Structure-sensitive epoxidation of dicyclopentadiene over TiO₂ catalysts

Abstract

Epoxidation of dicyclopentadiene (DCPD) is studied on a series of TiO₂ catalysts using hydrogen peroxide as an oxidant. DCPD derivatives have applications in several areas including polymer, pharmaceutical and pesticide products. The control of selectivity leading to the desired product is important for many of these applications. Using experimental and computational studies, we show that the surface crystalline phases of TiO₂ play crucial roles not only in the formation of peroxo species but also in the selective epoxidation of two different C=C double bonds in DCPD.

---

Given the versatility of the two different double bonds in its chemical structure, dicyclopentadiene (DCPD) is one of the most interesting cyclic olefin compounds. Numerous DCPD derivatives can be found in the pharmaceutical, pesticide, and polymer industries.¹ Amongst the DCPD derivatives, DCPD epoxides find their main uses in adhesives and insulation materials.²

Some heterogeneous catalysts have been studied for DCPD epoxidation, such as H₃PW₁₂O₄₀ on SBA-15 and on chloromethylated polystyrene resin,^3,4 and the dispersion of phosphotungstic acid is responsible for the catalytic performance.⁴ Metal complexes intercalated in Zn/Al layered double hydroxide structures (Sulfonate-salen-M^III, M = Mn or Fe) showed higher activity than Fe.⁵ Besides the catalytic activity (conversion rate of DCPD), product selectivity is another crucial factor in the epoxidation of DCPD. The epoxidation of DCPD yields two different mono-epoxides (endo-4-oxatetracyclo-[6.2.1.0.^2,60^3,5]undec-9-ene (P1) and endo-9-oxatetracyclo-[5.3.1.0.^2,60^8,10]undec-3-ene (P2)), depending on the location of the epoxide group (in the cyclopentene ring or in the norbornene ring, respectively) (Fig. 1(a)). Due to the difficulty in the product separation unit, the development of a selective epoxidation of DCPD has been encouraged,⁶ but the product selectivity typically did not rely on the chemical properties of the metal center.^3–5

Fig. 1 (a) Reaction scheme of DCPD epoxidation using H₂O₂ as an oxidant and reaction products P1 and P2. Catalytic performances of different TiO₂ catalysts in the epoxidation of DCPD: (b) TiO₂-anatase and TiO₂-rutile; (c) TiO₂-blue and TiO₂-black.

TiO₂ has been widely used as a catalyst for various reactions such as photocatalysis,⁷ CO oxidation,⁸ and H₂O₂ decomposition.⁹ In particular, the crystallinity of the TiO₂ catalysts (anatase and rutile) is responsible for their geometric and electronic properties¹⁰ and plays key roles in view of the product selectivities in the reactions of the decomposition of hydrogen sulphide¹¹ and the photo-oxidation of water.¹²

Our preliminary results demonstrated that the molecular oxygen (as well as the dissolved oxygen in the liquid phase) could not activate the double bonds of DCPD in methanol at 333 K. For example, insignificant conversion of DCPD (<1%) was obtained at 9 bar of pure oxygen gas in an autoclave. Thus, we studied selective DCPD epoxidation on a series of TiO₂ catalysts using hydrogen peroxide (H₂O₂) as an oxidant. Indeed, the epoxidation of cyclic olefin can proceed with H₂O₂ in two sequential steps: (i) the formation of peroxo-species on the metal sites and (ii) the oxygen transfer from the surface to an olefin to form an epoxide.¹³

