Mechanistic investigation of benzene esterification by K₂CO₃/TiO₂: the catalytic role of the multifunctional interface

Abstract

Potassium carbonate dispersed over a defective TiO₂ support (K₂CO₃/TiO₂) is an efficient catalyst for benzene esterification with CO₂ and CH₃OH. Density functional theory calculations reveal that this unique catalytic reactivity originates from the cooperation of the Ti³⁺/K⁺ surface sites. The K₂CO₃ promotor steers the stabilization of surface intermediates thus preventing catalyst deactivation.

---

CO₂ conversion into valuable chemicals has received much attention due to the environmental concerns associated with the growing atmospheric concentrations of greenhouse gases. Many efforts have been invested to develop the economic carbon-neutral system by recycling the carbon resource of CO₂ from industrial emission to the production of chemicals.^1–3 The carboxylation of aromatics by CO₂ is one of the attractive routes for the valorization of CO₂, because the produced aromatic carboxylic acids and their derivatives can serve as important chemical feedstocks.^4,5 Conventionally, such a carboxylation coupling reaction is carried out in the presence of a strong base or Lewis acid such as NaH,⁶ AlCl₃,⁴ and AlBr₃.⁷ The strong base is needed to cleave the C–H bond of arene to form a nucleophilic carbon atom which can further interact with the weak CO₂ electrophile. The role of the Lewis acid is to activate the CO₂ molecule before the arene C–H bond carboxylation.⁸ However, both these strategies usually provide rather low yields of the target products due to the low electrophilicity of CO₂ and side reactions caused by excessive reactivity of the mediators.^7,8 Therefore, the development of alternative catalytic procedures avoiding the usage of strong base or acid is highly desired but also represents one of the great challenges for this reaction.

Recently, Kanan et al. reported that alkali carbonates (K₂CO₃ and Cs₂CO₃) finely dispersed over a TiO₂ support can promote the two-step cycle of benzene esterification with CO₂ and CH₃OH to produce methyl benzoate with both high yield (80%) and high selectivity (100%) in the absence of stoichiometric additives.⁹ It is important to note that bare TiO₂ can also promote the first step of benzene C–H bond carboxylation, however, the catalyst became deactivated after just one catalyst recycling. In contrast, no carboxylation products were observed when K₂CO₃ or Cs₂CO₃ powders as the only catalyst component were used. Thus, it was hypothesized that dispersing alkali carbonate over TiO₂ would engender catalytic carboxylation activity towards hydrocarbon substrates because of the disruption of the bulk alkali carbonate structure. However, the mechanistic aspects of this system, such as the nature of the active site, initiation of the reaction and the exact role of the different catalyst components, remained moot.⁸ This inspired us to carry out a comprehensive mechanistic study of benzene carboxylation with CO₂ and subsequent reaction with CH₃OH over the K₂CO₃/TiO₂ catalyst by periodic density functional theory (DFT) calculations. Our main objective was to identify the role and function of each catalyst component and to propose the origin of the deactivation of bare TiO₂ catalyst.

In this work, all calculations were performed using the Vienna Ab Initio Simulation Package (VASP) 5.3.5.^10,11 DFT calculations were carried out using PBE functional based on the generalized gradient approximation (GGA).¹² Grimme's DFT-D3 method with Becke-Jonson damping was used to account for the dispersion interactions.¹³ The DFT+U method was applied to the 3d orbitals of Ti to correct the on-site Coulomb interactions. The U value used in this work is 4.2 eV.¹⁴ The energy cutoff and convergence criteria for the electronic and ionic loops were 400 eV, 10⁻⁵ eV, and 0.05 eV Å⁻¹, respectively. Transition states were determined by the nudged-elastic band method with the improved tangent estimate (CI-NEB) and subsequent frequency analysis. The model of K₂CO₃/TiO₂ catalyst was built following the experimental evidence of the very fine dispersion of K₂CO₃ on the surface of TiO₂.⁹ Fig. S1 in the ESI† shows the catalyst model featuring the K₂CO₃ species deposited on the defective anatase TiO₂(101) surface. We hypothesized that the interface of coordination-unsaturated surface Ti site (Ti³⁺) together with the adjacent K₂CO₃ cluster form the reactive ensemble because neither the bulk crystalline K₂CO₃ nor the pristine TiO₂ surface are active.^9,15

