Defective Amorphous Carbon‐Coated Carbon Nanotube‐Loaded Ruthenium Nanoparticles as Efficient Electrocatalysts for Hydrogen Production

Compared with hydrogen evolution reaction (HER) under acidic conditions, the kinetic steps of alkaline HER are more complex, which involves the adsorption and cleavage of water molecules. Defect and interface engineering are two important means to enhance basic HER. In order to prepare catalysts with high activity and stability under both acidic and basic conditions, a defective amorphous carbon loaded with Ru nanoparticles coaxially wrapped around a carbon nanotube (CNT) sponge network is presented. The presence of amorphous carbon can promote the uniform loading of Ru nanoparticles and limit the growth of Ru nanoparticles, and also the introduction of oxygen defects can regulate the electronic structure of the metal and improve its charge transport capacity, thus enhancing the catalytic performance. The catalyst CNT/C/Ru0.37 wt%‐700 prepared under optimized conditions exhibits excellent stability and activity. At a current density of 10 mA cm−2, the overpotential of HER is as low as 38.3 and 36.2 mV under alkaline and acidic conditions, respectively, which is better than most reported Ru‐based catalysts. This study details innovative and feasible ideas for the design and preparation of loaded catalysts, which can contribute to the development of high‐performance electrochemical catalysts.


Introduction
Electrochemical water decomposition as a promising sustainable hydrogen production technology has attracted wide attention.[3] However, Ru tends to aggregate due to its high cohesion energy, and is prone to fall off during testing. [4]To solve this problem, the strategy of evenly dispersing Ru nanoparticles in carbon structure was studied, which showed excellent hydrogen evolution reaction (HER) performance under both acidic and alkaline conditions.Li et al. reported the synthesis of a carbonloaded ruthenium nanoparticle electrocatalyst (Ru@CQDs) using carbon quantum dots (CQDs), which required an overpotential of 10 mV to achieve a current density of 10 mA cm À2 in alkaline solution. [5]owever, Li et al.only adjusted the catalytic performance of power catalyst Ru@CQDs under alkaline conditions.The chemical structure of CQDs and the catalytic mechanism of catalysts were not clear.Sun and co-workers dispersed Ru nanoparticles on S-doped graphene (Ru/S-rGO) and observed an overpotential of 14 mV for Ru/S-rGO-24 at a current density of 20 mA cm À2 . [6]In HER catalyst with GO as carbon support, metals were easy to oxidize and agglomerate and lose in the process of use.Porous nitrogen-doped carbon (PNC) prepared by Gao et al. using ZIF-8 as a precursor can expose more active sites and improve the utilization rate of precious metals. [7]At a current density of 10 mA cm À2 , Ru/PNC showed an overpotential of 40 mV.However, the preparation process of ZIF was complex and the yield was relatively low.In addition to the above carbon materials, carbon nanotubes (CNTs) were also commonly used as carbon carriers to compound metal Ru for HER. [8]Kweon et al. reported that a catalyst uniformly loaded with Ru nanoparticles on multiwalled carbon nanotubes (MWCNTs) showed an overpotential of 13 and 17 mV, respectively, in 0.5 M H 2 SO 4 and 1.0 M KOH with current densities of 10 mA cm À2 . [3]Zhang et al. prepared the Ru 2 P@Ru/CNT heterogeneous catalyst by a solvent-free microwave method, which showed HER performance in both alkaline electrolyte and real seawater. [9]owever, the dispersion of CNT powder was very difficult.So far, the uniform and stable dispersion of CNT powder has not been achieved, which hinders the application of CNT composite materials.
[12] They have excellent chemical stability under acidic or alkaline conditions, large surface area, and low mass density.The large specific surface area of CNT sponge is beneficial to the uniform dispersion of catalyst and improves the loading capacity of the catalyst.Our team has used CNT sponges as conductive substrates for HER.Wang et al. deposited amorphous Ni x P catalyst onto 3D porous CNT sponge substrate by electrodeposition, and prepared catalyst Ni x P/CNT with HER performance in acidic, neutral, and alkaline media. [13]Han et al. proposed a seaweed-like structure selfassembled from transition metal sulfide nanoplates on a CNT sponge network (SW-CoS@CNT) as a freestanding bifunctional catalytic electrode. [14]The above two works have not explored the real reaction mechanism of HER, and HER performance needs to be improved.