Fig. 1(b) displays the DCPD epoxidation results of two different crystalline phases of TiO₂. TiO₂-anatase effectively converted DCPD into the related mono-epoxides, showing ca. 2 times higher activity than TiO₂-rutile (C_DCPD = 13% and 7% in 6 h, respectively). The selectivity of the epoxidation products (towards P1 and P2) is greatly influenced by the crystalline phase of the TiO₂ catalysts. For example, on TiO₂-anatase, both P1 and P2 were produced with higher selectivity toward P1 (Table 1). This indicates that the double bond in the cyclopentene ring is preferably reacted on the TiO₂ anatase phase, which is in line with the previous results over the Ti-incorporated SBA-15 catalyst.^14,15 The DCPD di-epoxide was not observed in this study, suggesting that the epoxidation sites were utilized for the reactant (DCPD) and not occupied by the mono-epoxides, similar to the results of Bhattacharjee et al.⁵ Interestingly, despite its lower catalytic performance, TiO₂-rutile solely produced P2, i.e., the double bond in the norbornene moiety is selectively reacted to form mono-epoxide.

Table 1 BET surface area and average pore size for different TiO₂ catalysts, and the catalytic performance results for the epoxidation of DCPD

Catalyst	BET surface area (m^2 g^−1)	Average pore size (nm)	Normalized activity^a (molDCPD m^−2 h^−1)	S P1/P2
a For TiO2-anatase and TiO2-rutile, the normalized activity was calculated in 6 h of reaction. For TiO2-blue and TiO2-black, the activity values were calculated in 8 h, which is ca. 6 h from the point of observed conversion.
TiO2-anatase	60.8	21.8	28	1.7
TiO2-rutile	2.4	16.8	138	0
TiO2-blue	58.6	29.2	37	1.1
TiO2-black	2.6	16.3	350	0

---

We further prepared two additional TiO₂ catalysts (TiO₂-blue and TiO₂-black),^16–18 and the colors of the catalysts are attributed to the oxygen vacancies, based on the crystalline phases (anatase and rutile) (Fig. S1, ESI†).¹⁹ At the initial reaction stage (until ca. 2 h), no conversion of DCPD was observed over TiO₂-blue and TiO₂-black. This suggests that (i) DCPD is not reactive with H₂O₂ in solution and (ii) a certain delay (or lag-phase) is necessary to initiate DCPD epoxidation over the catalysts with oxygen vacancies. Since the textural properties of TiO₂-blue and TiO₂-black are similar to those of TiO₂-anatase and TiO₂-rutile, respectively (Table 1 and Fig. S1, S2, ESI,† as also reported by previous works^16–18,20), we expected that the observed lag-phase is related to the surface oxygen vacancies, which might need to be modified by the oxygen from H₂O₂. In terms of the surface area normalized activity (Table 1), TiO₂-blue and TiO₂-black showed superior catalytic performance compared to TiO₂-anatase and TiO₂-rutile. We expect that the few nanometer layers of the disordered TiO₂ surface²¹ can be attributed to the enhanced epoxidation performance. Similarly, the formation of reactive oxygen species is preferred on the amorphous ZrO₂ than the crystalline monoclinic-ZrO₂.²² Amorphous Nb₂O₅ and Ta₂O₅ also showed higher performance in the catalytic oxidation of glycerol and cyclohexene than their crystalline forms.^23,24 For TiO₂-blue, both P1 and P2 were produced with higher selectivity to P1 than P2. We suggest that the newly formed oxygen functionalities are highly reactive to convert the double bond in the norbornene ring as well as the one in the cyclopentene ring, similar to the fact that the oxygen vacancies are known to be responsible for the altered activity in oxidation reactions.²⁵ Meanwhile, only P2 was observed on TiO₂-black, indicating that the selectivity towards P1 or P2 is largely dependent on the surface crystalline phase.