Scheme 1 presents a proposed reaction mechanism for the esterification of benzene with CO₂ and CH₃OH by the K₂CO₃/TiO₂ catalyst. The computed reaction energy profile is shown in Fig. 1. Firstly, the two possible mechanisms of C₆H₆ C–H bond deprotonation were evaluated and it was found that the direct C–H carboxylation of benzene with activated CO₂ is preferred over the C₆H₆ deprotonation (Fig. S2 and discussion, ESI†). Therefore, we proposed that the reaction starts with the adsorption of CO₂ through bidentate coordination with the interface Ti³⁺ and K⁺ sites. The presence of K₂CO₃ prevents a bidentate adsorption mode of CO₂ with two Ti³⁺ surface atoms which occur on the bare defective TiO₂ surface. In the next step, the C–C bond formation between the co-adsorbed benzene and CO₂ occurs to form the C₆H₆COO* intermediate (C₆H₆_CO₂* → C₆H₆COO*). This step is endothermic with ΔE = 0.46 eV and it proceeds with an activation energy (E^‡) of 0.72 eV. Next, the C–H bond of the activated C₆H₆ fragment is cleaved to form benzoic acid (C₆H₅COOH*) or benzoate (C₆H₅COO*) surface intermediate. The former is formed via a direct intramolecular H-transfer from the C₆H₆ moiety of C₆H₆COO* to terminal O of the carboxylate moiety. The computed activation barrier for this step is 1.31 eV. An alternative path involves a two-step surface-assisted H*-transfer, upon which the C₆H₆COO* intermediate is first deprotonated by the vicinal basic surface O sites (E^‡ = 1.19 eV) followed by the C₆H₅COO* and H* recombination (E^‡ = 1.19 eV). The highest activation energy of the indirect route is only ca. 0.10 eV lower than that of the direct pathway, indicating that both reaction routes can contribute to the catalytic reaction. The hydrogen transfer from the C₆H₆COO* to form C₆H₅COO* is predicted to be more difficult than the initial coupling of CO₂ and benzene, which is consistent with the experimentally observed kinetic isotopic effect results.⁹ Electronic analysis further indicated that effective charge transfer between intermediates and the defective catalyst surface facilitates the C–C coupling and deprotonation reaction processes (Fig. S3 and S4, ESI†). The desorption of benzoic acid to regenerate the catalytic interface is endothermic by 0.95 eV.

Scheme 1 Proposed mechanism of benzene esterification with CO₂ and CH₃OH on K₂CO₃/TiO₂ catalyst. The A + CO₂ + C₆H₆ → D conversion involves the carboxylation of benzene with CO₂via the one-step direct (dashed) or stepwise indirect (solid path) mechanisms to yield the adsorbed benzoic acid. The latter is methylated with CH₃OH (D + CH₃OH → F + C₆H₅COOCH₃) to yield methyl benzoate product and the adsorbed water. Subsequent desorption of H₂O by-product (F → A + H₂O) regenerates the catalytic surface ensemble.

Fig. 1 DFT-computed reaction energy diagram for the benzene esterification with CO₂ and CH₃OH on K₂CO₃/TiO₂.

The closure of the catalytic cycle can be facilitated in the presence of methanol, which reacts with the surface benzoate intermediate (CH₃OH_C₆H₅COOH*) to produce methyl benzoate product (C₆H₅COOCH₃*_HO*_H*). During the methylation process, a CH₃OH molecule is added to the system and co-adsorbed at the neighboring surface oxygen atom of benzoic acid. Then, the methyl benzoate is generated by the formation of the C–O bond between CH₃OH and benzoic acid. The simultaneous deprotonation of CH₃OH* and cleavage of the C–OH bond of C₆H₅COOH* result in the generation of two hydroxyl groups on the surface. This concerted step is slightly exothermic and proceeds with an activation energy of 0.73 eV. In the next step, the methyl benzoate product is desorbed from the surface with ΔE of 0.30 eV. The last step is the recombination of surface OH groups to form H₂O and regenerate the catalytic interface sites (HO*_H*^O2 → H₂O*). This dehydration step is barrierless but proceeds with a barrier of 1.00 eV associated with the surface migration of H*. The hydrogen transfer step following the C–C bond formation is identified as the most difficult step of the carboxylation reaction having the highest activation barrier of 1.31 eV along the reaction path.

Previous experimental studies⁹ showed that the carbonate-free defective TiO₂ can also promote benzene carboxylation but it loses the activity already after the first reaction cycle. We therefore hypothesized that the reaction intermediates or the reaction products (e.g. methyl benzoate or water) might block the active site of the bare TiO₂ catalyst, while the presence of K₂CO₃ species protects catalyst from such poisoning. To check this hypothesis, the DFT analysis was extended to the mechanism of benzene carboxylation followed by methylation on the bare and defective anatase TiO₂ (101) surface. Fig. 2 presents the respective DFT-computed reaction energy diagram. In this case, the C–C bond formation between CO₂ and benzene is thermodynamically and kinetically more favorable than on the interface site (ΔE = −0.63 eV, E^‡ = 0.12 eV). However, the subsequent H* transfer to directly form adsorbed benzoic acid is in this case 0.3 eV higher than over the K₂CO₃/TiO₂. A much more favorable reaction is the benzoate formation via the hydroxylation of the TiO₂ surface. This step has a barrier of 1.1 eV and stabilizes the system by ΔE of −1.3 eV. Almost identical energetics was observed for the K₂CO₃/TiO₂ catalyst. For the subsequent concerted benzoic acid methylation reaction, the activation energy over defective TiO₂ is 0.36 eV higher than that of K₂CO₃/TiO₂ catalyst reaction (TS_B5 = 1.09 eV vs. TS_K5 = 0.73 eV).

