First-principles study of rocksalt early transition-metal carbides as potential catalysts for Li–O2 batteries

Yingying Yang , Yuelin Wang , Man Yao *, Xudong Wang and Hao Huang *
School of Materials Science and Engineering, Dalian University of Technology, Dalian 116085, China. E-mail: yaoman@dlut.edu.cn; huanghao@dlut.edu.cn; Fax: +86 411 84709248; Tel: +86 411 84707347

Received 30th October 2018 , Accepted 19th November 2018

First published on 19th November 2018


A series of early transition-metal carbides (TMCs) in the NaCl structure have been constructed to compare the catalytic activity in Li–O2 batteries by first-principles calculations. The reasonable interfacial models of LixO2 (x = 4, 2, and 1) molecules adsorbed on early TMCs surfaces were used to simulate oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) processes. Taking overpotentials as a merit parameter of catalytic activity, more relationships between material properties relative to the adsorption/desorption behavior of active molecules and catalytic activity are constructed for early TMCs. The equilibrium and charging potentials used to calculate the OER overpotentials of early TMCs are inversely proportional to the adsorption energies of (Li2O)2 and LiO2, respectively. The ORR overpotentials are inversely proportional to the adsorption energies of (Li2O)2 and LiO2 for early TMCs, but the relationship between OER overpotentials and the adsorption energies of reactive intermediates is unclear. Additionally, the overpotentials of early TMCs for ORR and OER are proportional to the desorption energies of Li+ and O2, respectively. In general, both the adsorption energy of (Li2O)2/LiO2 and desorption energy of Li+/O2 are effective characterization parameters of catalytic activity. By providing the comprehensive valuable parameters on electrochemical performance to compare the catalytic activity of early TMCs and establishing more correlations between material properties relative to the adsorption/desorption behavior of active molecules with their catalytic activity, our investigation is helpful for knowing more about the catalytic process and beneficial to screen and design novel highly active catalysts for Li–O2 batteries.


1. Introduction

In recent years, rechargeable lithium-oxygen (Li–O2) batteries have received widespread attention because of their high practically available energy density (1700 W h kg−1), which is nearly comparable to that of gasoline.1–4 Hence, they have potential applications in electric vehicles and other high-energy storage devices. The net chemistry of nonaqueous aprotic Li–O2 batteries is 2Li+ + 2e + O2 ↔ Li2O2, involving positive discharging reaction and reverse charging reaction. However, many issues such as low ORR/OER rate, poor round-trip efficiency, and short cycle life restrict their practical application, which are caused by the sluggish ORR/OER kinetics during the discharge/charge process.5–7 Electrocatalysts play an important role in accelerating the sluggish electrochemical reactions, including noble metals,8,9 bimetallic alloys,10–12 metal oxides,13–15 perovskites,16,17 graphene,18–20 transitional metal dichalcogenides (TMDCs),21,22 and transition metal nitrides (TMNs) and carbides (TMCs),23,24 which have been extensively investigated in experimental and theoretical studies.

In particular, some TMC materials reduce side reactions greatly and exhibit better reversible formation/decomposition of Li2O2. TiC-based cathodes efficiently reduce side reactions arising from electrolyte and electrode decomposition, and even maintain more than 98% capacity after 100 cycles in a rechargeable aprotic Li–O2 battery.24 The oxide layer upon a TiC(100) surface as the potential reactive state of an OER catalyst to active Li2O2 decomposition in Li–O2 batteries has also been verified by our theoretical calculation.25 The ordered mesoporous carbon material decorated by TiC as a bifunctional catalyst obviously reduces charge/discharge overpotentials and increases the specific capacity of a Li–O2 battery.26 N-doped MoxC porous nanorods were prepared as a cathode of Li–O2 batteries, in which the α-MoC1−x electrode exhibits better electrochemical performance due to its lower charge transfer resistance and better O2 adsorption.27 In addition, the hierarchically porous carbon nanofibers decorated with Mo2C nanoparticles as a cathode show a high discharge capacity (10509.6 mA h g−1), a low discharge–charge voltage gap (about 1 V), and long cycle life (up to 124 cycles) in Li–O2 batteries.28 A Mo2C/carbon nanotube composite cathode also shows high electrical efficiency (88%) and reversibility (more than 100 cycles) for Li–O2 batteries.29 Some TMCs not only exhibit good catalysis in Li–O2 batteries, but also show decent electrochemical properties in fuel cells, Li–ion batteries, Na-ion batteries, and supercapacitors.30 WC was considered as a candidate electrocatalyst to replace noble metals in low temperature fuel cells31 and WC, Mo2C, TaC, NbC were evidenced as potential electrocatalysts for hydrogen evolution reaction (HER) at medium temperatures.32 The mass activity of ORR at 0.9 V for a Pt/VC@C electrocatalyst is 2.4 times larger than that of Pt/C, which is ascribed to the synergistic effect of nanosized VC and Pt nanoparticles.33 A series of high surface area TMCs were synthesized and applied for HER and ORR, and the results indicate that Mo2C, WC, and V8C7 have enhanced HER activity and Mo2C, Cr3C2, and V8C7 have significant ORR activity.34 To summarize, some TMCs show good electrocatalytic performance in ORR, OER, and HER. Inspired by these findings, we found that all these mentioned TMCs belong to early TMCs, and there is no relevant systematic research on early TMCs for Li–O2 batteries. The reaction mechanism of ORR in fuel cells is like that of ORR and OER in Li–O2 batteries, whether the catalysis is still available for early TMCs requiring systematic investigation.

