New horizons of MBenes: highly active catalysts for the CO oxidation reaction

Abstract

The search for materials with high intrinsic carbon monoxide oxidation reaction (COOR) catalytic activity is critical for enhancing the efficiency of reducing CO contamination. COOR catalysts, however, have long relied heavily on noble metals and CeO₂. Herein, in order to search for non-noble COOR catalysts that are more active than CeO₂, 18 oxygen-functionalized MBenes with orthorhombic and hexagonal crystal structures, denoted as orth-M₂B₂O₂ and hex-M₂B₂O₂ (M = Ti, V, Cr, Zr, Nb, Mo, Hf, Ta and W), were investigated in terms of their COOR catalytic activity by high-throughput first-principles calculations. Hex-Mo₂B₂O₂, orth-Mo₂B₂O₂, hex-V₂B₂O₂ and hex-Cr₂B₂O₂ were found to be more active than CeO₂ and possess structural stability below 1000 K, showing the potential to replace CeO₂ as the substrates of COOR catalysts. Moreover, orth-Mo₂B₂O₂, hex-V₂B₂O₂ and hex-Cr₂B₂O₂ exhibit even higher COOR catalytic activity than Pt–CeO₂ and Au–CeO₂, and are expected to be applied as COOR catalysts directly. Further investigations showed that the formation energy of oxygen vacancies could be used as the descriptor of COOR catalytic activity, which would help to reduce the amount of calculations significantly during the catalyst screening process. This work not only reports a series of 2D materials with high COOR catalytic activity and opens up a new application area for MBenes, but also provides a reliable strategy for highly efficient screening for COOR catalysts.

---

Introduction

Carbon monoxide (CO) is one of the major air pollutants and is not only toxic to human beings, but also results in the poisoning of Pt-based catalysts in proton-exchange membrane fuel cells (PEMFC).^1,2 CO originates mostly from the exhaust gas of internal combustion engines,³ and one of the most common methods for CO treatment is the carbon monoxide oxidation reaction (COOR), in which CO is oxidized to carbon dioxide (CO₂) at low temperature in an O₂ atmosphere without combustion.^4–7 There exist three typical mechanisms for COOR catalysis: the Langmuir–Hinshelwood (L–H) mechanism, the Eley–Rideal (E–R) mechanism and the Mars–van Krevelen (M–vK) mechanism.³ In the L–H and E–R mechanisms, all the CO molecules are oxidized by O₂ molecules and the lattices of catalysts remain intact during the whole catalytic process. As illustrated in Scheme 1, in the L–H mechanism, the catalysts adsorb both CO and O₂ molecules, while in the E–R mechanism, the catalysts only adsorb O₂ molecules. However, in the M–vK mechanism, the CO molecules are oxidized by both O₂ molecules and the lattice oxygen atoms on the surfaces of the catalysts, which means that the catalysts are involved in the elementary reactions directly. During the catalytic process, oxygen vacancies (Va_O) are generated when the lattice oxygen atoms are used to oxidize CO molecules, and they will be filled with O₂ molecules in the next elementary step.^8,9 In general, on catalysts with metal active sites (including single atoms, clusters and nanoparticles) and inert substrates (such as graphene⁷), the COOR occurs via the L–H and E–R mechanisms, while only on catalysts containing accessible lattice oxygen atoms could the COOR occur via the M–vK mechanism.

Scheme 1 Three different mechanisms of COOR catalysis.

So far, noble metals, such as Pt, Pd, and Au, are indispensable for most COOR catalysts. To reduce the use of noble metals and control the cost, one popular strategy is to load single atoms or nanoparticles of noble metals or their oxides on non-noble substrates, such as Pt/CeO₂, Pd/CeO₂ and Au/CeO₂,^5,8–23 which have been universally adopted in three-way catalytic converters (TWCs) in the exhaust systems of vehicles to remove CO, which is one of the major harmful gases. The most commonly used COOR catalysts is CeO₂-supported noble metals because the lattice oxygen atoms in CeO₂ could be easily utilized, making the COOR occur via the M–vK mechanism with high activity.^{8,11,14,15,19,24–26} The intrinsic catalytic activity of substrates contributes a lot to the overall performance of COOR catalysts since the substrates are involved in the oxidation of CO directly via the M–vK mechanism. However, the intrinsic catalytic activity of CeO₂ is relatively poor.³ Therefore, the search for materials with higher intrinsic COOR catalytic activity to replace CeO₂ will be beneficial for enhancing the COOR catalytic performance of combined catalysts.