Although powder catalysts are often used for the electrolysis of water to produce hydrogen, the use of adhesive (e.g., Nafion solution) is required during the preparation of the hydrogen evolution electrodes.Adhesive coating of some active centers increases the charge transfer resistance, which reduces the catalytic performance of the catalyst.However, the 3D self-supporting catalyst eliminates adhesive influences, reduces electron transfer resistance, and simplifies electrode preparation.There are relatively less studies on the application of 3D porous carrier catalysts in HER, and further studies are needed on the reaction mechanism of catalysts and the regulation of uniform distribution of catalysts on the carrier.Here, to solve the problem of catalyst aggregation tendency on the support, we design a 3D CNT sponge electrode loaded with Ru.We introduce an amorphous carbon layer to uniformly disperse Ru nanoparticles in particular.The presence of amorphous carbon layer promotes the loading of Ru nanoparticles and limits their further growth.The presence of defects can also regulate the electronic structure of metal, improving its charge-transport capacity and catalytic performance.The synergistic effect of defects and Ru nanoparticles makes CNT/C/Ru 0.37 wt% -700 catalysts exhibit excellent catalytic activity in alkaline and acidic media.The catalyst CNT/C/ Ru 0.37 wt% -700 prepared after optimization requires an overpotential as low as 38.3 and 36.2 mV when the current density is 10 mA cm À2 under alkaline and acidic conditions.Combining defective amorphous carbon layer with 3D self-supporting substrate materials can expose more active sites and promotes electron transfer.This strategy may provide a means to prepare catalysts with excellent water splitting performance.

Results and Discussion
The CNT sponge is a conductive substrate material, prepared by chemical vapor deposition (CVD). [11]It is a 3D porous network structure consisting of interlaced CNTs. Figure 1a shows the preparation process of one of the catalytic electrode termed as CNT/C/Ru 0.37 wt% -700, where the Ru loading (0.37 wt%) and annealing temperature (700 °C) are denoted.Through a first hydrothermal process with glucose as precursor, amorphous carbon layers are coaxially wrapped around CNTs (CNT/C).The CNT/C/Ru 0.37 wt% -700 catalysts are then prepared by a second hydrothermal treatment of CNT/C and RuCl 3 followed by high-temperature annealing to uniformly load the sponge with Ru.During the process, defects such as disordered layers and oxygen intercalation have been introduced into the amorphous carbon layer, which can promote electron transfer and further improve the catalytic performance.The CNT sponge is also extremely flexible and could return to its original shape after both bending and pressing (Figure 1b).It is also hydrophobic with a water contact angle of 141°, whereas the prepared CNT/C/ Ru 0.37 wt% -700 has excellent hydrophilicity, which facilitates the penetration of electrolyte during electrochemical testing (Figure 1c).Next, we conduct morphological characterization of the material.CNT sponge has a randomly interlaced 3D structure and abundant porous (Figure S1, Supporting Information).This highly porous structure provides large surface area for loading catalyst.CNT sponges, as conductive substrates, provide a network framework for fast electron transport in electrocatalysis tests.The average diameter of CNTs is 36 nm.After hydrothermal treatment and annealing, an amorphous carbon is coaxially coated on CNTs and loaded with Ru nanoparticles, which increases the average diameter of the nanotubes to about 46 nm (Figure 2a,b).Therefore, the thickness of the loaded amorphous carbon is about 5 nm.Afterward, the catalyst CNT/C/ Ru 0.37 wt% -700 is morphologically characterized.The catalyst CNT/C/Ru 0.37 wt% -700 does not change much in its SEM image before and after annealing at 700 °C (Figure S2, Supporting Information).The transmission electron microscopy (TEM) image (Figure 2c) shows that the Ru nanoparticles are uniformly distributed on the coaxial amorphous carbon-coated CNT.The lattice spacing of 0.205 nm in the high-resolution transmission electron microscopy (HRTEM) image (Figure 2d) corresponds to the (101) crystal plane of Ru.