It is important to establish a basic relationship between material properties and the catalytic activity of ORR and OER in Li–O2 batteries. For Pt3M bimetallic alloys (M = 3d, 4d, and 5d transition metals), the correlations of overpotentials with adsorption energies of intermediates and d-band center for ORR and OER, respectively, are built in Li–O2 batteries.35 Surface acidity is also identified as a descriptor of catalytic activity for OER in a Li–O2 battery.36 In our previous work, we have concluded that the adsorption energies of Li and LiO2, surface energy, and the binding energy of the O atom are intrinsic properties affecting the catalytic activity of 3d-TMCs in Li–O2 batteries.37 However, whether the above mentioned properties are applicable for early TMCs needs to be verified and other relationships with catalytic activity still need to explore in Li–O2 batteries, all of which are valuable for screening and designing more active catalysts.

In this paper, a systematic first-principles study of Li2O2 formation/dissociation on early TMCs with the NaCl structure in Li–O2 batteries is presented. The focus is to understand the catalytic process more detailed by correlating catalytic activity with the adsorption/desorption behaviors of active molecules. The ORR and OER overpotentials were calculated firstly to evaluate the catalytic activity, then relationships between (Li2O)2/LiO2 adsorption energy and Li+/O2 desorption energy with ORR and OER overpotentials were established, respectively. These studies deepen our understanding of catalytic mechanisms and provide an illuminating strategy of screening and designing new catalysts that can be used in Li–O2 batteries.

2. Computational details

First-principles calculations with spin polarization were performed using periodic density functional theory (DFT) using the Vienna Ab-initio Simulation Package (VASP).38 The Perdew–Burke–Ernzerhof (PBE) generalized gradient approximation (GGA) was used to formulate the electron–electron exchange–correlation functional,39 and the projector augmented wave (PAW) potential was used to describe valence electron–ion interaction. No dipole and van der Waals (vdW) corrections were used because the dipole and vdW effects on total energies were verified to be negligible. A plane-wave cutoff energy of 500 eV and a suitable k-point Monkhorst–Pack grid of 8 × 8 × 1 were chosen by testing preferences. The stopping-criterion of energy and force for geometric optimization were set as less than 10−4 eV and 0.02 eV Å−1, respectively.

The early transition-metal carbides considered in this study crystallize in the rocksalt structure, among which ScC (at normal pressures), TiC, VC, ZrC, NbC and TaC are stable compounds, while δ-MoC and WC are metastable compounds.40 Our calculation is more inclined to choose the NaCl structure of early TMCs since smaller variation parameters should be kept when conducting trend studies. A stable (100) surface with both carbon and metal reaction sites was considered as the catalyst surface to examine their catalytic activity in this study. A 2 × 2 supercell with 7 atom layers added a vacuum of 15 Å in the Z direction to avoid unexpected interaction with neighboring slabs. The slab thickness is determined by increasing atom layers until the convergence of surface energy, which can effectively show the bulk characteristic of the slab. The top four layers allow to relax freely and the bottom three layers are fixed during geometric optimization. The parameter settings of the supercell and odd number layer keep the top layer consistent with the bottom layer in configuration for the nonpolar early TMC(100) surface.