Lattice oxygen atoms are indispensable to the intrinsic COOR catalytic activity of non-noble catalysts.^3,27 In the past decade, two-dimensional transition metal carbides/nitrides/carbonitrides, MXenes, have been widely studied as catalysts for various reactions. The exposed transition metal atoms on the surfaces of MXenes are easily functionalized by oxygen atoms which can tune their properties.^28–30 According to the previous research by Tahini et al.,²⁷ the interaction between the transition metal atoms and the surface oxygen atoms in Cr₃C₂O₂, V₃C₂O₂ and Sc₃C₂O₂ MXenes is relatively weak, and thus the surface oxygen atoms could be easily utilized to oxidize CO molecules. By studying the COOR catalytic activity, it was found that these O-functionalized MXenes are more active than CeO₂via the M–vK mechanism. In 2017, MBenes, a family of two-dimensional transition metal borides, were reported,³¹ and they possess similar structures to MXenes.³² MBenes are widely studied as catalysts or substrates of various reactions, including the hydrogen evolution reaction (HER),^31,33–36 the oxygen evolution reaction (OER)/oxygen reduction reaction (ORR),^34,37 the nitrogen reduction reaction (NRR)^38–40 and the CO₂ reduction reaction (CO₂RR).^41,42 Similar to the situation for MXenes, the exposed transition metal surfaces of pristine MBenes can be easily functionalized by O atoms.^35,37,43,44 Therefore, it is anticipated that the surface oxygen atoms on O-functionalized MBenes could also be utilized to oxidize CO molecules leading to high COOR catalytic activity via the M–vK mechanism.

In this work, 18 O-functionalized MBenes, denoted as orth-M₂B₂O₂ and hex-M₂B₂O₂, where M represents Ti, V, Cr, Zr, Nb, Mo, Hf, Ta, and W, and “orth/hex” indicates orthorhombic/hexagonal crystal systems, were investigated to explore their intrinsic COOR catalytic activity and mechanism by means of the first-principles method. Our calculated results show that five M₂B₂O₂ compounds demonstrate higher COOR catalytic activity than CeO₂ at room temperature and good thermal stability at a high temperature of 1000 K, among which, orth-Mo₂B₂O₂, orth-W₂B₂O₂, hex-V₂B₂O₂ and hex-Cr₂B₂O₂ are eligible to serve as independent COOR catalysts. Moreover, it has been found that oxygen vacancy formation energy could reflect the maximum reaction energy for the COOR directly. Therefore, by taking oxygen vacancy formation energy as the descriptor for COOR catalytic activity, the screening efficiency for COOR catalysts could be highly improved.

Computational methods

All first-principles calculations were performed based on density functional theory (DFT) using the Vienna ab initio Simulation Package (VASP)^45,46 with the projector-augmented wave (PAW) method.⁴⁷ The interaction between ion cores and valence electrons was described using the Perdew–Burke–Ernzerhof (PBE) functional.^48,49 All the atomic structures were fully relaxed under the energy and force convergence criteria of 10⁻⁴ eV per atom and 0.02 eV per angstrom, respectively. For the calculation of energy, 3 × 3 × 1 supercells were used, while for ab initio molecular dynamics (AIMD) simulation, 4 × 4 × 1 supercells were applied, during which a canonical NVT ensemble was used, and the temperature was controlled with the algorithm of Nosé.⁵⁰

In this work, COOR catalysis on 18 orth/hex-M₂B₂O₂via the M–vK mechanism was studied, which involves three phases: the adsorption of O₂ (I:* + O₂ → *OO), the formation of the first CO₂ (II: *OO + CO → *OOCO and III: *OOCO → CO₂ + *O) and the formation of the second CO₂ (IV: *O + CO → *CO₂ and V: *CO₂ → * + CO₂). Here * represents M₂B₂O₂ with O vacancies. The Gibbs free energy of steps I∼V is denoted as ΔG₁ ∼ ΔG₅. The computational details of the energy terms are included in the ESI.†