After the epoxidation of DCPD, the color of the spent TiO₂-anatase catalyst was changed from white to yellow, indicating the formation of the additional surface oxygen functional groups on titania.^26,27 To identify the oxygen functionalities responsible for the catalytic activity, we characterized TiO₂ catalysts with Raman spectroscopy (Fig. 2). For TiO₂-anatase and TiO₂-blue, the Raman features at 148, 395, 515 and 630 cm⁻¹ are attributed to the anatase phase (the modes of E_g, B_1g, A_1g or B_1g, and E_g, respectively).²⁸ Meanwhile, the five Raman bands are observed for TiO₂-rutile and TiO₂-black at 140, 235, 445, 610 and 825 cm⁻¹, due to the modes of B_1g, multi-phonon process, E_g, A_1g and B_2g, respectively. After the treatment of TiO₂ with H₂O₂, the formation of oxo species on TiO₂ (yellow coloration of TiO₂) by H₂O₂ was observed with the possibility of three different forms on the Ti metal centers (oxo, peroxo, and superoxo species).²⁹ The O–O stretching frequencies are typically observed between 800 and 930 cm⁻¹, depending on the coordination with the environment.³⁰ For example, the Raman band of H₂O₂ is typically positioned at 880 cm⁻¹.³¹ On TiO₂-anatase, the Raman bands of peroxo species were observed at 871 cm⁻¹, indicating that the peroxo species are coordinated to the Ti sites (Fig. 3(a)).³² On the contrary, for TiO₂-blue, the Raman band of peroxo species was observed at a higher wavenumber (910 cm⁻¹) (Fig. 2(b)), related to the perturbation of the chemical structure of the adsorbed molecule on the catalyst surface.³³ For the rutile phase TiO₂ catalysts (e.g., TiO₂-rutile and TiO₂-black), however, no additional Raman bands are observed after treatment with H₂O₂ (Fig. 2(b) and (d)), possibly due to (i) the low surface areas of the rutile phase TiO₂ catalysts and (ii) the short lifetime of peroxo intermediates on the rutile phase catalysts.³⁴

Fig. 2 Raman spectra of TiO₂ catalysts. (a) TiO₂-anatase, (b) TiO₂-rutile, (c) TiO₂-blue and (d) TiO₂-black. In each figure, the Raman spectra of H₂O₂ treated samples are displayed on top of the spectra of the untreated samples.

Fig. 3 DFT-calculated free-energy profile (kcal mol⁻¹) for the DCPD epoxidation with H₂O₂ on TiO₂-anatase (b, orange, magenta) and TiO₂-rutile (c, navy, blue). The free energies are calculated based on the models in Fig. S5 and S6 (ESI†). The numbers in Fig. 4b and c denote the energy barriers. Coordinates of -P2 and -P1 in (b) and (c) are provided in Tables S2–S5, ESI† as representative structures.

We have explored the reaction pathway of DCPD epoxidation and the selectivity differences among the crystalline TiO₂ phases using density functional theory (DFT) calculation (Fig. S3 and S4, ESI†). Fig. 3(a) shows the calculated free energy profiles for DCPD epoxidation with H₂O₂ catalyzed by the TiO₂-anatase and TiO₂-rutile. Raman spectroscopy indicates that peroxo species are formed on the surface of TiO₂, and the reaction starts from the TiO₂ surface with adsorbed H₂O₂ (Ti–H₂O₂*). Formation of the surface peroxo species (Ti–O₂*) proceeds via the generation of hydroperoxo compounds. Once the surface of TiO₂ adsorbed H₂O₂, it firstly generates η¹-coordinated Ti-hydroperoxo compounds, Ti-η¹(OOH), following the release of a water molecule for TiO₂-anatase and TiO₂-rutile, which requires +13.0 (Fig. 3(b)) and +19.8 kcal mol⁻¹ (Fig. 3(c)) at Transition State 1 (TS1), respectively. Subsequently, the thermodynamically more stable η²-coordinated Ti-hydroperoxo compound (Ti-η²(OOH)) is formed. The protonated Ti-peroxo species (Ti–O₂*) can be formed from Ti-η²(OOH) via hydrogen transfer to the adjacent Ti atom overcoming moderate energy barriers of +13.9 (Fig. 3(b)) and +16.3 kcal mol⁻¹ (Fig. 3(c)) at TS2 (Fig. 3(a)), respectively. Along the overall reaction steps, TiO₂-anatase (Fig. 3(a), red color) has lower free energies than TiO₂-rutile (Fig. 3(a), blue color), indicating a higher DCPD conversion rate of TiO₂-anatase, which is in good agreement with our experimental results (Fig. 1(b)).