Two different reaction pathways for Li2O2 formation and decomposition involving either in solution or on the electrode surface have been proposed.41,42 Due to the different solubilities of LiO2 in electrolyte, it has been demonstrated that O2 reduction is more likely to occur in electrolyte in high-donor-number solvents, while occur on the electrode surface in low-donor-number solvents.43 We are concerned with the influence of substrate on the intrinsic reaction of Li2O2 formation/decomposition, so mainly study the surface pathway of ORR/OER, which can be described by the stepwise adsorption/desorption of Li+ and O2. The adsorbed LixO2 (x = 4, 2, and 1) molecules on early TMCs surfaces were considered as the intermediates of ORR and OER processes. The reaction free energy of each step in Li–O2 batteries can be formulated as follows,44

ΔG = EE0 + ΔnLi(μLieU) + ΔnO2μO2
where E is the total energy of the slab at a certain step, and E0 is the total energy of the initial slab. ΔnLi and ΔnO2 are the adsorbed/desorbed numbers of Li+ and O2, respectively. μLi and μO2 represent the chemical potentials of Li and O2, which are referenced to the DFT energy of one Li atom in the Li bulk and the corrected DFT energy of an isolated O2 molecule.10,11 For the well-known overbinding error of DFT-calculated O2 binding energy,45 in the present work, the experimental value of O2 binding energy (5.12 eV)46 and the calculated O atomic energy were used to deduce the total energy of O2. The same solution was widely adopted in previous publications.14,18,19,36U is the electromotive force corresponding to the external potential in an electrochemical reaction, which can easily change the chemical potential of an electron. To ensure energy conservation, the term −eU is added on μLi to account for the energy of electron transfer at a certain external potential, this is commonly adopted in cell research,44,47,48 water photooxidation and other areas related to ORR or OER.49–51 It is worth noting that the Li2O2 formation/decomposition reaction occurs at low temperature and pressure, thus ignoring the effects of entropy (-TS) and volume (PV) on E and E0 in the above formula.

The calculated overpotentials for ORR (ηORR) and OER (ηOER) were expressed by ηORR = U0UDc and ηOER = UCU0, where UDc, U0 and UC are the discharging, equilibrium and charging potentials, respectively. More specifically, UDc is the highest potential where all discharging steps are energetically downhill, while UC is the lowest potential for which all charging steps are energetically downhill. U0 in theory is equivalent to the thermodynamic equilibrium potential, which drives ORR/OER to occur spontaneously (ΔG ≤ 0). This theoretical calculation procedure for overpotentials has been widely applied in previously published works.10,11,35,48,52

The adsorption energies (Eb) were described as Eb = Esub–adsEsubEads, where Esub–ads, Esub, Eads are the total energies of the adsorption system, substrate slab and adsorbed molecules, respectively.

The desorption energies (Ed) of Li+ were defined by the average Li+ desorption energy during LiO2* + 3(Li+ + e) ↔ (Li2O)2*, and the desorption energy of O2 is determined based on the estimated Li+ desorption energy in the step of (1) O2 + (Li+ + e) ↔ LiO2*.

3. Results and discussion

We investigated the ORR and OER overpotentials of early TMCs in the rocksalt structure. As mentioned above, some early TMCs play the same positive role in enhancing the catalysis of Li–O2 batteries and fuel cells. Here, both the ORR and OER were examined on the early TMC(100) surface firstly to determine the discharging, equilibrium and charging potentials by plotting the reaction free energy of each step, then the ORR and OER overpotenticals were obtained to evaluate the catalytic activity of early TMCs.