Results and discussion

The 18 orth/hex-M₂B₂ compounds studied in this work have been proved to be stable dynamically according to the previous research studies.^{31–33,37,51–55} Experimentally, oxygen functional groups have been detected on the surface of hex-Mo_4/3B₂ MBenes.⁵⁶ Meanwhile, theoretical studies of O-functionalized MBenes have been carried out in many studies.^{31,35,36,43,54,57,58} First of all, we studied the atomic structures of 18 M₂B₂O₂ and the corresponding formation energy of oxygen functional groups. There exist four and two sites with different chemical environments for O atoms binding on the surfaces of orth-M₂B₂ and hex-M₂B₂, respectively (Fig. S1(a) and S2(a)†). Considering the symmetry of orth-M₂B₂ and hex-M₂B₂, we investigated four types of atomic structures for orth-M₂B₂, including types I–III shown in Fig. 1(a–c) and type IV shown in Fig. S1(b),† and three types of atomic structures for hex-M₂B₂, including types A and B shown in Fig. 1(d and e) and type C shown in Fig. S2(b).† After studying the total energy of all the structures of the 18 M₂B₂O₂, the most stable structures for each are shown in Fig. 1. There are three types of atomic structures for orth-M₂B₂O₂: type-I for M = Ti, Zr, Hf, Mo; type-II for M = V, Nb, Ta, W; and type-III for M = Cr, while there are two different atomic structures for hex-M₂B₂O₂: type-A for M = Ti, Zr, Hf, Mo, W and type-B for M = V, Nb, Ta, Cr. For the most stable structure for each M₂B₂O₂, the formation energy of the oxygen functional group (ΔG^O_f) was calculated with eqn (S4),† and the results are listed in Table S2.† For all 18 M₂B₂O₂, the values of ΔG^O_f are negative, indicating the spontaneity of the formation of oxygen functional groups under an O₂ atmosphere thermodynamically. These surface oxygen atoms could serve as the oxidants of CO molecules in COOR catalysis.

Fig. 1 The optimized atomic structures of (a) type-I, (b) type-II, (c) type-III for orth-M₂B₂O₂ and (d) type-A, (e) type-B for hex-M₂B₂O₂ in side (up) and top (down) views. Green and red balls represent B and O atoms, respectively, and the balls with other colors represent transition metal atoms.

In general, the surface vacancies on O-functionalized MXenes are inevitable.^28,59 Therefore, we considered the existence of oxygen vacancies on the surface of M₂B₂O₂ in this work. Here the concentration of O vacancies is considered to be 5.56%, corresponding to one O vacancy in a 3 × 3 × 1 supercell of M₂B₂O₂. The formation energy of oxygen vacancies, ΔG^Va_f, was calculated with eqn (S5)† and the results are listed in Table S2.† The results show that the values of ΔG^Va_f for the 18 M₂B₂O₂ range from 2.55 eV to 5.59 eV. In addition, the values of ΔG^Va_f for two O-functionalized MXenes, Ti₂CO₂ and Ti₃C₂O₂, were calculated to be 4.5 eV and 5.0 eV (the concentration of O vacancies = 5.56%), respectively, which are higher than those of most of the M₂B₂O₂, as displayed in Fig. 2. Since O vacancies on Ti₂CO₂ and Ti₃C₂O₂ are very common in experiments,^28,60 the M₂B₂O₂ studied here can also easily become defective with O vacancies, which could capture CO molecules.

Fig. 2 The formation energy of oxygen vacancies of 18 M₂B₂O₂, in comparison with those of Ti₂CO₂ and Ti₃C₂O₂.

From Fig. 2, it is found that for orth/hex-M₂B₂O₂ with M in the same group, the ΔG^Va_f values increase with increasing atomic number (Z), while for M₂B₂O₂ with M in the same period, the ΔG^Va_f values decrease with increasing electron layer number (n). Therefore, orth/hex-Cr₂B₂O₂ has the lowest ΔG^Va_f value, while orth/hex-Hf₂B₂O₂ has the highest ΔG^Va_f value. Apparently, the variation of ΔG^Va_f is in accordance with the periodic law of elements: the larger Z and the lower n M has, the lower ΔG^Va_f value M₂B₂O₂ has.