The two different C=C bonds in DCPD (cyclopentene or norbornene moiety) have different reactivity to TiO₂ surface structures. Thus, the selectivity towards P1 or P2 strongly depends on the direction of oxygen transfer from Ti-peroxo (Ti–O₂*). Adsorption of the DCPD molecule on the surface of TiO₂-rutile is exothermic for both cyclopentene and norbornene with a very small energy difference (−3.6 and −5.1 kcal mol⁻¹ (Fig. 3(c)), respectively). However, O–O bond cleavage with cyclopentene requires a significantly higher energy barrier compared with norbornene (+19.6 and +12.8 kcal mol⁻¹ (Fig. 3(c)) at TS3 (Fig. 3(a)), respectively). This leads to the selective, one-sided oxygen transfer to the C=C double bond in norbornene, which yields nearly 100% P2 selectivity. On the other hand, in the case of TiO₂-anatase, P1 is preferably formed since the adsorption towards the norbornene moiety requires a higher energy barrier for O–O cleavage than cyclopentene (+6.5 and +4.6 kcal mol⁻¹, respectively) (see Fig. 3(b)).

In summary, the surface crystalline phase of TiO₂ catalysts plays a crucial role in selective DCPD epoxidation, not only in the conversion rate of DCPD but also in the product selectivity towards mono-epoxides in cyclopentene and the norbornene moiety. Moreover, the surface oxygen vacancies of TiO₂-blue and TiO₂-black are responsible for the lag-phases at the initial reaction stages, indicating that the formation of the surface peroxo functionalities (Ti–O₂*) is the rate-determining step.

We acknowledge the Dutch research council (NWO) for the funding (LIFT project 731.017.412), the Institute for Basic Science (IBS-R011-D1), and the Korea Medical Device Development Fund (KMDF_PR_20200901_0004). We thank V. Lachman, N. J. Geels, P. F. Collignon and M.C. Mittelmeijer–Hazeleger (UvA) for the technical support.

Conflicts of interest

There are no conflicts to declare.

