Tuning the transition barrier of H₂ dissociation in the hydrogenation of CO₂ to formic acid on Ti-doped Sn₂O₄ clusters

Abstract

A density functional theory study has been performed to investigate cation-doped Sn₂O₄ clusters for selective catalytic reduction of CO₂. We study the influence of Si and Ti dopants on the height of the H₂ dissociation barrier for the doped systems, and then the subsequent mechanism for the conversion of CO₂ into formic acid (FA) via a hydride pinning pathway. The lowest barrier height for H₂ dissociation is observed across the ‘Ti–O’ bond of the Ti-doped Sn₂O₄ cluster, with a negatively charged hydride (Ti–H) formed during the heterolytic H₂ dissociation, bringing selectivity towards the desired FA product. The formation of a formate intermediate is identified as the rate-determining step (RDS) for the whole pathway, but the barrier height is substantially reduced for the Ti-doped system when compared to the same steps on the undoped Sn₂O₄ cluster. The free energy of formate formation in the RDS is calculated to be negative, which reveals that the hydride transfer would occur spontaneously. Overall, our results show that the small-sized Ti-doped Sn₂O₄ clusters exhibit better catalytic activity than undoped clusters in the important process of reducing CO₂ to FA when proceeding via the hydride pinning pathway.

---

Introduction

Combustion of fossil fuels is a major contributing factor to the increasing atmospheric concentration of CO₂, which is regarded as one of the major anthropogenic contributors to global warming.^1,2 To address the current “Climate Emergency”, mitigation of CO₂ either by conversion or trapping is now essential. For this purpose, reduction of CO₂ into commercially valuable chemicals such as formic acid (FA), methanol and longer-chained alcohols is highly advantageous.^3–5 Furthermore, hydrogenation of CO₂ to liquid phase HCOOH and CH₃OH also addresses challenges associated with storage and transport of H₂, with the low volumetric density of H₂ in the gas phase considered prohibitive for its potential use as an energy vector; HCOOH is a non-toxic liquid at room temperature, making it easy to transport and allowing direct implementation in fuel cells.⁶ Currently, however, chemical reduction of CO₂ to formic acid is limited by the availability of suitable catalysts to activate the thermodynamically stable C=O bond, which has a bond energy of 5.52 eV (533 kJ mol⁻¹).⁷ There are several reported methods of CO₂ reduction, including catalytic hydrogenation, electrochemical reduction, photocatalysis and biological reduction; among these, electrochemical and photochemical reductions are promising for facilitating product selectivity, lower production costs and high catalytic efficiency.^8–10

During the last decade, the study of Sn-based electrodes for electrochemical CO₂ reduction has grown rapidly due to the higher overpotentials necessary for competing H₂ evolution reactions, thus bringing selectivity to the reduction of CO₂ to HCOOH.^11–13 Like other non-noble metals, tin undergoes corrosion and degradation, but the oxide layer formed has been identified as highly catalytic towards the electrochemical reduction of CO₂. The importance of the oxide layer was highlighted by Chen and Kanan, who reported that the catalytic activity of SnO₂ depends on the oxide layer percentage, showing that the depth of the oxide layer was proportional to the reduction efficiency.¹⁴ Particle size also has an important effect on catalyst efficacy, with Xu et al. describing the significance of quantum confinement for SnO₂ quantum dots (QDs) within the range of 0.5–2.5 nm, as illustrated for gas sensing properties towards ethanol; QDs show higher sensitivity than larger SnO₂ nanowires.¹⁵ Liu et al. compared the electrocatalytic properties of SnO₂ quantum wires (1.7 nm) with SnO₂ nanoparticles (5.5 nm) for HCOOH formation and concluded that the quantum wires have exposed grain boundaries that enhance the current density, as well as faradaic efficiency, by over 80% for HCOOH formation.¹⁶ This also reveals the importance of small-sized clusters in the catalytic reduction of CO₂ to HCOOH and it influences the overpotential of formic acid production.