The electrons on the surfaces of defective M₂B₂O₂ are not evenly distributed due to the existence of O vacancies, which are easy to fill with gaseous molecules. According to the results of Bader charge calculations, the transition metal atoms nearest to the O vacancies have lower coordination number than the other transition metal atoms, and thus they possess more electrons. The extra electrons could transfer to the 2p orbitals of the O atoms in O₂ or CO molecules to drive the electron distribution of the transition metal layers equally. Theoretically, both O₂ and CO molecules could adsorb at the O vacancy sites. However, if CO molecules fill the O vacancies, the C atoms would be left on the surface, which could not be removed by combining with O₂ to form CO₂. This would result in the blockage of O₂ adsorption sites and deactivation of the catalysts. Therefore, we investigated the adsorption selectivity of O vacancies by calculating the adsorption energy of O₂ and CO, as shown in Fig. 3(a). Apparently, the adsorption of O₂ on all 18 defective M₂B₂O₂ is highly spontaneous with the O₂ adsorption energy being lower than −1 eV, while the values of CO adsorption energy are all higher than 0. Taking orth-Cr₂B₂O₂ and hex-Cr₂B₂O₂ as examples, after adsorbing CO, the O atoms of CO could not fill the O vacancies and the CO molecules would stay above the surfaces, as displayed in Fig. S3(a and b).† Therefore, all 18 defective M₂B₂O₂ compounds have higher adsorption selectivity towards O₂ than towards CO, avoiding the deactivation of catalysts.

Fig. 3 (a) The adsorption energy of O₂ and CO on 18 defective M₂B₂O₂. (b)The structures after O₂ adsorption on hex-Cr₂B₂O₂ (up) and orth-Cr₂B₂O₂ (down). Blue, green and red balls represent transition metal, B and O atoms, respectively. (c) The relationship between the O–O bond length (Δd) and the numbers of electrons that the O₂ molecules obtains from the substrates (N_e).

As shown in Fig. 3(b), an O₂ molecule adsorbs on the O vacancy site perpendicularly to the surfaces of most M₂B₂O₂. However, for orth-Ta₂B₂O₂ and hex-Zr₂B₂O₂, O₂ molecules are parallel with the surfaces (Fig. S3(c and d)†), while for hex-Hf₂B₂O₂, after the vacancy is filled with one O atom of O₂, the other O atom adsorbs on the top of the Hf atom (Fig. S3(e)†). To study the mechanism, we further calculated the change in the O–O bond length (Δd) and the number of electrons that the O₂ molecules obtain from the substrates (N_e), which are listed in Table S3.† The O–O bond length of O₂ increases on all 18 M₂B₂O₂ after adsorption, indicating that the O₂ molecules are activated by M₂B₂O₂ and the O–O bonds become weaker. By analyzing the relationship between Δd and N_e, it is found from Fig. 3(c) that Δd is approximately proportional to N_e, implying that the more electrons the O₂ molecules obtain, the stronger the O₂ activation degree is, contributing to a longer O–O bond.

After the adsorption of O₂, the outer O atom combines with the gaseous CO molecule, thus forming the OOCO intermediates (step II). This step is highly spontaneous for all 18 M₂B₂O₂ compounds, with ΔG₂ ranging from −4 eV to −6 eV. It is notable that the OOCO intermediates exist as separate adsorbed O and isolated OCO from the M₂B₂O₂ substrates, as illustrated in Fig. 4(a). The isolated OCO is located at a distance of over 2.8 Å above the surfaces of M₂B₂O₂, which suggests that the binding strength between the M₂B₂O₂ substrates and the outer O atom is extremely weak. Consequently, it is very easy for the outer O atoms to detach from the substrates and combine with the CO molecules, which is attributed to the O₂ activation by substrates. Further study of Bader charge indicated that there is barely electron transfer between the substrates and OCO. In other words, no chemisorption exists between the two moieties and the binding strength is relatively low. Therefore, step III is also exothermic, generating gaseous CO₂ molecules with ΔG₃ ranging from 0 to −0.5 eV. The values of Gibbs free energy change for the formation of the first CO₂ (ΔG₂ + ΔG₃) are listed in Table S4.†

Fig. 4 (a) Gibbs free energy profile for steps II and III. (b) ΔG_max for COOR of orth-V₂B₂O₂, orth-Cr₂B₂O₂, orth-Nb₂B₂O₂, orth-Mo₂B₂O₂, orth-W₂B₂O₂, hex-V₂B₂O₂, hex-Cr₂B₂O₂, hex-Mo₂B₂O₂, compared with CeO₂, Pt–CeO₂ and Au–CeO₂.