Fig. 2 Reaction energy diagram for the benzene esterification with CO₂ and CH₃OH on defective TiO₂ surface.

The comparison of the reaction profiles in Fig. 1 and 2 reveals that the activation energies of all elementary steps (except the initial CO₂ coupling with benzene) over the defective TiO₂ surface and K₂CO₃-promoted TiO₂ are quite close and comparable suggesting that indeed both catalysts can enable the esterification reaction of benzene with CO₂ and CH₃OH to form methyl benzoate. However, we find that most of the reaction intermediates on the defective TiO₂ surface are significantly more stable than those on K₂CO₃/TiO₂. The energies of all reaction intermediates on the defective TiO₂ surface are in the range of 0.00 to −3.00 eV, while those on K₂CO₃/TiO₂ catalyst fall in the range of +1.00 to −2.00 eV relative to the reactant state. The stronger binding interaction on the bare TiO₂ surface impedes the desorption of the products and regeneration of the active site for the next catalytic cycle (Fig. S5, ESI†). Specifically, the desorption energies of methyl benzoate and H₂O on the defective TiO₂ surface are 0.39 and 0.32 eV higher than that on K₂CO₃/TiO₂, respectively (ΔE_C6H5COOCH3: 1.37 eV vs. 0.98 eV; ΔE_H2O: 0.62 eV vs. 0.30 eV).

Based on these results, we conclude that both bare TiO₂ with oxygen vacancy and K₂CO₃/TiO₂ catalysts are able to activate CO₂ and benzene to form benzoate products. However, the excessive reactivity of the surface sites in the former results in surface poisoning by the reaction products/byproducts. The presence of potassium carbonate species partially deactivates the reactive sites on the titanium surface to facilitate the product desorption and the regeneration of the catalytic interface sites. Although the K₂CO₃ species cannot act as an active site alone for this reaction, it steers the local structures of the transition states and reaction intermediates, and thus facilitates the products desorption and catalyst regeneration. These insights shed light onto the role of multicomponent reaction environments on the catalytic surface for the efficient CO₂ valorization and they can form a base for further development of efficient and stable catalysts for the direct carboxylation with CO₂ of other more challenging hydrocarbon substrates such as ethane, methane and ethylene.

Authors acknowledge financial support from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement No. 725686) J. M. gratefully acknowledges the Royal Thai Government Scholarships for the financial support. The use of supercomputer facilities was sponsored by NWO Domain Science.

Conflicts of interest

The authors declare no conflicts of interest.

Notes and references

1. M. D. Burkart, N. Hazari, C. L. Tway and E. L. Zeitler, ACS Catal., 2019, 9, 7937–7956.
2. A. D. N. Kamkeng, M. Wang, J. Hu, W. Du and F. Qian, Chem. Eng. J., 2021, 409, 128138.
3. C. Hepburn, E. Adlen, J. Beddington, E. A. Carter, S. Fuss, N. Mac Dowell, J. C. Minx, P. Smith and C. K. Williams, Nature, 2019, 575, 87–97.
4. G. A. Olah, B. Török, J. P. Joschek, I. Bucsi, P. M. Esteves, G. Rasul and G. K. Surya Prakash, J. Am. Chem. Soc., 2002, 124, 11379–11391.
5. A. S. Lindsey and H. Jeskey, Chem. Rev., 1957, 57, 583–620.
6. J. Luo, S. Preciado, P. Xie and I. Larrosa, Chem. – Eur. J., 2016, 22, 6798–6802.
7. K. Nemoto, H. Yoshida, N. Egusa, N. Morohashi and T. Hattori, J. Org. Chem., 2010, 75, 7855–7862.
8. J. Luo and I. Larrosa, ChemSusChem, 2017, 10, 3317–3332.
9. D. J. Xiao, E. D. Chant, A. D. Frankhouser, Y. Chen, A. Yau, N. M. Washton and M. W. Kanan, Nat. Chem., 2019, 11, 940–947.
10. G. F. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
11. G. F. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
12. J. P. B. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 3865–3868.
13. G. S. Stefan, E. Stephan and G. Lars, J. Comput. Chem., 2011, 32, 1456–1465.
14. W. Song, S. Ma, L. Wang, J. Liu and Z. Zhao, ChemCatChem, 2017, 9, 4340–4344.
15. D. C. Sorescu, W. A. Al-Saidi and K. D. Jordan, J. Chem. Phys., 2011, 135, 124701.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1cc02513a

---