In contrast to the catalytic application of pure SnO₂ clusters, there is limited literature for the application of doped SnO₂ clusters towards the hydrogenation of CO₂ to HCOOH. H₂ dissociation on doped metal oxide clusters and surfaces such as Al-doped ZnO, CeO₂ and MgO has been reported, as well as on Ni clusters doped with Rh, Pt, Pd and Au metals.^17,18 However, the mechanism of CO₂ reduction using doped-SnO₂ nanoparticles remains relatively unexplored. Saravanan et al. performed theoretical calculations of SnO₂ surfaces to try and understand the consequences of transition metal doping,¹⁹ considering two types of dopants: Ti, V, Nb, Ta and Zr replacing the six coordinated Sn(IV) in a pristine crystal; and Cd, Co, Pb, Sb and Zn accompanied by a neighbouring O vacancy, i.e. at an Sn(II) site. The extrinsic dopants at the Sn(IV) site resulted in a material showing potential to reduce CO₂ to HCOOH at lower overpotentials than pure SnO₂, with Ti-doped SnO₂ identified as the best catalyst. Such Ti-doped SnO₂ nanoparticles are used currently as sensing materials for ethanol, and both Ti- and Si-doped SnO₂ nanoparticles have been studied to understand the influence of the doping metal on the electrical conductivity.^20–23

Previously, CO₂ activation on the catalyst surface has been reported as occurring by the formation of a carbonate-like structure, followed by the transfer of a hydrogen atom to the carbon atom, producing formate and subsequently formic acid.^14,23,24 Earlier studies show that H₂ can dissociate heterolytically on ZnO clusters, which leads to the formation of a metal-hydride bond.^17,25 Similarly, another study on SnO₂ reported a hydride pinning pathway whereby heterolytic H₂ dissociation leads to the formation of an Sn-hydride bond, and CO₂ activation occurs during the (rate-determining) hydride transfer step; the observed outcome was total selectivity of HCOOH formation at an overpotential of 0.25 eV.²⁶ In the present work, we consider whether dopants in a model dimeric Sn₂O₄ cluster modify the key steps in this favourable hydrogenation pathway of CO₂ to HCOOH. Structures for the Sn₂O₄ clusters have been taken from previous studies²⁶ and, for the newly considered doped system, we chose two dopants: Si, which is isovalent with Sn, with a stable +4 oxidation state as well as vacant 3d orbitals; and Ti, which has an atomic radius similar to Sn, vacant 3d electrons, and stability in the +4 oxidation state (though somewhat reducible). As both dopants are considered formally in +4 oxidation states, their introduction to the cluster does not require compensating oxygen vacancies.

Results and discussion

Structures of Si- and Ti-doped Sn₂O₄ clusters

Si and Ti dopants replace a Sn atom in the small dimeric Sn₂O₄ cluster, and the structures were geometry-optimized using density functional theory (DFT) (Fig. 1). The structure of the SnSiO₄ cluster is similar to that of the pure Sn₂O₄ cluster, maintaining a planar 2-dimensional configuration; in contrast, the SnTiO₄ cluster is 3-dimensional, with the Ti-center taking a distorted T_d shape surrounded by two bridging and one terminal oxygens. The shape of the TiO₃ unit in the SnTiO₄ cluster resembles the Ti-center of a pure Ti₂O₄ cluster.²⁷

Fig. 1 Optimized geometries of SnSiO₄ (left) and SnTiO₄ (right) clusters, along with the HOMOs and LUMOs, as calculated at the MN12-SX/def2tzvpp level of theory. The different colours (aqua/magenta) of the orbitals indicate the phases of the orbitals taken at the cutoff of 0.03.

