky mechanism with the Volmer step as rate-limiting step for HER^6,53. The exchange current density (j₀) values were obtained from Tafel plots²⁹. As read from Supplementary Fig. 17, the catalysts showed an increase in j_o in the following order: S-4 (10^−2.48)> Ru powder (10^−2.74)> Pt/C (10^−2.94). Durability was another critical parameter to assess the electrocatalytic performance. The durability of S-4 was also evaluated by measuring polarization curves after 10,000 CV sweeps between −1.1 and −0.9 V (versus Ag/AgCl) at 100 mV s⁻¹. As illustrated in Fig. 5b, the polarization curve of S-4 after 10,000 cycles retained an almost similar performance to the initial test, just with the overpotential increased by 4 mV, which is also slightly superior to the durability of Pt/C (8 mV) as well as much better than Ru powder. The graphene shell and the alloying of Ru and Co made great contributions to protect the catalysts from corrosion during cycling^8,37. Besides, Ru (42 $ per oz) is further more economically advantageous in price than Pt (992 $ per oz)²⁸. The Ru content of S-4 was 3.58 wt. % obtained from ICP in Table 1, which was much lower than the Pt content of 20 wt. % in Pt/C catalysts. Specific to the field of noble metal’s cost, the S-4 was only 0.76% of commercial Pt/C in price. Therefore, the S-4 sample performed higher activity, durability and economic competitiveness than the commercial Pt/C catalysts, demonstrating that it had a great potential to be a substitution of Pt/C catalyst for HER in alkaline media.

Table 2 HER electrocatalysts in alkalic media reported recently.

HER enhancement mechanism

To investigate the origin of excellent activity of RuCo alloy encapsulated in nitrogen-doped graphene layers for HER, density functional theory calculations were carried out using software of Vienna Ab Initio Simulation Package. A graphitic carbon cage C240 encapsulated 55 metal atoms was used as the basic model of graphene-encapsulated alloys, which worked well in previous studies^37,58. The detailed calculation information could be read in the Supplementary Note 1. In general, for HER performed in both acid and alkaline electrolyte, ΔG_H* is one of the key descriptors in theoretical prediction of the activity for HER. Previous studies have used |ΔG_H*| as a catalytic descriptor for HER and proposed the optimal value should be close to 0 (refs 22, 28). Therefore, the ΔG_H* of different models were calculated, including pure graphene (C₂₄₀), graphene doped with nitrogen atoms (C₂₃₉N₁), graphene-encapsulated Co (C₂₃₉N₁Co₅₅) as well as graphene-encapsulated RuCo alloys. In order to find the influence of Ru content in RuCo alloys, three established different models were abbreviated to Ru₁Co (C₂₃₉N₁Ru₁Co₅₄), Ru₂Co (C₂₃₉N₁Ru₂Co₅₃) and Ru₃Co (C₂₃₉N₁Ru₃Co₅₂), in which the metal atom ratio is comparable to experimental materials measured by ICP. The optimized adsorption structures are shown in Fig. 6, which exhibited the C atom next to the doping N atom in graphene shell is the active adsorption site for H*, which is also consistent with previous work^8,37,53. As illustrated in Fig. 7, the nitrogen doping and the combination of metal Co with graphene can significantly reduce ΔG_H*, this result was consistent with our previous work⁸. Moreover, the introduction of Ru atoms into Co metal can further decrease ΔG_H* and properly increasing Ru content will further reduce the value of ΔG_H*. As a result, Ru₃Co model exhibit the lowest ΔG_H* (0.31 eV) among graphene-encapsulated models. We also calculated H* adsorbed on Pt₅₅ and Ru₅₅ metal cluster models. Besides, some possible H* adsorption sites such as top, bridge, hcp and hollow sites are taken into considerations in our cluster models (Supplementary Figs 19 and 20). It should be noted that, during optimization process the hollow site of Pt₅₅ and top site of Ru₅₅ are tended to be bridge and hollow sites, respectively, indicating the latter are more moderate adsorption sites. The calculated ΔG_H* of different models and sites are illustrated in Supplementary Tables 2 and 4.