After desorption of the first CO₂, the surfaces of substrates become complete and flat, and all the O atoms could serve as the adsorption sites for CO molecules. During this process, the values of ΔG₄ are negative for orth-V₂B₂O₂ (−0.59 eV), orth-Cr₂B₂O₂ (−0.56 eV), orth-Mo₂B₂O₂ (−0.60 eV) and hex-Cr₂B₂O₂ (−0.87 eV). For these materials, the subsequent CO₂ desorption process requires more energy input with ΔG₅ ranging from −0.28 eV to 0.1 eV, and thus ΔG₅ is the maximum Gibbs free energy change (ΔG_max) for the whole COOR. For the other M₂B₂O₂ compounds, the adsorption of the second CO molecules is endothermic and the most energy-consuming step, among which orth-Nb₂B₂O₂, orth-W₂B₂O₂, hex-V₂B₂O₂ and hex-Mo₂B₂O₂ have ΔG_max lower than 1 eV. The values of ΔG₄ and ΔG₅ for all 18 M₂B₂O₂ are listed in Table S5.† Besides, the COOR catalytic activity of CeO₂ was also studied. The supercell of CeO₂ contains 18 surface O atoms, which is the same as for the M₂B₂O₂ studied in this work, and the ΔG_max for CeO₂ is calculated to be 0.93 eV. It was found that the features in terms of reaction energy are similar for CeO₂ and M₂B₂O₂: easy adsorption of O₂, easy formation of the first CO₂, but hard formation of the second CO₂. As shown in Fig. 4(b), the eight M₂B₂O₂ mentioned above are more active than CeO₂ for COOR catalysis. Moreover, except hex-Mo₂B₂O₂, these M₂B₂O₂ compounds possess COOR catalytic activities even higher than those of CeO₂-based single atom catalysts, Pt–CeO₂ and Au–CeO₂ (ΔG_max = 0.60 eV and 0.45 eV, respectively), which have been considered as highly active COOR catalysts experimentally according to the previous studies.^11,26 Since the COOR is widely used for CO removal in exhaust gas, the catalysts should maintain the stability of their structures at high temperature. By ab initio molecular dynamics (AIMD) simulation at 1000 K for 10 ps, hex-Mo₂B₂O₂, orth-Mo₂B₂O₂, hex-V₂B₂O₂ and hex-Cr₂B₂O₂ were found to be stable with tiny energy fluctuation and structural distortion, as seen in Fig. S4.† Therefore, we conclude that hex-Mo₂B₂O₂, orth-Mo₂B₂O₂, hex-V₂B₂O₂ and hex-Cr₂B₂O₂ could all serve as the substitutes for CeO₂ as the substrates of COOR catalysts. Furthermore, orth-Mo₂B₂O₂, hex-V₂B₂O₂ and hex-Cr₂B₂O₂ are expected to be used as COOR catalysts directly.

As the COOR catalysis involves five elementary steps, it is very time-consuming to calculate the Gibbs free energy changes for all the elementary steps to screen the desired COOR catalysts from many candidates. Therefore, to simplify the screening process and reduce the amount of calculations, here we tried to find out a general descriptor for the COOR catalytic activity by investigating the relationship between ΔG_max of M₂B₂O₂ and species of M. In Fig. 5(a and b) the ΔG_max values of orth-M₂B₂O₂ and hex-M₂B₂O₂ are shown in the form of a three-dimensional bar graph. It can be seen that the values of ΔG_max vary with the species of transition metal elements, in line with the periodic law of elements: for M in the same period, ΔG_max of M₂B₂O₂ decreases with increasing atomic number (Z), while for M in the same group, ΔG_max of M₂B₂O₂ becomes higher with increasing electron layer number (n). Therefore, orth/hex-Cr₂B₂O₂ has the smallest ΔG_max while orth/hex-Hf₂B₂O₂ has the largest ΔG_max.

Fig. 5 ΔG_max for COOR of (a) orth-M₂B₂O₂ and (b) hex-M₂B₂O₂ with M in different groups and periods. (c) Relationship between ΔG_max and ΔG^Va_f of orth/hex-M₂B₂O₂.