The formation energy (E_f) of the doped clusters upon replacement of one Sn by the Si or Ti atom is calculated via the equation as follows:

E_f = (E_SnMO4 + E_SnO2) − (E_Sn2O4 + E_MO2),

where M corresponds to the dopant Si or Ti species. The formation energy of the SnSiO₄ and SnTiO₄ clusters with respect to the Sn₂O₄ and MO₂ (SiO₂ and TiO₂) clusters are calculated to be −0.80 eV (−18.55 kcal mol⁻¹) and −0.73 eV (−16.83 kcal mol⁻¹), respectively, indicating that doping of these small clusters is thermodynamically feasible. SnSiO₄ has a lower energy than SnTiO₄, due to the greater strength of the Si–O bond.²⁸

Natural bond orbital (NBO) calculations have been performed to interpret the bonding interactions of the metal dopants with the oxygen atoms in the clusters, and diagrams of the highest occupied molecular orbitals (HOMOs) and the lowest unoccupied molecular orbitals (LUMOs) are included accordingly in Fig. 1. The HOMOs of both of the doped clusters are concentrated on the lone pairs of oxygen, i.e. the non-bonding electrons. The LUMO of SnSiO₄ has the π* character of an O–Si bond, and some character of the Sn-center, which provides us the evidence of having an O–Si π interaction resulting from the back donation of electron density from the oxygen lone pairs to the empty 3d orbitals of silicon. To verify the involvement of Si d orbitals in the bonding interactions, we have analyzed the percentage of individual atomic contributions towards the participating bonds in the SnSiO₄ and SnTiO₄ clusters (Table 1, column 4), along with the orbital contributions of the s, p and d orbitals (Table 1, column 5). Some occupancy of the Si d orbitals is observed in both the terminal and bridging Si–O bonds, along with the s and p orbitals, which reveals that the O–Si π bonding interaction includes the contributions from the Si d orbitals; however, for the SnTiO₄ cluster, the percentage contribution of the d orbitals is much more than those of the s and p orbitals. These observations lead us to conclude that the more covalent-like Si interactions in the SnSiO₄ cluster favourably stabilise the system when compared to those of the Ti-dopant in SnTiO₄. The individual atomic contributions (column 4) of the participating bonds show that O contributes significantly to bonding compared to the Sn, Si and Ti atoms.

Table 1 Percentage contribution of orbitals to the bridging and terminal metal-oxygen bonds as calculated via an NBO calculation at the MN12-SX/def2TZVPP level of theory

Cluster	Bond type	Atom	Elemental contribution (%)	Angular contribution (%)
SnSiO4	Bridging Si–O	Si	12.65	s (27.98) p (48.21) d (23.78)
		O	87.35	s (33.69) p (65.90)
	Terminal Si–O	Si	19.53	s (43.39) p (55.60) d (0.99)
		O	80.47	s (34.36) p (65.0%)
	Bridging Sn–O	Sn	12.82	s (26.54) p (72.66) d (0.45)
		O	87.18	s (12.24) p (87.36)
	Terminal Sn–O	Sn	29.85	s (46.92) p (52.58) d (0.33)
		O	70.15	s (6.94) p (92.93)
SnTiO4	Bridging Ti–O	Ti	12.86	s (23.60) p (23.94) d (51.79)
		O	87.14	s (29.51) p (70.39)
	Terminal Ti–O	Ti	20.57	s (10.69) p (7.40) d (81.33)
		O	79.43	s (19.70) p (80.22)
	Bridging Sn–O	Sn	13.65	s (27.35) p (71.28) d (0.89)
		O	86.35	s (9.23) p (90.65)
	Terminal Sn–O	Sn	28.95	s (45.08) p (54.31) d (0.43)
		O	71.05	s (7.29) p (92.58)

---

H₂ dissociation on the SnMO₄ clusters

Previously, an undoped Sn₂O₄ cluster has been observed to dissociate H₂ at Sn–O bridging sites, with the hydride pinning pathway leading to selective reduction of CO₂ to HCOOH via a formate intermediate.²⁶ Considering this observation, we calculated the transition states and barrier heights of H₂ dissociation on all possible bridging sites in the doped cluster; the subsequent optimized minima and transition state geometries are shown in Fig. 2, and the reaction energy profiles in Fig. 3.

Fig. 2 The optimized structures of doped clusters along with the transition states and H₂ dissociated products. Top: SnSiO₄; bottom: SnTiO₄.