Figure 6: Optimized structures of H* adsorbed on nitrogen-doped graphene-encapsulated Co and RuCo alloys.

(a) Pure Co model (C₂₃₉N₁Co₅₅), (b) Ru₁Co alloy model (C₂₃₉N₁Ru₁Co₅₄) and (c) Ru₂Co alloy model (C₂₃₉N₁Ru₂Co₅₃) (d) Ru₃Co alloy model (C₂₃₉N₁Ru₃Co₅₂). The black, blue, pink, brown and white balls refer to C, N, Co, Ru and H atoms, respectively.

Figure 7: HER-free energy diagram.

ΔG_H* calculated at the equilibrium potential of different models.

The calculated results show Ru₅₅ cluster exhibited even better ΔG_H* than RuCo and Pt₅₅, which is not in good agreement with our previous experimental results based on purchased Ru powder. Unlike bulk metal, metal cluster and nano-sized metal particles have more high active sites such as facets, corners, edges. However, the Ru powder measured before was purchased from manufacturer and the size of them was in micron order as illustrated in Supplementary Fig. 18a. Besides, recent researches report nano metallic Ru-based electrocatalysts with very active performance in basic media^59,60. Therefore, we synthesized nano-sized Ru nanoparticles (Supplementary Fig. 18b) by ourselves through RuCl₃ and NaBH₄ as Precursors. The performance of nano-sized Ru catalyst was even better than our S-4 at high overpotential (Supplementary Fig. 18d–f), which was in good agreement with our calculated cluster result. For Pt cluster, even its calculated result was as good as our CoRu models; however, the measured activity was inferior to our S-4 sample. Similar results also existed in a recently published research⁵⁹, Qiao and his co-workers have proved that, apart from the ΔG_H*, the water dissociation kinetics would also affect the overall reaction rate, especially under basic electrolyte. According to their calculated results, when the kinetics of water dissociation from the Volmer step is considered, Pt exhibits a significant higher energy barrier than Ru and others, indicating sluggish water dissociation during catalytic process. Therefore, from the kinetic viewpoint, conversely, Pt did not demonstrate as good catalytic performance as depicted by calculated ΔG_H*, which might explain the inferior measured activity than our CoRu alloys. We also tried to calculate energy barrier of water dissociation process for our CoRu alloys; however, due to the very large models (C₂₃₉N₁Co₅₂Ru₃ containing 295 atoms), it’s very difficult and time-consuming to search for the exact transition state and energy barrier of ours. However, combing our calculated ΔG_H* results with Qiao’s result, it’s sufficient to give a reasonable explanation. Therefore, both the experimental and the calculated results indicated that alloying Co with small amount of Ru could obtain highly active eletrocatalyst comparable to pure Ru catalyst.

The charge-density difference of model Co and Ru₃Co was also calculated and illustrated in Fig. 8. It is shown that electrons transferred from alloy core to C atoms of graphene shell, which was beneficial to enhance C–H bond thereby lowing ΔG_H* of the model as a whole³⁷. As shown in Supplementary Table 3, the exact number of transferred electron was calculated using bader charge analysis. Ru3Co model showed the largest number of transferred electrons (5.91e⁻) among graphene-encapsulated models, therefore, ΔG_H* of it was also the lowest among them. The calculated result is consistent with the experimental results, proving the excellent HER activity originates from Ru alloying with Co atoms as well as the unique structure derived from MOFs.

Figure 8: Calculated charge-density differences of different models.

(a) Co and (b) Co₃Ru models. The isosurface value of the colour region is 0.01 e Å⁻³. The yellow and cyan regions refer to increased and decreased charge distributions, respectively.

Discussion

In summary, a novel RuCo alloy catalysts with high catalytic activity and stability for HER in basic solutions was prepared, which exceeds almost all the documented electrocatalysts including 20 wt.% Pt/C catalysts, hence enable it a cheaper alternative to Pt-based electrocatalysts for HER in basic media. The RuCo alloy structures with a low content of Ru provide it a huge economical advantage in price over commercial Pt/C catalysts. The density functional theory calculations indicate that RuCo alloy core can transfer more number of electrons to the graphene shell than pure Co metal core so as to enhance C–H bond, which will significantly decrease ΔG_H*, and thereby improve electrocatalysis activity. The results here show a new way for the development of high-performance HER electrocatalysts in alkaline media while reducing the cost of noble metal electrocatalysts.