The ORR and OER energy diagrams of early TMC surfaces are shown in Fig. 1. Three elementary reaction steps are considered to describe the ORR/OER pathway: (1) O2 + (Li+ + e) ↔ LiO2*, (2) LiO2* + (Li+ + e) ↔ Li2O2* and (3) Li2O2* + 2(Li+ + e) ↔ (Li2O)2*, where the asterisk denotes the molecules adsorbed on early TMC surfaces. These adsorbed LixO2 (x = 4, 2, and 1) molecules on the catalyst surfaces represent the ORR/OER intermediate products. For actual Li–O2 batteries, Li2O2 formation/decomposition reaction occurs in a complex electrochemical environment, so the corresponding catalytic mechanism will be more complicated than those considered in this work. Considering that a direct simulation of the whole catalytic system is impractical, the effective interface model accurately describes all the pivotal electrochemical reactions to some extent. Indeed, all of these adsorbed intermediates have been reported in previous experimental research on reaction mechanisms,53–55 and the catalytic performance is predicted by analyzing the interaction between intermediates and the catalyst surface in relevant theoretical studies.10,11,34,48–52 Top views of the optimized LixO2 (x = 4, 2, and 1) adsorption geometries along the reaction coordinate for early TMCs are shown in Fig. 1a–h, respectively. The stable adsorption structure with a relatively low total energy is obtained after testing about 20 active sites for each adsorbed molecule, and the horizontal adsorption is more stable than the vertical adsorption. Different early TMCs have different stable adsorption structures of LixO2 (x = 4, 2, and 1) molecules as shown in Fig. 1a–h, leading to the difference of reaction free energy, therefore resulting in different potentials and overpotentials. Our calculated discharging/charging potential and ORR/OER overpotential are based on the intrinsic electrochemical reaction of Li–O2 batteries in an ideal electrochemical environment, regardless of the conductivity of Li2O2 and polarization phenomena and so on. However, if the substrate has a good catalysis of intrinsic electrochemical reaction, it is also effective in reducing experimental overpotentials under a similar electrochemical environment. So our calculated results can in essence reflect the catalytic activity of the substrate. As described above, the overall ORR process became energetically downhill from the left to right side after applying UDc, while the overall OER process became energetically downhill from the right to left side after applying UC. The specific values of discharging (UDc), equilibrium (U0), and charging (UC) potentials for each early TMC are presented in Fig. 1 and also listed in Table 1.


image file: c8cp06745g-f1.tif
Fig. 1 The calculated energy diagrams of ORR and OER processes with a top view of the optimized adsorption system on (a) ScC, (b) TiC, (c) VC, (d) ZrC, (e) NnC, (f) MoC, (g) TaC and (h) WC. Green, red, brown and other spheres indicate Li, O, C and early transition-metal atoms, respectively.
Table 1 The potentials of discharging (UDc), equilibrium (U0), and charging (UC), overpotentials of ORR (ηORR), OER (ηOER), and their sum (ηTOT) for early TMCs
ScC TiC VC ZrC NbC MoC TaC WC
U Dc (V) 1.60 1.82 1.32 1.43 1.25 1.20 1.68 1.09
U 0 (V) 2.35 2.51 2.25 2.24 2.36 2.60 3.08 3.11
U C (V) 4.07 3.70 4.01 4.27 4.51 4.69 4.94 5.17
η ORR (V) 0.75 0.69 0.93 0.81 1.11 1.40 1.40 2.02
η OER (V) 1.72 1.19 1.76 2.03 2.15 2.09 1.86 2.06
η TOT (V) 2.47 1.88 2.69 2.84 3.26 3.49 3.26 4.08


Among them, TiC has the largest UDc, while WC has the smallest UDc. The maximum value of U0 corresponds to WC, while ZrC corresponds to the smallest U0. TiC has the smallest UC, while WC has the largest UC. The largest UDc and smallest UC correspond to TiC. The minimum value of UDc and the maximum values of U0 and UC correspond to the same TMC surface. However, potentials are not the general-purpose parameters to evaluate the catalytic activity of Li–O2 batteries, which is usually evaluated by overpotentials in experimental and theoretical researches.

As mentioned above, the sluggish kinetics of OER during the charging process is a major factor affecting the catalytic activity in Li–O2 batteries. In the following, two potential parameters for calculating OER overpotentials will be discussed in Fig. 2. It is well known that the adsorption energies of intermediates in elementary reactions are common descriptors of catalytic activities in Li–O2 batteries.35,37 Meanwhile, a high-energy-density lithium-oxygen battery based on a reversible four-electron conversion to LiO2 was reported, showing that the intermediate of LiO2 is more effective than Li2O2 in improving the battery performance.56 Then we attempt to build the relationships between electrical potentials and the adsorption energies of intermediates based on the calculated results of early TMCs. As shown in Fig. 1, the determination of UDc requires considering both the reactions of (3) Li2O2* + 2(Li+ + e) ↔ (Li2O)2* and (2) LiO2* + (Li+ + e) ↔ Li2O2*, so no direct relationship was found. However, the equilibrium potential is determined by the adsorption energies of the (Li2O)2 molecule with a relational expression of y = −0.250x + 1.665 in Fig. 2(a). This indicates that equilibrium potentials are inversely proportional to the (Li2O)2 adsorption energies, namely, the weaker adsorption of (Li2O)2 leads to higher equilibrium potential of early TMCs. The energies before and after reaction of (1) O2 + (Li+ + e) ↔ LiO2* are the same in the OER process for most early TMCs except TaC, which suggest that charging potentials can be described by the adsorption energies of LiO2 species. The corresponding functional relation is y = −0.998x + 1.678 in Fig. 2(b). This reveals that the charging potentials are inversely proportional to the LiO2 adsorption energies, or in other words the weaker adsorption of LiO2 induces higher charging potentials of early TMCs. As a result, the equilibrium potentials are related to the adsorption energies of (Li2O)2 and charging potentials are associated with the LiO2 adsorption energies for early TMCs in Li–O2 batteries.