Fig. 3 The potential energy profile diagrams comparing the dissociation of H₂ on the ‘Sn–O’ and ‘M–O’ sites of the SnSiO₄ (left) and SnTiO₄ (right) clusters, calculated at the MN12-SX/def2tzvpp level of theory.

For the SnSiO₄ cluster, the activation energy for H₂ dissociation is 1.76 eV (40.61 kcal mol⁻¹) on a Si–O bond, which is 0.26 eV (5.99 kcal mol⁻¹) higher than at the Sn–O bond [1.50 eV, 34.73 kcal mol⁻¹]. The transition states corresponding to H₂ dissociation on Si–O (A–B) and Sn–O (A–B′) sites (Fig. 2) are characterized by the presence of imaginary frequencies of 1524i cm⁻¹ and 1438i cm⁻¹, respectively. The dissociated products ‘B’ and ‘B′’, formed on the Si–O and Sn–O sites, respectively, are also shown in Fig. 2. Interestingly, despite dissociation occurring at the bridging position of the Si–O site for product B, as confirmed via the transition state (A–B) and IRC calculation (Fig. S1, ESI†), the product has a geometry based on the terminal Si–O bond. The convergence graph is shown in Fig. 4 for the reaction A → A–B → B. Fig. 4 indicates that, whilst the H₂ dissociation transition state is at the bridging site, the product H is unstable at this bridging location and converts to a more stable terminal-positioned structure (Fig. 2, structure B). We attribute the formation of the terminal H to the Si–O dπ–pπ interaction observed for the bridging Si–O bond in the SnSiO₄ cluster, which provides extra stability to the bridging Si–O bonds; the dπ–pπ interaction is less pronounced in the terminal Si–O bond, with the contribution of the Si d orbital reduced (0.99%, compared to 23.78% for the bridging bond). The H₂ dissociation on the Sn–O site results in similar structural arrangement to H₂ dissociation on the undoped Sn₂O₄ cluster investigated previously.²⁶

Fig. 4 Structural convergence from H₂ dissociation at the bridging ‘Si–O’ site to terminal ‘Si–O’ site.

In contrast, the SnTiO₄ cluster shows a smaller barrier on the Ti–O bridge (C–D) of 0.98 eV (22.62 kcal mol⁻¹), with an imaginary frequency of 1373i cm⁻¹. The barrier height is 0.50 eV (11.68 kcal mol⁻¹) lower than that of the Sn–O bridge (C–D′), as shown in Fig. 3. The lower barrier over the Ti–O bridge can be explained via the orbital structures: The LUMO of the SnTiO₄ cluster is located on the Ti-center and thus, when H₂ dissociation takes place over the Ti–O bridge, it occurs via interaction with the LUMO; however, on the Sn–O bridge, H₂ dissociation cannot take place through the Ti-centered LUMO, interacting instead with the LUMO+1 (Fig. S2, ESI†). Therefore, a higher orbital reorganization energy is required for H₂ dissociation over the Sn–O bridge, which leads to a larger activation energy for the process. Notably, H₂ dissociation on the Ti–O site has a lower activation energy than on the Sn–O site in doped and undoped Sn₂O₄ clusters (barrier height = 1.25 eV).²⁶

To understand the strength of the M–H bond, and confirm whether the hydride ion has formed, we analyzed the Bader charge of the H₂ dissociated products (B, B′, D and D′) in Table 2. A hydride is observed in all the models, as the hydrogen of newly formed M–H possesses a negative formal charge. The calculated percentages of orbital contributions of H and M in the M–H bond are shown in Table 3; in the bonding of Si–H, the contribution of the Si 3p orbital is the greatest (46.86%), whereas the Ti–H bond is formed most prominently of 3d orbitals (44.90%). In the latter case, the contribution of d is more than those of valence 4s and 4p orbitals, which contribute 25.81% and 29.12%, respectively, and as a consequence the Ti–H bond is weaker than the Si–H bond.