As discussed above, the periodic law of elements was also found in the variation of ΔG^Va_f with species of M. Therefore, we correlate ΔG_max with ΔG^Va_f, and the results are demonstrated in Fig. 5(c). The data dots in the figure indicate that there exists close correlation between ΔG_max and ΔG^Va_f. By linearly fitting the dot data, it is found that the determination coefficient, i.e., R², reached 0.94, indicating the high degree of linear relationship between the two parameters. Meanwhile, the ΔG^Va_f value of CeO₂ with an O vacancy density of 5.56% was calculated to be 3.94 eV. The coordinates (ΔG^Va_f, ΔG_max) for CeO₂ are indicated with the blue star in Fig. 5(c), which is clearly very close to the fitted line. Therefore, the oxygen vacancy formation energy can directly reflect the COOR catalytic activity of materials. Generally, the lower the ΔG^Va_f, the lower the ΔG_max, which implies higher COOR catalytic activity.

By using ΔG^Va_f as the descriptor for the COOR catalytic activity, the screening of COOR catalysts with high activity will become much easier because only the values of ΔG^Va_f need to be calculated while the time-consuming calculations of ΔG for all the elementary steps are unnecessary. Subsequently, the investigation of the COOR catalytic activity of 9 M₂CO₂ MXenes was conducted, where M represents the same transition metal atoms as those of M₂B₂O₂ in this work. Here, we calculated the ΔG^Va_f of the 9 M₂CO₂ first, which shows the periodic variation with M, the same as in the case of M₂B₂O₂. Among the 9 M₂CO₂, only Cr₂CO₂ and V₂CO₂ have lower ΔG^Va_f than CeO₂. Then, by calculating ΔG_max, it was found that Cr₂CO₂ and V₂CO₂ are the two most active catalysts with the values of ΔG_max being −0.2 eV (step II) and 0.26 eV (step I), respectively, as illustrated in Fig. S5.† Except Cr₂CO₂ and V₂CO₂, the other M₂CO₂ compounds have relatively low intrinsic COOR catalytic activity via the M–vK mechanism. For example, when used as a substrate of COOR catalysts, the surface oxygen atoms of Ti₂CO₂ ⁶ are not involved in the elementary reactions and thus the COOR occurs on Ti₂CO₂-based catalysts via the L–H or E–R mechanism rather than the M–vK mechanism. Therefore, our findings are applicable to the case for M₂CO₂, which once again supports the conclusion that ΔG^Va_f can serve as the descriptor for COOR catalytic activity.

Conclusions

By investigating the COOR catalytic activity of 18 M₂B₂O₂ with first-principles methods, hex-Mo₂B₂O₂, orth-Mo₂B₂O₂, hex-V₂B₂O₂, and hex-Cr₂B₂O₂ were found to possess high COOR catalytic activity via the M–vK mechanism and high thermal stability under 1000 K. Hex-Mo₂B₂O₂ is expected to serve as the substrate of COOR catalysts as an alternative to CeO₂, and orth-Mo₂B₂O₂, hex-V₂B₂O₂, and hex-Cr₂B₂O₂ are expected to be used as direct COOR catalysts with higher activity. Furthermore, it was found that the formation energy of O vacancies (ΔG^Va_f) can reflect the value of maximum Gibbs free energy change (ΔG_max) for COOR catalysis and serve as the descriptor for COOR catalytic activity. In general, materials with lower ΔG^Va_f have lower ΔG_max. Thus, during the process of COOR catalyst screening, only ΔG_max needs to be calculated, thereby saving large amounts of calculation cost and contributing to high screening efficiency.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (no. 51871009 and 51872017).