By changing the amount of added glucose, the coating of amorphous carbon layer and the loading of Ru nanoparticles can be adjusted.When the addition of glucose (1 mmol) is low, bare pure CNT is present in the material and the carbon layer thickness is relatively thin (Figure 2e).There is no loading of Ru particles on pure CNT, and the average size of Ru nanoparticles loaded on the thin carbon layer is 7.48 nm.For comparison, we also prepared CNT/Ru catalyst without glucose, which is almost free of Ru nanoparticles on the surface (Figure S3, Supporting Information).Therefore, the presence of amorphous carbon layer can promote the loading of Ru nanoparticles and limit the growth of too large Ru nanoparticles.When a relatively large amount of glucose (6 mmol) is added, the thickness of the carbon layer is about 15 nm, and there are carbon spheres in the material (Figure 2f ).Next, the addition amount of glucose is fixed at 3 mmol, and the load of Ru nanoparticles is adjusted by changing the addition amount of RuCl 3 .When the addition of RuCl 3 (50 mg) is too small, the loading of Ru is relatively low.When the addition of RuCl 3 (150 mg) is too much, Ru nanoparticles tend to agglomerate (Figure S4, Supporting Information).Two diffraction rings clear in the selected area electron diffraction (SAED) image of CNT/C/Ru 0.37 wt% -700 correspond to the (002) crystal plane of graphite and the (101) crystal plane of Ru, respectively (Figure 2g).To confirm the presence of binding elements, energy dispersive X-ray spectroscopy (EDX) elemental distribution analysis is performed for CNT/C/Ru 0.37 wt% -700.The figure shows that CNT/C/Ru 0.37 wt% -700 contains C, O, and Ru elements, and Ru is evenly dispersed on the tube (Figure 2h,i).The average size of Ru nanoparticles is 4.08 nm (Figure 2j).
Next, we studied the influence of several key parameters on the defect structure and the loaded Ru catalyst, including the precursor amount of glucose and RuCl 3 , and the annealing temperature.The presence of Ru diffraction peak in X-ray diffraction (XRD) indicates the successful preparation of catalyst CNT/C/ Ru 0.37 wt% -700 (Figure 3a).The strongest diffraction peak of Ru is located at 44.0°, corresponding to the (101) crystal plane of Ru, and the results are consistent with HRTEM and SAED.During the catalyst preparation, we first adjust the amorphous carbon by varying the amount of glucose added (e.g., 1, 3, and 6 mmol, respectively) for hydrothermal treatment.From the corresponding XRD (Figure S5a, Supporting Information), it can be seen that as the addition of glucose increases to 6 mmol, the amorphous carbon content in the catalyst increases and the intensity of the carbon diffraction peak decreases (26.1°).The XRD of the catalysts annealed at different temperatures all have the diffraction peaks of Ru (Figure S5b, Supporting Information).Subsequently, we also adjust the amount of RuCl 3 added (50, 100, and 150 mg).The Ru diffraction peak (JCPDS no.06-0663) gradually increases in the XRD spectrum with increasing RuCl 3 content (Figure S5c, Supporting Information).In the Raman spectrum, 1345 and 1580 cm À1 correspond to the D and G peaks of the graphitic CNTs, respectively.The intensity ratios (I D /I G ) of CNT sponge and CNT/C/Ru 0.37 wt % -700 are 0.54 and 0.81, respectively (Figure 3b).The presence of the D peak indicates that the CNTs have an appreciable amount of defects.The peak intensity of the catalyst CNT/C/Ru 0.37 wt% -700 is greatly enhanced by high temperature annealing, which means that the defects increase significantly.The catalyst I D / I G ratio obtained by changing the amount of added glucose and RuCl 3 precursor, and annealing temperature, is shown in Figure S6 and Table S1, Supporting Information.In order to further explore the electron transfer ability of Ru nanoparticles to CNT substrates with different defect contents, the work function and valence band maximum values of catalysts are calculated by ultraviolet photoelectron