Table 2 Bader charges and bond distances of reacting atoms on all sites of the doped clusters

Structure	Bond	Bond distances, Å	Bader charges, e
B	Si–H	1.46	Si = +3.15, H = −0.71
	O–H	0.96	O = −1.45, H = +0.70
B′	Sn–H	1.67	Sn = +1.98, H = −0.25
	O–H	0.96	O = −1.53, H = +0.70
D	Ti–H	1.60	Ti = +2.17, H = −0.55
	O–H	0.96	O = −1.37, H = +0.68
D′	Sn–H	1.67	Sn = +1.99, H = −0.27
	O–H	0.96	O = −1.36, H = +0.67

---

Table 3 Percentage of orbital contribution of all metal-hydride bonds obtained via NBO calculation performed at the MN12-SX/def2TZVPP level of theory

Cluster	Atom	Elemental contribution (%)	Angular contribution (%)
SnSiO4	Si	31.60	s (30.85) p (46.86) d (22.21)
	H	68.40	s (99.93) p (0.07)
	Sn	37.57	s (31.12) p (68.44) d (0.33)
	H	62.43	s (99.85) p (0.14)
SnTiO4	Ti	36.88	s (25.81) p (29.12) d (44.90)
	H	63.12	s (99.85) p (0.14)
	Sn	38.10	s (42.59) p (56.78) d (0.56)
	H	61.90	s (99.86) p (0.14)

---

Mechanistic study via hydride pinning pathway

After dissociation of H₂ on the doped Sn₂O₄ cluster, we extended our study to the hydrogenation of CO₂via the hydride pinning pathway. As already discussed, the H₂ dissociation barrier height is lower on SnTiO₄ than that on SnSiO₄, and hence CO₂ reduction is considered only on the SnTiO₄ cluster. The overall pathway follows a two-step catalytic mechanism:²⁶

(a) First, gas phase CO₂ takes up the hydride from the catalytic site following the ‘Eley Rideal’ (ER) mechanism and binds with the Sn-center in the cluster.

(b) Second, the remaining hydrogen transfers to the co-adsorbed oxygen of HCOO by following the ‘Langmuir Hinshelwood’ (LH) mechanism, to form the product HCOOH.

The potential energy pathway for hydrogenation of CO₂ on the SnTiO₄ cluster is presented in Fig. 5. In the mechanism studied, the hydride pinning step is initiated by the product ‘D’ (Fig. 2) where the hydride is transferred from the Ti-site to the carbon of CO₂ to form the formate (HCOO*) intermediate that further binds with the catalyst by one of the intermediate's oxygens. This hydride transfer step proceeds via the transition state TS1 (607i cm⁻¹) to form the intermediate IM2 with an activation energy of 0.49 eV (11.49 kcal mol⁻¹). The transition barrier of the hydride transfer step for the doped system is lower than for the undoped cluster [1.15 eV (26.51 kcal mol⁻¹)],²⁶ which is attributed to two factors: first, the weakness of the Ti–H bond relative to the Sn-H bond, which results in hydride transfer to the carbon being less energetically demanding; second, the Ti-centre in the doped SnTiO₄ cluster retains its tetrahedral shape throughout i.e energy is not required for structure changes of SnTiO₄, unlike for bare Sn₂O₄ where the Sn-centre changes its shape from tetrahedral to planar.²⁶ In the transition state TS1, the Ti–H bond elongates from 1.60 Å to 1.78 Å, while the bond distances of the newly formed C–H and Sn–O bonds are 1.73 Å and 2.33 Å, respectively. After the formation of the HCOO* intermediate, the orientation spontaneously changes to IM2′ by flipping the –CHO group of HCOO* around the C–O(–Sn) bond, with a ΔG of −0.30 eV (−6.91 kcal mol⁻¹), which then allows uptake of the second hydrogen from the cluster. In the optimized IM2′, we observe a new Sn⋯O interaction of 2.36 Å (along with the remaining Sn–O bond); the O is now nearer to the remaining H on the cluster, and this hydrogen then transfers from the cluster to form HCOOH. During the formation of HCOOH, IM2′ overcomes an activation barrier of 0.35 eV (8.17 kcal mol⁻¹) via the transition state TS2 (920i cm⁻¹), which is smaller than that of the pinning step (TS1) and so indicating that the reaction will proceed rapidly to the equilibrium once the HCOO* intermediate is formed. The HCOOH product leaves the catalytic cycle in the same step and hence the catalyst is recycled for the next H₂ dissociation step.