References

1. S. Jiménez, J. Soler, R. X. Valenzuela and L. Daza, J. Power Sources, 2005, 151, 69–73.
2. J. J. Baschuk and X. Li, Int. J. Energy Res., 2001, 25, 695–713.
3. H. J. Kim, M. G. Jang, D. Shin and J. W. Han, ChemCatChem, 2020, 12, 11–26.
4. B. K. Min and C. M. Friend, Chem. Rev., 2007, 107, 2709–2724.
5. B. Qiao, A. Wang, X. Yang, L. F. Allard, Z. Jiang, Y. Cui, J. Liu, J. Li and T. Zhang, Nat. Chem., 2011, 3, 634–641.
6. X. Zhang, J. Lei, D. Wu, X. Zhao, Y. Jing and Z. Zhou, J. Mater. Chem. A, 2016, 4, 4871–4876.
7. Y. Li, Z. Zhou, G. Yu, W. Chen and Z. Chen, J. Phys. Chem. C, 2010, 114, 6250–6254.
8. G. Spezzati, A. D. Benavidez, A. T. DeLaRiva, Y. Su, J. P. Hofmann, S. Asahina, E. J. Olivier, J. H. Neethling, J. T. Miller, A. K. Datye and E. J. M. Hensen, Appl. Catal., B, 2019, 243, 36–46.
9. P. Schlexer, D. Widmann, R. J. Behm and G. Pacchioni, ACS Catal., 2018, 8, 6513–6525.
10. C. Cheng, X. Zhang, Z. Yang and Z. Zhou, ACS Appl. Mater. Interfaces, 2018, 10, 32903–32912.
11. Z. Jiang, P. Wang, X. Jiang and J. Zhao, Nanoscale Horiz., 2018, 3, 335–341.
12. F. Li, Y. Li, X. C. Zeng and Z. Chen, ACS Catal., 2015, 5, 544–552.
13. Y. Lou, Y. Cai, W. Hu, L. Wang, Q. Dai, W. Zhan, Y. Guo, P. Hu, X.-M. Cao, J. Liu and Y. Guo, ACS Catal., 2020, 10, 6094–6101.
14. Y.-Q. Su, I. A. W. Filot, J.-X. Liu and E. J. M. Hensen, ACS Catal., 2018, 8, 75–80.
15. X.-m. Zhang, P. Tian, W. Tu, Z. Zhang, J. Xu and Y.-F. Han, ACS Catal., 2018, 8, 5261–5275.
16. C. Cheng, X. Zhang, Z. Yang and K. Hermansson, Adv. Theory Simul., 2019, 2, 1900006.
17. S. Y. Hwang, E. Yurchekfrodl, C. Zhang and Z. Peng, ChemCatChem, 2016, 8, 97–101.
18. X. Zhang, C. Xu, Y. Zhang, C. Cheng, Z. Yang and K. Hermansson, Int. J. Hydrogen Energy, 2021, 46, 8477–8485.
19. J.-X. Liu, Y. Su, I. A. W. Filot and E. J. M. Hensen, J. Am. Chem. Soc., 2018, 140, 4580–4587.
20. M. Moses-DeBusk, M. Yoon, L. F. Allard, D. R. Mullins, Z. Wu, X. Yang, G. Veith, G. M. Stocks and C. K. Narula, J. Am. Chem. Soc., 2013, 135, 12634–12645.
21. A. J. Therrien, A. J. R. Hensley, M. D. Marcinkowski, R. Zhang, F. R. Lucci, B. Coughlin, A. C. Schilling, J.-S. McEwen and E. C. H. Sykes, Nat. Catal., 2018, 1, 192–198.
22. E. J. Peterson, A. T. DeLaRiva, S. Lin, R. S. Johnson, H. Guo, J. T. Miller, J. Hun Kwak, C. H. F. Peden, B. Kiefer, L. F. Allard, F. H. Ribeiro and A. K. Datye, Nat. Commun., 2014, 5, 4885.
23. C. Cheng, X. Zhang, M. Wang, S. Wang and Z. Yang, Phys. Chem. Chem. Phys., 2018, 20, 3504–3513.
24. Y.-L. Song, L.-L. Yin, J. Zhang, P. Hu, X.-Q. Gong and G. Lu, Surf. Sci., 2013, 618, 140–147.
25. G. Glaspell, H. M. A. Hassan, A. Elzatahry, V. Abdalsayed and M. S. El-Shall, Top. Catal., 2008, 47, 22–31.
26. Y. Feng, Q. Wan, H. Xiong, S. Zhou, X. Chen, X. I. Pereira Hernandez, Y. Wang, S. Lin, A. K. Datye and H. Guo, J. Phys. Chem. C, 2018, 122, 22460–22468.
27. H. A. Tahini, X. Tan and S. C. Smith, ChemCatChem, 2020, 12, 1007–1012.
28. M. Naguib, M. Kurtoglu, V. Presser, J. Lu, J. Niu, M. Heon, L. Hultman, Y. Gogotsi and M. W. Barsoum, Adv. Mater., 2011, 23, 4248–4253.