image file: c8cp06745g-f2.tif
Fig. 2 The equilibrium potentials as a function of adsorption energies of (Li2O)2 (a), and the charging potentials as a function of adsorption energies of LiO2 (b) for early TMCs.

At the same time, the adsorption energies of (Li2O)2 and LiO2 are also key factors affecting the ORR overpotentials during the discharge process for early TMCs. Fig. 3(a) shows that the ORR overpotentials are inversely proportional to the adsorption energies of (Li2O)2 with a high correlation coefficient of R2 = 0.76. This correlation means that weaker adsorption of (Li2O)2 leads to higher catalytic activity of ORR for early TMCs. Fig. 3(b) exhibits that the ORR overpotentials are inversely proportional to the LiO2 adsorption energies with a high correlation coefficient of R2 = 0.81. This relationship reveals that weaker adsorption of LiO2 induces higher catalytic activity of ORR for early TMCs. However, the relationship between OER overpotentials and the adsorption energies of the intermediates is not clear, which is not shown in Fig. 3. As a result, the catalytic activities of early TMCs for ORR can be explained by the adsorption energies of (Li2O)2 and LiO2. In our study of catalytic activities for 3d-TMCs, the calculated results also indicate that the ORR overpotentials have strong linear correlations with the adsorption energies of LiO2.37 Meanwhile, ORR overpotentials also show a negative correlation with the adsorption energy of Li2O4 in the study of rate-determining step for N-, B-doped graphene as the potential catalysts in nonaqueous Li–O2 batteries.20 Consequently, the catalytic activities of ORR for early TMCs in Li–O2 batteries are associated with the adsorption energies of LiO2.


image file: c8cp06745g-f3.tif
Fig. 3 The ORR overpotentials as a function of adsorption energies of (a) (Li2O)2 and (b) LiO2 for early TMCs.

Next, the relationships between Li+/O2 desorption energies and ORR/OER overpotential are discussed. In fact, the theoretical assumption of the OER process is based on the multistep reaction, including Li+ decomposition after breaking the Li–O bonds one by one and O2 evolution. So to some extent, it is valuable to construct the relationship between Li+/O2 desorption energies and catalytic activity for early TMCs. In this paper, the defined Li+ desorption energy is an average, and the O2 desorption energy is determined based on the estimated desorption energy of Li+. In fact, the O2 desorption energy in theory is a reasonable approximation even without considering the effect of electrolyte types, which cannot be measured experimentally. Fig. 4(a) shows that the ORR overpotentials are proportional to the Li+ desorption energies with a correlation coefficient of R2 = 0.67. This relationship means that weaker desorption of Li+ contributes to generate products easily, leading to higher catalytic activity of ORR for early TMCs. Similarly, the OER overpotentials are proportional to the O2 desorption energy with a functional relation of y = 0.758x − 0.021 in Fig. 4(a). This correlation reveals that weaker desorption of O2 helps to overcome the O2 evolution barrier, which induces higher OER catalytic activity for early TMCs. By comparing the calculated results, the O2 desorption energies are all higher than the desorption energies of Li+, which means that the O2 evolution step is more difficult to occur than the decomposition of Li+. Similarly, we have found that the O2 evolution barrier is the rate-determining barrier of OER in our study of Li2O2 decomposition on the TiC(100) surface.25 The remarkable catalytic activity of TiC is closely associated with the smallest desorption energies of Li+ and O2. Accordingly, the catalytic activities of early TMCs can be fully explained by the desorption energies of Li+ and O2. Consequently, the catalytic activities of early TMCs in Li–O2 battery are associated with the desorption energies of Li+ and O2.


image file: c8cp06745g-f4.tif
Fig. 4 The ORR overpotentials as a function of desorption energies of Li+ (a), and the OER overpotentials as a function of desorption energies of O2 (b) for early TMCs.