Methods

RuCo@NC hybrids synthesis

The Co₃[Co(CN)₆]₂ MOF was prepared according to our previous researchs^41,42. The as-prepared Co₃[Co(CN)₆]₂ nanoparticles (25 mg) were dispersed in a 20 ml distilled water system under agitated stirring to get an absolutely homogeneous mixed solution, followed by the addition of 0.55, 1.1, 1.65, 2.2, 2.75 or 3.3 ml RuCl₃ solution (0.01 g ml⁻¹). After agitated stirring for 10 h in dark, the brown products were collected and rinsed several times by distilled water, and finally dried under oven at 60 °C. The thermal decomposition of the MOF precursor was performed at 600 °C for 4 h under nitrogen atmosphere in the oven with a heating rate of 10 °C min⁻¹. The obtained sample should be kept in the vacuum drying oven.

Nano Ru synthesis

Briefly, 100 mg sodium borohydride (NaBH₄) was dissolved in a 20 ml distilled water system under agitated stirring to get an absolutely transparent solution. And then the above solution was added into 20 ml RuCl₃ solution (5 mg ml⁻¹) slowly. After agitated stirring for 20 h, the black products were collected and rinsed several times by distilled water and ethanol, and finally dried under oven at 60 °C in the vacuum drying oven.

Characterization

The powder X-ray diffraction patterns of the samples were analysed with an X-ray diffractometer (Japan Rigaku D/MAX-γA) using Cu-Kα radiation (λ=1.54178 Å) with 2θ range of 20–80°. The morphology and size of all as-synthesized samples were characterized by a JEOL JSM-6700 M field-emission scanning electron microscopy and a scanning TEM (STEM; Talos F200X) with energy dispersive X-ray (EDX) spectroscopy. Raman spectrum was carried out using a LabRAM HR Evolution ranging from 1,000 to 3,000 cm⁻¹. XPS was measured on an ESCALAB 250 X-ray photoelectron spectrometer using Al Ka radiation. The specific surface and pore diameters were obtained from the results of N₂ physisorption at 77 K (Micromeritics ASAP 2020) by using the BET (Brunauer-Emmet-Teller) and BJH (Barrett-Joyner-Halenda) methods, respectively. The content of Co and Ru were obtained via the inductively coupled plasma-atomic emission spectrometer (ICP-AES) (Optima 7300 DV).

HER electrochemical measurements

The HER electrochemical perform measurements were performed in a three-electrode system on an electrochemical workstation (CHI 660D) in 1 M KOH electrolyte. Typically, 4 mg of catalyst and 30 μl Nafion solution (Sigma Aldrich, 5 wt.%) were dispersed in 1 ml of 3:1 v/v water/isopropanol mixed solvent by at least 30 min sonication to form a homogeneous ink. Then 5 μl of the dispersion (containing ∼19.4 μg of catalyst) was loaded onto a glassy carbon electrode with 3 mm diameter (loading ∼0.275 mg cm⁻²). An Ag/AgCl (filled with 3 M KCl solution) electrode and a platinum wire were served as the reference electrode and counter electrode, respectively. All of the potentials were calibrated to a RHE. The working electrode was polished with Al₂O₃ powders with size ranging from 1 to 0.05 μm. Commercial Ru powders catalysts (99.9%, Aladdin) and commercial Pt/C catalysts (20%, Alfa Aesar) were used as a reference to evaluate the electrocatalytic performance of various samples. Linear sweep voltammetry with a scan rate of 2 mV s⁻¹ was conducted between −1.1 V and −0.9 V versus AgCl/Ag electrode into N₂ saturated 1 M KOH electrolyte. CV was conducted in 1 M KOH solution in the potential region from −1.1 V to −0.9 V versus AgCl/Ag electrode at a sweep rate of 100 mV s⁻¹ for 10,000 times to investigate the cycling stability.

Data availability

The authors declare that the data supporting data supporting the findings of this study are available within the article and its Supplementary Information files.