29. M. Naguib, V. N. Mochalin, M. W. Barsoum and Y. Gogotsi, Adv. Mater., 2014, 26, 992–1005.
30. Y. Wei, P. Zhang, R. A. Soomro, Q. Zhu and B. Xu, Adv. Mater., 2021, 33, 2103148.
31. Z. Guo, J. Zhou and Z. Sun, J. Mater. Chem. A, 2017, 5, 23530–23535.
32. B. Zhang, J. Zhou and Z. Sun, J. Mater. Chem. A, 2022, 10, 15865–15880.
33. B. Zhang, J. Zhou, Z. Guo, Q. Peng and Z. Sun, Appl. Surf. Sci., 2020, 500, 144248.
34. T. Zhang, B. Zhang, Q. Peng, J. Zhou and Z. Sun, J. Mater. Chem. A, 2021, 9, 433–441.
35. B. Li, Y. Wu, N. Li, X. Chen, X. Zeng, Arramel, X. Zhao and J. Jiang, ACS Appl. Mater. Interfaces, 2020, 12, 9261–9267.
36. F. Li and Q. Tang, ACS Appl. Nano Mater., 2019, 2, 7220–7229.
37. E. Wang, B. Zhang, J. Zhou and Z. Sun, Appl. Surf. Sci., 2022, 604, 154522.
38. X. Guo, S. Lin, J. Gu, S. Zhang, Z. Chen and S. Huang, Adv. Funct. Mater., 2021, 31, 2008056.
39. Y. Xiao, C. Shen and T. Long, Chem. Mater., 2021, 33, 4023–4034.
40. X. Yang, C. Shang, S. Zhou and J. Zhao, Nanoscale Horiz., 2020, 5, 1106–1115.
41. X. Liu, Z. Liu and H. Deng, J. Phys. Chem. C, 2021, 125, 19183–19189.
42. Y. Xiao, C. Shen and N. Hadaeghi, J. Phys. Chem. Lett., 2021, 12, 6370–6382.
43. M. Yao, Z. Shi, P. Zhang, W.-J. Ong, J. Jiang, W.-Y. Ching and N. Li, ACS Appl. Nano Mater., 2020, 3, 9870–9879.
44. S. Xing, J. Zhou, B. Zhang and Z. Sun, J. Phys. Chem. C, 2022, 126, 14275–14282.
45. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
46. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
47. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
48. J. P. Perdew and Y. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 45, 13244–13249.
49. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
50. S. Nosé, J. Chem. Phys., 1984, 81, 511–519.
51. S. Qi, Y. Fan, L. Zhao, W. Li and M. Zhao, Appl. Surf. Sci., 2021, 536, 147742.
52. T. Bo, P.-F. Liu, J. Xu, J. Zhang, Y. Chen, O. Eriksson, F. Wang and B.-T. Wang, Phys. Chem. Chem. Phys., 2018, 20, 22168–22178.
53. N. Miao, J. Wang, Y. Gong, J. Wu, H. Niu, S. Wang, K. Li, A. R. Oganov, T. Tada and H. Hosono, Chem. Mater., 2020, 32, 6947–6957.
54. K. Rasoul, K. Mohammad, W. Vei, M. Nanxi, S. Chen, W. Jianfeng and W. Junjie, J. Phys.: Condens. Matter, 2020, 33, 155503.
55. M. Zafari, A. S. Nissimagoudar, M. Umer, G. Lee and K. S. Kim, J. Mater. Chem. A, 2021, 9, 9203–9213.
56. J. Zhou, J. Palisaitis, J. Halim, M. Dahlqvist, Q. Tao, I. Persson, L. Hultman, P. O. Å. Persson and J. Rosen, Science, 2021, 373, 801–805.
57. J. Wang, T.-N. Ye, Y. Gong, J. Wu, N. Miao, T. Tada and H. Hosono, Nat. Commun., 2019, 10, 2284.
58. T. Hu, M. Wang, X. Wang, Y. Zhou and C. Li, Comput. Mater. Sci., 2021, 199, 110810.
59. B. Anasori, M. R. Lukatskaya and Y. Gogotsi, Nat. Rev. Mater., 2017, 2, 16098.
60. W. Sun, S. A. Shah, Y. Chen, Z. Tan, H. Gao, T. Habib, M. Radovic and M. J. Green, J. Mater. Chem. A, 2017, 5, 21663–21668.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2nr05705k

---