Fig. 5 Potential energy diagram representing the hydrogenation of CO₂ to HCOOH on the SnTiO₄ cluster via the hydride pinning pathway.

For the overall reaction, the hydride pinning step (TS1) is the rate-determining step (RDS) and hence the free energy difference between IM2′ and IM1 equates to the necessary overpotential for realising the hydrogenation process. This overpotential factor can be calculated from the computational hydrogen electrode (CHE) model given by Nørskov et al.,³⁷ which is discussed in the computational methodology. In the outlined reaction mechanism, the difference in free energies is −0.31 eV, which reveals that the formation of HCOO* intermediate is calculated to be thermodynamically spontaneous. This spontaneity was not observed in the case of undoped Sn₂O₄, where the required overpotential of HCOO* formation is 0.25 eV.²⁶

Comparing the present results with previous studies on the hydrogenation of CO₂ to HCOOH on dimeric SnO₂,²⁶ we can propose the following:

(a) The H₂ dissociation barrier height on the Ti-doped Sn₂O₄ is 0.98 eV, which is 0.54 eV lower than that on the undoped Sn₂O₄ (1.52 eV). The doped cluster is therefore a potentially more active catalyst due to the reduced kinetic H₂ dissociation barrier.

(b) In the hydride transfer step, the barrier height for formate formation is 0.49 eV on the Ti-doped cluster, which is 0.66 eV greater than that on the undoped Sn₂O₄ (1.15 eV). The doped cluster therefore has potentially improved catalytic selectivity towards reduction of CO₂ to HCOOH via the hydride pinning pathway.

(c) Lastly, the overpotential factor drops from 0.25 eV when considering the undoped Sn₂O₄ cluster to −0.31 eV for SnTiO₄, correlating with the previous work reported by Saravanan et al.¹⁹ and illustrating that the reaction would proceed favourably.

Summary and conclusions

Titanium and silicon dopants have been considered on the Sn⁴⁺ centers of a pure Sn₂O₄ cluster, and subsequently tested in the context of CO₂ hydrogenation. Negative formation energies for SnSiO₄ and SnTiO₄ reveal the replacement of Sn by Si or Ti to be energetically favourable. H₂ dissociation barrier heights are calculated as favourable on all the bridging sites available on the clusters, with the lowest activation energy on the bridging Ti–O site of the SnTiO₄ cluster. Our observations of the potential role for Si and Ti dopants indicates that metals providing π character in the M–O bond will be less suitable catalysts for CO₂ reduction, while first and second row early transition metals (e.g. Ti, V, Nb and Zr), are good candidates as dopants in SnO₂ clusters because π bonding is less common. Bader charge analysis of the metal–hydride (Ti–H) shows that one H on each cluster possesses a formal negative charge. The hydride is the key in the hydride-assisted pathways for CO₂ reduction, triggering the total selectivity for the formation of HCOOH over CO at a lower overpotential. The SnTiO₄ cluster has a reduced barrier relative to the pure Sn₂O₄ cluster in the rate-determining step (HCOO* formation), and also requires a lower overpotential for the formation of HCOOH, delivering product selectivity. Experimentally, small-sized bare SnO₂ quantum dots are used for the conversion of CO₂ to HCOOH at low overpotentials but the use of Ti-doped SnO₂ particles is not yet reported; therefore, our results should be considered as a stimulus for further experimental investigation of HCOOH formation via the hydride pinning pathway.

Methodology

Kohn–Sham density functional theory (DFT), as implemented in the Gaussian 09 software package,²⁹ has been used to obtain the lowest energy structures and thermochemical properties involved in the mechanism of CO₂ reduction. For geometry relaxation and frequency calculations, a range-separated hybrid non-separable meta-NGA, MN12-SX,³⁰ is used; the accuracy of the MN12-SX density functional when predicting structures and energies has been extensively studied previously.^26,31,32 A combination of density fitting triple-ξ def2TZVPP basis sets is used,³³ as these minimise the basis set superposition error (BSSE) and give results close to the DFT basis set limit;³⁴ the BSSE is <0.04 eV with the triple-ξ basis set, and energies reported in the manuscript include this correction.