4. Conclusions

In this study, the catalytic activity of early transition-metal carbides in the rocksalt structure has been systematically investigated in Li–O2 batteries using first-principles calculations. Then more correlations between the adsorption and desorption behaviors of active molecules with catalytic activity are established based our calculated results of early TMCs. The equilibrium potentials of early TMCs are inversely proportional to the adsorption energies of (Li2O)2, while the charging potentials are inversely proportional to the adsorption energies of LiO2. The ORR overpotentials correlate well with the adsorption energies of (Li2O)2 and LiO2, but the relationship between OER overpotentials and catalytic activity is not intelligible. The ORR overpotentials are proportional to the desorption energies of Li+, while the OER overpotentials are proportional to the desorption energies of O2. Among them, the lowest ORR and OER overpotentials of TiC with the lowest Li+ desorption energy of 1.44 eV and O2 desorption energy of 1.58 eV. This indicates that O2 evolution is one key step of the whole electrochemical reaction. Consequently, the highly active catalyst should have weaker adsorption energies of (Li2O)2 and LiO2 and desorption energies of Li+ and O2. Our calculations and analysis on the relationship of catalytic activity and material properties yield a deep insight into understanding the catalytic process, which provide a reliable foundation for new efficient catalyst screening and design in Li–O2 batteries.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (Grant No. 21233010), the Fundamental Research Funds for the Central Universities (Grant No. DUT16ZD102), the Key Laboratory of Solidification Control and Digital Preparation Technology (Liaoning Province) and the Supercomputing Center of Dalian University of Technology.