References

1. T. T. P. Cheung, Kirk-Othmer Encyclopedia of Chemical Technology, John Wiley & Sons, Inc., Hoboken, NJ, USA, 2001, vol. 2.
2. M. Worzakowska, Polym. Bull., 2012, 68, 775–787.
3. R. Gao, Q. Zhu, W. L. Dai and K. Fan, RSC Adv., 2012, 2, 6087–6093.
4. Y. Shen, X. H. Lu, C. C. Wei, X. T. Ma, C. Peng, J. He, D. Zhou and Q. H. Xia, Mol. Catal., 2017, 433, 185–192.
5. S. Bhattacharjee and J. A. Anderson, J. Mol. Catal. A: Chem., 2006, 249, 103–110.
6. M. Sheng and R. Rosenthal, US Pat., 3 631 072, 1971.
7. J. Yan, G. Wu, N. Guan, L. Li, Z. Li and X. Cao, Phys. Chem. Chem. Phys., 2013, 15, 10978–10988.
8. J. Li, G. Lu, G. Wu, D. Mao, Y. Guo, Y. Wang and Y. Guo, Catal. Sci. Technol., 2014, 4, 1268–1275.
9. W. F. Huang, P. Raghunath and M. C. Lin, J. Comput. Chem., 2011, 32, 1065–1081.
10. V. Paunović, M. Rellán-Piñeiro, N. López and J. Pérez-Ramírez, Catal. Today, 2021, 369, 221–226.
11. R. W. Siegel, J. Mater. Res., 1992, 7, 2840–2845.
12. T. Ohno, D. Haga, K. Fujihara, K. Kaizaki and M. Matsumura, J. Phys. Chem. B, 1997, 101, 6415–6419.
13. N. S. Antonova, J. J. Carbó, U. Kortz, O. A. Kholdeeva and J. M. Poblet, J. Am. Chem. Soc., 2010, 132, 7488–7497.
14. Y. C. Lin, C. C. Chang, K. H. Sung, J. F. Lee and S. Cheng, Microporous Mesoporous Mater., 2018, 272, 276–285.
15. P. R. Khoury, J. D. Goddard and W. Tam, Tetrahedron, 2004, 60, 8103–8112.
16. Y. Kim, H. M. Hwang, L. Wang, I. Kim, Y. Yoon and H. Lee, Sci. Rep., 2016, 6, 25212–25221.
17. C. T. K. Nguyen, N. Quang Tran, S. Seo, H. Hwang, S. Oh, J. Yu, J. Lee, T. Anh Le, J. Hwang, M. Kim and H. Lee, Mater. Today, 2020, 35, 25–33.
18. S. Oh, J. S. J.-H. J. Kim, H. M. Hwang, D. Kim, J. S. J.-H. J. Kim, G. H. Park, J. S. J.-H. J. Kim, Y. H. Lee and H. Lee, J. Mater. Chem. A, 2021, 9, 4822–4830.
19. A. Naldoni, M. Altomare, G. Zoppellaro, N. Liu, Š. Kment, R. Zbořil and P. Schmuki, ACS Catal., 2019, 9, 345–364.
20. J. Lee, X. Liu, A. Kumar, Y. Hwang, E. Lee, J. Yu, Y. D. Kim and H. Lee, Chem. Sci., 2021, 12, 9619–9629.
21. X. Chen, L. Liu, P. Y. Yu and S. S. Mao, Science, 2011, 331, 746–750.
22. K. Sobańska, P. Pietrzyk and Z. Sojka, ACS Catal., 2017, 7, 2935–2947.
23. M. Ziolek, I. Sobczak, P. Decyk, K. Sobańska, P. Pietrzyk and Z. Sojka, Appl. Catal., B, 2015, 164, 288–296.
24. M. Ziolek, I. Sobczak, P. Decyk and L. Wolski, Catal. Commun., 2013, 37, 85–91.
25. A. Y. Zhang, T. Lin, Y. Y. He and Y. X. Mou, J. Hazard. Mater., 2016, 311, 81–90.
26. A. H. Boonstra and C. A. H. A. H. A. Mutsaers, J. Phys. Chem., 1975, 79, 1940–1943.
27. J. Zou, J. Gao and Y. Wang, J. Photochem. Photobiol., A, 2009, 202, 128–135.
28. W. F. Zhang, Y. L. He, M. S. Zhang, Z. Yin and Q. Chen, J. Phys. D: Appl. Phys., 2000, 33, 912–916.
29. G. L. Gutsev, B. K. Rao and P. Jena, J. Phys. Chem. A, 2000, 104, 11961–11971.
30. J. Cho, R. Sarangi, J. Annaraj, S. Y. Kim, M. Kubo, T. Ogura, E. I. Solomon and W. Nam, Nat. Chem., 2009, 1, 568–572.
31. A. A. Mikhaylov, A. G. Medvedev, A. V. Churakov, D. A. Grishanov, P. V. Prikhodchenko and O. Lev, Chem. – Eur. J., 2016, 22, 2980–2986.
32. L. Bonato, M. Virot, T. Dumas, A. Mesbah, P. Lecante, D. Prieur, X. Le Goff, C. Hennig, N. Dacheux, P. Moisy and S. I. Nikitenko, Chem. – Eur. J., 2019, 9580–9585.
33. K. Tomobe, E. Yamamoto, D. Kojić, Y. Sato, M. Yasui and K. Yasuoka, Sci. Adv., 2017, 3, e1701400.
34. Á. Ganyecz, P. D. Mezei and M. Kállay, Comput. Theor. Chem., 2019, 1168, 112607.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2cc05305e

‡ These authors contributed to the manuscript equally.

---