Frequency calculations are used to confirm the stability of the reactants (Rs), intermediates (IMs) and products (Ps) along the reaction pathway, and to verify the transition states (TSs) identified. The presence of one imaginary frequency, consistent with the eigenvector along with the reaction coordinate, confirms that the TS structure is a first-order saddle point, whereas only real vibrational frequencies are observed for the R, IM and P minima. Changes in the free energy of each reaction step (ΔG) and free energy of activation (ΔG^‡) are calculated from the free energy differences between the final and initial states, assuming room temperature and atmospheric pressure (298 K, 1 atm). Formic acid is formed in the liquid phase at room temperature and therefore, as discussed elsewhere previously, a correction of −0.12 eV (taken from the NIST database) is included in our calculated energy of the FA product to account for the free energy difference between its liquid- and gas-phase forms.^26,35 Bader charge analysis is performed on charge density distribution using the AIMALL software package.³⁶

To interpret the overpotential factor associated with the formic acid formation, we have used the computational hydrogen electrode (CHE) model, which correlates the reaction free energy change with the electrochemical potential. We calculate the free energy change at the rate-determining step (RDS), determined as the hydride transfer step to form the HCOO* intermediate, where the CHE gas-phase hydrogen is at equilibrium with a proton and electron pair (H⁺ + e⁻ → 1/2H₂) at a potential of 0 V against the reversible hydrogen electrode (RHE) for all pH values, all temperatures and 1 atm of pressure.³⁷

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors are thankful to the Science & Engineering Research Board (SERB) (EMR/2016/003195), the Department of Science and Technology (DST) nanomission ([SR/NM/NS-1147/2016(G)]) and the Department of Biotechnology (DBT) project (BT/PR16182/NER/95/92/2015) New Delhi, India, and Tezpur University for the financial support. AJL acknowledges funding by the UKRI Future Leaders Fellowship program (MR/T018372/1). The authors gratefully acknowledge Dr Sanjeev P. Mahanta and Dr Dharitri Das of Tezpur University for their valuable suggestions while carrying out the work.