Notes and references

  1. G. Girishkumar, B. McCloskey, A. C. Luntz, S. Swanson and W. Wilcke, J. Phys. Chem. Lett., 2010, 1, 2193–2203 CrossRef CAS.
  2. P. G. Bruce, S. A. Freunberger, L. J. Hardwick and J. M. Tarascon, Nat. Mater., 2012, 11, 19–29 CrossRef CAS PubMed.
  3. J. Christensen, P. Albertus, R. S. Sanchez-Carrera, T. Lohmann, B. Kozinskyb and R. Liedtke, J. Electrochem. Soc., 2012, 159, R1–R30 CrossRef CAS.
  4. F. Y. Cheng and J. Chen, Chem. Soc. Rev., 2012, 41, 2172–2192 RSC.
  5. Y. C. Lu, D. G. Kwabi, K. P. C. Yao, J. R. Harding, J. G. Zhou, L. Zuin and Y. Shao-Horn, Energy Environ. Sci., 2011, 4, 2999–3007 RSC.
  6. H. G. Jung, J. Hassoun, J. B. Park, Y. K. Sun and B. Scrosati, Nat. Chem., 2012, 4, 579–585 CrossRef CAS PubMed.
  7. T. Zhang and H. S. Zhou, Nat. Commun., 2013, 4, 1817 CrossRef PubMed.
  8. Z. Q. Peng, S. A. Freunberger, Y. H. Chen and P. G. Bruce, Science, 2012, 337, 563–566 CrossRef CAS PubMed.
  9. A. Kushima, T. Koido, Y. Fujiwara, N. Kuriyama, N. Kusumi and J. Li, Nano Lett., 2015, 15, 8260–8265 CrossRef CAS PubMed.
  10. R. Choi, J. Jung, G. Kim, K. Song, Y. I. Kim, S. C. Jung, Y. K. Han, H. Song and Y. M. Kang, Energy Environ. Sci., 2014, 7, 1362–1368 RSC.
  11. H. J. Kim, S. C. Jung, Y. K. Han and S. H. Oh, Nano Energy, 2015, 13, 679–686 CrossRef CAS.
  12. B. G. Kim, H. J. Kim, S. Back, K. W. Nam, Y. Jung, Y. K. Han and J. W. Choi, Sci. Rep., 2014, 4, 4225 CrossRef PubMed.
  13. H. G. Jung, Y. S. Jeong, J. B. Park, Y. K. Sun, B. Scrosati and Y. J. Lee, ACS Nano, 2013, 7, 3532–3539 CrossRef CAS PubMed.
  14. J. Z. Zhu, X. D. Ren, J. J. Liu, W. Q. Zhang and Z. Y. Wen, ACS Catal., 2015, 5, 73–81 CrossRef CAS.
  15. S. R. Cai, M. S. Zheng, X. D. Lin, M. Lei, R. M. Yuan and Q. F. Dong, ACS Catal., 2018, 8, 7983–7990 CrossRef CAS.
  16. J. G. Kim, Y. Kim, Y. Noh, S. Lee, Y. Kim and W. B. Kim, ACS Appl. Mater. Interfaces, 2018, 10, 5429–5439 CrossRef CAS PubMed.
  17. J. B. Zhang, C. F. Zhang, W. Li, Q. Guo, H. C. Gao, Y. You, Y. T. Li, Z. M. Cui, K. C. Jiang, H. J. Long, D. W. Zhang and S. Xin, ACS Appl. Mater. Interfaces, 2018, 10, 5543–5550 CrossRef CAS PubMed.
  18. X. D. Ren, J. Z. Zhu, F. M. Du, J. J. Liu and W. Q. Zhang, J. Phys. Chem. C, 2014, 118, 22412–22418 CrossRef CAS.
  19. X. D. Ren, B. Z. Wang, J. Z. Zhu, J. J. Liu, W. Q. Zhang and Z. Y. Wen, Phys. Chem. Chem. Phys., 2015, 17, 14605–14612 RSC.
  20. H. R. Jiang, T. S. Zhao, L. Shi, P. Tan and L. An, J. Phys. Chem. C, 2016, 120, 6612–6618 CrossRef CAS.
  21. P. P. Zhang, X. Y. Lu, Y. Huang, J. W. Deng, L. Zhang, F. Ding, Z. Q. Su, G. Wei and O. G. Schmidt, J. Mater. Chem. A, 2015, 3, 14562–14566 RSC.
  22. M. Asadi, B. Sayahpour, P. Abbasi, A. T. Ngo, K. Karis, J. R. Jokisaari, C. Liu, B. Narayanan, M. Gerard, P. Yasaei, X. Hu, A. Mukherjee, K. C. Lau, R. S. Assary, K. A. Fatemeh, R. F. Klie, L. A. Curtiss and S. K. Amin, Nature, 2018, 555, 502–506 CrossRef CAS PubMed.
  23. P. He, Y. G. Wang and H. S. Zhou, Chem. Commun., 2011, 47, 10701–10703 RSC.
  24. M. M. O. Thotiyl, S. A. Freunberger, Z. Q. Peng, Y. H. Chen, Z. Liu and P. G. Brace, Nat. Mater., 2013, 12, 1050–1056 CrossRef PubMed.
  25. Y. Y. Yang, X. W. Xue, Y. Qin, X. D. Wang, M. Yao, Z. H. Qin and H. Huang, J. Phys. Chem. C, 2018, 122, 12665–12672 CrossRef CAS.
  26. F. L. Qiu, P. He, J. Jiang, X. P. Zhang, S. F. Tong and H. S. Zhou, Chem. Commun., 2016, 52, 2713–2716 RSC.
  27. H. Yu, K. N. Dinh, Y. M. Sun, H. S. Fan, Y. H. Wang, Y. Jing, S. Z. Li, M. Srinivasanb and Q. Y. Yan, Nanoscale, 2018, 10, 14877–14884 RSC.