spectroscopy (UPS).The results show that with the increase of the temperature, the work function of the catalyst decreases gradually, and the electrons in the metal are more easily transferred (Figure S7 and Table S2, Supporting Information).To determine the type of defects, electron paramagnetic resonance (EPR) is performed on both CNT and CNT/C/Ru 0.37 wt% -700 catalysts (Figure S8a, Supporting Information).17] In addition, we also analyze the elements and their states of existence in CNT/C/Ru 0.37 wt% -700 catalysts.There are three peaks C 1s þ Ru 3d, O 1s, and Ru 3p in the X-ray photoelectron spectroscopy (XPS) spectrum of catalyst measurement (Figure 3c).Thus, the catalyst CNT/C/Ru 0.37 wt% -700 mainly contains three elements, C, O, and Ru.Then its fine spectrum is analyzed.The C 1s þ Ru 3d has five peaks in the fine spectra at 289.3, 285.7, 284.8, 285.5, and 280.7 eV corresponding to C═O, C─O, C─C, Ru 3d 3/2 , and Ru 3d 5/2 , respectively (Figure 3d). [18]owever, two peaks, C─O and ─O─C═O, are also present in the fine spectrum of O 1s at 533.2 and 531.5 eV, respectively (Figure S8b, Supporting Information).The presence of the C─O bond is evidence of the excellent hydrophilicity of the catalyst.The fine spectrum of Ru 3p is shown in Figure 3e.Due to the low Ru content in the catalyst, its XPS signal is weak.The peaks at 484.2 and 462.5 eV belong to Ru 3p 1/2 and Ru 3p 3/2 of metallic Ru, respectively.486.0 and 467.2 eV are the satellite peaks of Ru xþ . [19]The specific surface area and porosity of CNT sponge and the catalyst CNT/C/Ru 0.37 wt% -700 are determined using the 77 K N 2 adsorption/desorption isotherm method.Brunauer-Emmett-Taylor (BET) has a surface area of 216 and 141 m 2 g À1 , respectively (Figure 3f ).The pore size distribution curve shows that there are micropores (pore size < 2 nm) and mesoporous pores (pore size 2-50 nm) in CNT/C/Ru 0.37 wt% -700 (Figure S9, Supporting Information).The presence of micropores and mesoporous pores can expose more active centers.We also test the specific surface area and pore size of catalysts with different Ru contents (Table S4, Supporting Information).These results indicate that high porosity and large specific surface area provide sufficient active sites for electrocatalysis and facilitation of proton and electron transfer.
The HER performance of the catalyst is tested in a 1 M KOH alkaline electrolyte using a three-electrode system.As a comparison, we test the HER performance of pristine CNT sponges under the same conditions and find that the catalytic activity of the CNT sponges is relatively low (Figure 4a,c).We first study the effect of adjusting Ru load on its performance by changing RuCl 3 addition.The overpotentials of the CNT/C/Ru 0.31 wt% -700, CNT/C/Ru 0.37 wt% -700, and CNT/C/Ru 0.54 wt% -700 catalysts at a current density of 10 mA cm À2 are 72.1, 38.3, and 54.4 mV, respectively.The CNT/C/Ru 0.37 wt% -700 has a relatively low overpotential and the best catalytic performance.The catalyst CNT/C/ Ru 0.31 wt% -700 has a relatively low content of the catalytic active substance Ru, while the catalyst CNT/C/Ru 0.54 wt% -700 has a high Ru content and the Ru particles are easily aggregated (Figure S4, Supporting Information).The aggregation of Ru nanoparticles can affect electron transport and reduce the leakage of active sites.Therefore, compared with CNT/C/Ru 0.37 wt% -700, the catalytic performance of the two catalysts is poor.The Tafel slopes of CNT/C/Ru 0.31 wt% -700, CNT/C/Ru 0.37 wt% -700, and CNT/C/Ru 0.54 wt% -700 catalysts are 117, 80, and 113 mV dec À1 , respectively (Figure 4b).As measured by inductively coupled plasma mass spectrometry (ICP-MS), Ru content in CNT/C/ Ru 0.37 wt% -700 catalyst is 0.37 wt% (Table S3, Supporting Information).The turnover frequency (TOF) of catalysts with different Ru contents under various overpotentials is calculated in 1 M KOH (Figure S10, Supporting Information).CNT/C/ Ru 0.37 wt% -700 catalyst has a TOF of 0.44 s À1 at 100 