References

1. J. C. Fyfe, N. P. Gillet and F. W. Zwiers, Nat. Clim. Change, 2013, 3, 767–769.
2. A. Stips, D. Macias, C. Coughlan, E. Garcia-Gorriz and X. S. Liang, Sci. Rep., 2016, 6, 21691–21699.
3. W. Li, H. Wang, X. Jiang, J. Zhu, Z. Liu, X. Guo and C. Song, RSC Adv., 2018, 8, 7651–7669.
4. Q. Li, J. Fu, W. Zhu, Z. Chen, B. Shen, L. Wu, Z. Xi, T. Wang, G. Lu, J. Zhu and S. Sun, J. Am. Chem. Soc., 2017, 139, 4290–4293.
5. A. Álvarez, A. Bansode, A. Urakawa, A. V. Bavykina, T. A. Wezendonk, M. Makkee, J. Gascon and F. Kapteijn, Chem. Rev., 2017, 117, 9804–9838.
6. A. K. Singh, S. Singh and A. Kumar, Catal. Sci., 2016, 6, 12–40.
7. K. Ahmad and S. Upadhyayula, Environ. Prog. Sustainable Energy, 2019, 38, 98–111.
8. Z. Sun, T. Ma, H. Tao, Q. Fan and B. Han, Chem, 2017, 3, 560–587.
9. F. Li, L. Chen, G. P. Knowles, D. R. MacFarlane and J. Zhang, Angew. Chem., Int. Ed., 2017, 56, 505–509.
10. J. Wu, Y. Huang, W. Ye and Y. Li, Adv. Sci., 2017, 4, 1–29.
11. J. Wu, F. G. Risalvato, S. Ma and X. D. Zhou, J. Mater. Chem. A, 2014, 2, 1647–1651.
12. W. Deng, L. Zhang, L. Li, S. Chen, C. Hu, Z. J. Zhao, T. Wang and J. Gong, J. Am. Chem. Soc., 2019, 141, 2911–2915.
13. S. Zhang, P. Kang and T. J. Meyer, J. Am. Chem. Soc., 2014, 136, 1734–1737.
14. Y. Chen and M. W. Kanan, J. Am. Chem. Soc., 2012, 134, 1986–1989.
15. X. Xu, J. Zhuang and X. Wang, J. Am. Chem. Soc., 2008, 130, 12527–12535.
16. S. Liu, J. Xiao, X. F. Lu, J. Wang, X. Wang and X. W. Lou, Angew. Chem., Int. Ed., 2019, 58, 8499–8503.
17. T. W. Keal, P. Sherwood, G. Dutta, A. A. Sokol and C. R. A. Catlow, Proc. R. Soc. London, Ser. A, 2011, 467, 1900–1924.
18. N. S. Venkataramanan, A. Suvitha, H. Mizuseki and Y. Kawazoe, Int. J. Quantum Chem., 2013, 113, 1940–1948.
19. K. Saravanan, Y. Basdogan, J. Dean and J. A. Keith, J. Mater. Chem. A, 2017, 5, 11756–11763.
20. A. P. S. Isabel, C. H. Kao, R. K. Mahanty, Y. C. S. Wu, C. Y. Li, C. Y. Lin and C. F. Lin, Ceram. Int., 2017, 43, 10386–10391.
21. J. Q. Wang, H. L. Kang and Y. Zhang, Mater. Res. Express, 2018, 5, 075902–075913.
22. J. Jang, H. Yim and J. W. Choi, Thin Solid Films, 2018, 660, 606–612.
23. Q. Tang, Y. Lee, D. Y. Li, W. Choi, C. W. Liu, D. Lee and D. E. Jiang, J. Am. Chem. Soc., 2017, 139, 9728–9736.
24. C. W. Lee, N. H. Cho, K. D. Yang and K. T. Nam, ChemElectroChem, 2017, 4, 2130–2136.
25. S. A. French, A. A. Sokol, S. T. Bromley, C. R. A. Catlow, S. C. Rogers, F. King and P. Sherwood, Angew. Chem., Int. Ed., 2001, 40, 4437–4440.
26. P. J. Sarma, S. D. Baruah, A. J. Logsdail and R. C. Deka, ChemPhysChem, 2019, 20, 680–686.
27. O. Lamiel-Garcia, A. Cuko, M. Calatayud, F. Illas and S. T. Bromley, Nanoscale, 2017, 9, 1049–1058.
28. F. Weinhold and R. West, Organometallics, 2011, 30, 5815–5824.
29. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci and G. A. Petersson, et al., Gaussian09 Revision D.01, Gaussian Inc., Wallingford CT. 2010.
30. R. Peverati and D. G. Truhlar, Phys. Chem. Chem. Phys., 2012, 14, 16187–16191.
31. F. Weigend, Phys. Chem. Chem. Phys., 2006, 8, 1057–1065.
32. F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005, 7, 3297–3305.
33. J. Frau and D. Glossman-Mitnik, Theor. Chem. Acc., 2018, 137, 1–10.
34. S. Paranthaman, J. Moon, J. Kim, D. E. Kim and T. K. Kim, J. Phys. Chem. A, 2016, 120, 2128–2134.
35. H. Afeefy, J. Liebman and S. Stein, Neutral thermochemical data. in ed. P. Linstrom and W. Mallard, NIST Chemistry WebBook, NIST Standard Reference Database Number. 69 (National Institute of Standards and Technology), Gaithersburg MD, USA, 2010.
36. AIMAll (Version 17.11.14), Todd A. Keith, TK Gristmill Software, Overland Park KS, USA, 2017 (aim.tk gristmill.com).
37. J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard and H. Jónsson, J. Phys. Chem. B, 2004, 108, 17886–17892.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0cp04472e

---