  28. Z. D. Yang, Z. W. Chang, Q. Zhang, K. K. Huang and X. B. Zhang, Sci. Bull., 2018, 63, 433–440 CrossRef CAS.
  29. W. J. Kwak, K. C. Lau, C. D. Shin, K. Amine, L. A. Curtiss and Y. K. Sun, ACS Nano, 2015, 9, 4129–4137 CrossRef CAS PubMed.
  30. Y. Zhong, X. H. Xia, F. Shi, J. Y. Zhan, J. P. Tu and H. J. Fan, Adv. Sci., 2016, 3, 1500286 CrossRef PubMed.
  31. D. J. Ham and J. S. Lee, Energies, 2009, 2, 873–899 CrossRef CAS.
  32. S. Meyer, A. V. Nikiforov, I. M. Petrushina, K. Köhler, E. Christensen, J. O. Jensen and N. J. Bjerrum, Int. J. Hydrogen Energy, 2015, 40, 2905–2911 CrossRef CAS.
  33. Z. F. Hu, C. Chen, H. Meng, R. H. Wang, P. K. Shen and H. G. Fu, Electrochem. Commun., 2011, 13, 763–765 CrossRef CAS.
  34. Y. N. Regmi, G. R. Waetzig, K. D. Duffee, S. M. Schmuecker, J. M. Thode and B. M. Leonard, J. Mater. Chem. A, 2015, 3, 10085–10091 RSC.
  35. Y. J. Kang, S. C. Jung, H. J. Kim, Y. K. Han and S. H. Oh, Nano Energy, 2016, 27, 1–7 CrossRef CAS.
  36. J. Z. Zhu, F. Wang, B. Z. Wang, Y. W. Wang, J. J. Liu, W. Q. Zhang and Z. Y. Wen, J. Am. Chem. Soc., 2015, 137, 13572–13579 CrossRef CAS PubMed.
  37. Y. Y. Yang, Y. Qin, X. W. Xue, X. D. Wang, M. Yao and H. Huang, J. Phys. Chem. C, 2018, 122, 17812–17819 CrossRef CAS.
  38. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186 CrossRef CAS.
  39. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868 CrossRef CAS PubMed.
  40. A. Vojvodic and C. Ruberto, J. Phys.: Condens. Matter, 2010, 22, 375501 CrossRef PubMed.
  41. C. O. Laoire, S. Mukerjee and K. M. Abraham, J. Phys. Chem. C, 2010, 114, 9178–9186 CrossRef CAS.
  42. B. D. McCloskey, R. Scheffler, A. Speidel, G. Girishkumar and A. C. Luntz, J. Phys. Chem. C, 2012, 116, 23897–23905 CrossRef CAS.
  43. L. Johnson, C. M. Li, Z. Liu, Y. M. Chen, S. A. Freunberger, P. C. Ashok, B. B. Praveen, K. Dholakia, J. M. Tarascon and P. G. Bruce, Nat. Chem., 2014, 6, 1091–1099 CrossRef CAS PubMed.
  44. Y. Mo, S. P. Ong and G. Ceder, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 84, 205446 CrossRef.
  45. L. Wang, T. Maxisch and G. Ceder, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 73, 195107 CrossRef.
  46. M. D. Radin, J. F. Rodriguez, F. Tian and D. J. Siegel, J. Am. Chem. Soc., 2012, 134, 1093–1103 CrossRef CAS PubMed.
  47. Y. Xu and W. A. Shelton, J. Chem. Phys., 2010, 133, 024703 CrossRef PubMed.
  48. J. S. Hummelshøj, J. Blomqvist, S. Datta, T. Vegge, J. Rossmeisl, K. S. Thygesen, A. C. Luntz, K. W. Jacobsen and J. K. Nørskov, J. Chem. Phys., 2010, 132, 071101 CrossRef PubMed.
  49. S. Lin, X. X. Ye, X. M. Gao and J. Huang, J. Mol. Catal. A: Chem., 2015, 406, 137–144 CrossRef CAS.
  50. X. X. Gao, S. J. Wang and S. Lin, ACS Appl. Mater. Interfaces, 2016, 8, 24238–24247 CrossRef PubMed.
  51. Q. Wan, F. F. Wei, Y. Q. Wang, F. T. Wang, L. S. Zhou, S. Lin, D. Q. Xie and H. Guo, Nanoscale, 2018, 10, 17893–17901 RSC.
  52. J. S. Hummelshøj, A. C. Luntz and J. K. Nørskov, J. Chem. Phys., 2013, 138, 034703 CrossRef PubMed.
  53. Z. Q. Peng, S. A. Freunberger, L. J. Hardwick, Y. H. Chen, V. Giordani, F. Bardé, P. Novák, D. Graham, J. M. Tarascon and P. G. Bruce, Angew. Chem., 2011, 123, 6475–6479 CrossRef.
  54. B. D. McCloskey, R. Scheffler, A. Speidel, G. Girishkumar and A. C. Luntz, J. Phys. Chem. C, 2012, 116, 23897–23905 CrossRef CAS.
  55. J. Hassoun, F. Croce, M. Armand and B. Scrosati, Angew. Chem., Int. Ed., 2011, 50, 2999–3002 CrossRef CAS PubMed.
  56. C. Xia, C. Y. Kwok and L. F. Nazar, Science, 2018, 361, 777–781 CrossRef CAS PubMed.

This journal is © the Owner Societies 2018