mV, which is much better than other catalysts.The effect of amorphous carbon content on catalytic performance is also studied.When the addition of glucose is 1 and 6 mmol, respectively, the corresponding overpotential at the current density of 10 mA cm À2 is 65.8 and 47.3 mV (Figure S11a, Supporting Information).The Tafel slopes are 166 and 160 mV dec À1 , respectively (Figure S11b, Supporting Information).When the addition of glucose (1 mmol) is low, the carbon layer thickness is thinner and exposed CNTs are present in the material (Figure 2e).Because the amorphous carbon layer is thin, the limiting effect on Ru nanoparticles is weak, so the size of Ru nanoparticles (7.48 nm) loaded on the thin carbon layer is large.When a relatively large amount of glucose (6 mmol) is added, the thickness of the carbon layer is about 15 nm, and there are carbon spheres in the material (Figure 2f ).In order to further investigate the intrinsic activity of the catalyst, HER properties of the catalysts annealed at 500 °C (94.8 mV, 218 mV dec À1 ), 600 °C (53.8 mV, 186 mV dec À1 ), 700 °C (38.3 mV, 80 mV dec À1 ), and 800 °C (70.8 mV, 207 mV dec À1 ) are tested under the same conditions (Figure S11c,d, Supporting Information).The overpotential shows a trend of decreasing first and then increasing.In order to further clarify the excellent electrocatalytic hydrogen production performance of CNT/C/Ru 0.37 wt% -700 composite, electrochemical impedance spectroscopy (EIS) was performed.Nyquist diagram shows that the charge transfer resistance (R ct ) of CNT/C/Ru 0.37 wt% -700 is 2 Ω (Figure S12, Supporting Information).Obviously, CNT/C/Ru 0.37 wt% -700 has fast interfacial electron transfer kinetics and high charge transfer ability in HER process.We attribute the efficiency to a synergistic effect between Ru nanoparticles and the amorphous carbon with defects, helping to reduce charge transfer resistance and thus improve electrochemical conductivity.In order to test the stability of the catalyst in alkaline electrolyte, the linear sweep voltammetry (LSV) curves before and after 5000 cyclic voltammetry (CV) cycles are compared, and the corresponding overpotential at the current density of 10 mA cm À2 only increases by 6.9 mV (Figure 4d).After 5000 CV cycles of CNT/C/Ru 0.37 wt% -700 in 1 M KOH, Ru nanoparticles are still uniformly dispersed, and no shedding or agglomeration is observed (Figure S13, Supporting Information).After 5000 cycles of CV testing, the catalyst still has Ru diffraction peak in XRD pattern (Figure S14, Supporting Information).The results indicate that the CNT/C/ Ru 0.37 wt% -700 catalyst has outstanding stability under alkaline conditions.
Subsequently, the HER performance of the catalyst is tested in 0.5 M H 2 SO 4 acidic electrolyte.CNT sponge, as a control sample, shows poor catalytic performance in the acidic electrolyte.We adjusted the catalytic performance of the catalyst by changing the amount of RuCl 3 .The overpotentials of the CNT/C/ Ru 0.31 wt% -700, CNT/C/Ru 0.37 wt% -700, and CNT/C/Ru 0.54 wt% -700 catalysts are 96.8,36.2, and 80.3 mV at a current density of 10 mA cm À2 , respectively (Figure 4e,g).The overpotential of CNT/C/Ru 0.37 wt% -700 is relatively low, and the catalytic performance is the best.The Tafel slopes of the CNT/C/Ru 0.31 wt% -700, CNT/C/Ru 0.31 wt% -700, CNT/C/Ru 0.37 wt% -700, and CNT/C/Ru 0.54 wt% -700 in 1 M KOH.c) Column diagrams of overpotential of different catalysts at 10 mA cm À2 current density.d) LSV curves of CNT/C/Ru 0.37 wt% -700 in 1 M KOH before and after 5000 cycles.e) HER LSV curve and f ) Tafel slope of CNT, CNT/Ru, CNT/C/Ru 0.31 wt% -700, CNT/C/Ru 0.37 wt% -700, and CNT/C/Ru 0.54 wt% -700 in 0.5 M H 2 SO 4 .g) Column diagrams of overpotential of different catalysts at 10 mA cm À2 current density.h) LSV curves of CNT/C/Ru 0.37 wt% -700 in 0.5 M H 2 SO 4 before and after 5000 cycles.i) Overpotentials of CNT/C/Ru 0.37 wt% -700 at 10 mA cm À2 , with other recently reported HER electrocatalysts.
CNT/C/Ru 0.37 wt% -700 and CNT/C/Ru 0.54 wt% -700 catalysts are 158, 127, and 142 mV dec À1 , respectively (Figure 4f ).The TOF of catalysts with different Ru contents under various overpotentials is calculated in 0.5 M H 2 SO 4 (Figure S15, Supporting Information).CNT/C/Ru 0.37 wt% -700 catalyst has a TOF of 0.48 s À1 at 100 mV, which is much better than other catalysts.Electrochemical active surface area (ECSA) can be obtained based on the electrochemical double-layer capacitance (C dl ) of the catalytic surface.[22] The C dl of catalyst CNT/C/ Ru 0.37 wt% -700 is 80.2 mF cm À2 , higher than that of CNT/C/ Ru 0.31 wt% -700 (61.5 mF cm À2 ) and CNT/C/Ru 0.54 wt% -700 (49.2 mF cm À2 ) (Figure S16, Supporting Information).According to the equation ECSA = C dl /C s (C s = 0.04 mF cm À2 ), the ECSA value of the catalyst CNT/C/Ru 0.37 wt% -700 is 1002.5 cm 2 , which is much higher than the other catalysts (Table S5, Supporting Information). [23,24]To compare their specific activity, iR-corrected LSV curves are normalized using the ECSA described above (Figure S17, Supporting Information).This indicates that CNT/C/Ru 0.37 wt% -700 has higher specific activity and intrinsic activity.In order to further explore the intrinsic activity of the catalyst, the catalysts prepared by calcination at different temperatures are tested under the same conditions.The overpotential of the catalysts annealed at 500, 600, 700, and 800 °C are 141.8, 99.3, 36.2, and 189.3 mV, respectively (Figure S18, Supporting Information).The overpotential shows a trend of first decreasing and then increasing.Nyquist curves tested in acidic electrolyte show that the catalyst CNT/C/ Ru 0.37 wt% -700 has a charge transfer resistance of 4.2 Ω (Figure S19, Supporting Information), showing a fast charge transfer capability.The interfacial electron transport between Ru nanoparticles and the amorphous carbon layer with defects improves the hydrogen evolution performance of the catalyst.The current density measured in 10 mV s À1 increases from 10 to 60 mV within a voltage window of 0.14-0.24V (E versus RHE), as shown in Figure S20, Supporting Information.In Figure S20e, Supporting Information, a linear relationship between Δj (0.19 V) and the scan rate can be observed.The C dl of the catalysts annealed at 500, 600, 700, and 800 °C are 47.9, 39.9, 80.2, and 15.1 mF cm À2 , respectively.To test the stability of the catalyst in the acidic electrolyte, we subject it to 5000 cycles of CV in 0.5 M H 2 SO 4 .Comparing the LSV curves before and after cycling of the material, it is found that the corresponding overpotential hardly decays at a current density of 10 mA cm À2 (Figure 4h).After 5000 CV cycles of CNT/C/Ru 0.37 wt% -700 in 0.5 M H 2 SO 4 , Ru nanoparticles remained uniformly dispersed (Figure S21, Supporting Information).Ru diffraction peak still exists in the XRD pattern of the catalyst after 5000 CV cycles (Figure S22, Supporting Information).The results show that the CNT/C/ Ru 0.37 wt% -700 catalyst has excellent stability in acidic electrolyte.HER results obtained here are comparable or better than those obtained with some of the best current electrocatalysts.[27][28][29][30][31][32] Flexibility is one of the potential advantages of CNT/C/ Ru 0.37 wt% -700, which will better adapt to environmental changes in applications and develop flexible energy equipment.
CNT/C/Ru 0.37 wt% -700 was pressed 50 times, and HER LSV curves were tested in alkaline and acid electrolytes.The results show that after 50 times of compression, HER overpotential increased by 6.2 and 7.4 mV, respectively (Figure S23, Supporting Information).HER overpotential changes are all within 10 mV, the deformation did not significantly change the electrochemical activity of the material, and there is no obvious macroscopic damage to the sample surface.Other experiments on CNT/C/Ru 0.37 wt% -700 by compression and measured electrochemical properties have shown a slight decrease in HER LSV curve, which indicates the flexibility of our material.
To further understand the specific catalytic mechanism, a theoretical model is established (Figure 5a).The existence of defects causes electron transfer and decreases the d-band center of Ru, which promotes the water molecule adsorption/dissociation and progressive dehydrogenation of HER.In order to understand the influence of amorphous carbon on HER performance of Ru cluster, we perform first-principles calculations and plot the hydrogenation Gibbs free energy ΔG H* diagram and projected density of states (PDOS) of Ru.A cluster consisting of 55 Ru atoms sitting in a monoatomic vacancy in a monolayer of graphene is used to quantitatively represent the system in the experiment.Three typical hydrogen adsorption sites are selected as illustrated in Figure 5b.According to our calculations, the site 2 which represents sites at the interface did not favor hydrogen adsorption probably due to the repulsion from the substrate so that the catalysis processes mainly occur in the top and equator parts of the Ru cluster.As shown in Figure 5c, site 3 represents the equator parts show ΔG H* closer to 0 eV, suggesting higher catalytic efficient.Because site 1 is far from the substrate so that its catalytic property can be used to approximate the bare Ru cluster, the superior performance of site 3 over site 1 demonstrates that our theoretical results suggest the catalytic performance is improved.In order to further illustrate the effect of oxygen defect on HER properties, we establish a model of Ru loaded by oxygen-free defect CNTs and carry out corresponding DFT calculations (Figure S24, Supporting Information).The ΔG H* at site 3 of the materials with oxygen defects is closer to 0 eV than that at site 3 of the materials without oxygen defects, indicating that the presence of oxygen defects can improve the catalytic performance.Figure 5d plots the PDOS of orbitals out of the cluster surface and reveals the origin of such improvements.Because Ru has a d 6 configuration, hydrogen tends to choose the unoccupied minority spin channel to bind with Ru atom.Under the influence of interaction between amorphous carbon and Ru cluster, Ru cluster loses electrons.We calculated the Bader charge of related Ru.The Ru cluster loses 1.53 electrons in total.The Ru atoms in the equator lose 0.031 electrons per atom and the Ru in the top gain 0.063 electrons per atom.This effect upshifts the PDOS of Ru.And because site 3 is closer to the interface, the upshift is generally stronger than that of site 1.This upshift increases the energy of the orbitals that interact with hydrogen atoms, thus weakening the bond between hydrogen and Ru atoms and thus accelerating the conversion of adsorbed hydrogen to hydrogen.In summary, our calculations suggest that the charge transfer between the amorphous carbon and Ru cluster can adjust the binding strength of Ru-H to a moderate level to optimize HER catalysis.

Conclusions
In summary, we have prepared CNT coaxially wrapped by amorphous carbon with oxygen defects and loaded Ru nanoparticles by a simple method.The introduction of amorphous carbon layer can increase the loading capacity of Ru nanoparticles, and limit the growth of Ru nanoparticles.The combination of amorphous carbon and ruthenium nanoparticles with oxygen defects increases the exposure of active sites by preventing the aggregation of nanoparticles.In 1.0 M KOH and 0.5 M H 2 SO 4 solutions, the overpotentials required for CNT/C/Ru 0.37 wt% -700 catalyst to deliver a current density of 10 mA cm À2 are 38.3 and 36.2 mV, respectively.The CNT/C/Ru 0.37 wt% -700 catalyst exhibits excellent electrochemical stability under both alkaline and acidic conditions, and after 5000 CV cycles, the overpotential of the catalyst is almost constant at a current density of 10 mA cm À2 .The 3D self-supporting catalyst with low overpotential and high stability designed by us can also be extended to batteries, supercapacitors, and other applications, contributing to the development of efficient, low-cost, and environmentally friendly energy storage devices in the future.

Figure 1 .
Figure 1.Synthesis diagram, photographs, and contact angle tests of the catalyst.a) Synthesis diagram of CNT/C/Ru 0.37 wt% -700.b) A series of photographs shows that CNT sponges can return to their original shape after being pressed and bended.c) Contact angle test of CNT sponge and CNT/C/Ru 0.37 wt% -700.

Figure 5 .
Figure 5. Reaction mechanism and theoretical calculation of catalyst CNT/C/Ru 0.37 wt% -700.a) Electrocatalytic hydrogen evolution mechanism diagram of catalyst CNT/C/Ru 0.37 wt% -700.b) Three typical hydrogen adsorption sites are mapped.c) Hydrogenated Gibbs free energy ΔG H* at three different sites of Ru. d) PDOS diagram of three adsorption sites.