Critical impacts of metal cocatalysts on oxidation kinetics and optimal reaction conditions of photocatalytic methane reforming

Abstract

Metal cocatalysts in photocatalysis are typically regarded as promoting only the reduction reactions. Here, we demonstrate that photocatalytic oxidation kinetics and optimal pressure of methane vary significantly with the loading amount of metal cocatalysts. These variations are well described by kinetic analyses treating molecular-level congestion of oxidation intermediates.

---

Photocatalysis represents a promising approach for driving redox reactions using light, surpassing thermodynamic limitations.^1–6 Most advanced photocatalyst systems rely on noble metal cocatalysts to achieve high photocatalytic activity.^7,8 Indeed, numerous studies have demonstrated that the incorporation of metal cocatalysts significantly improves the activity of semiconductor-based photocatalysts.^9,10 Conventionally, it has been widely accepted that the metal cocatalysts act mostly as sinks of photogenerated electrons (e⁻) and serve exclusively as reduction sites.^7,8

Some previous studies offered new insights into the role of metal cocatalysts in photocatalysis. For instance, previous density functional theory calculations¹¹ and infrared spectroscopy¹² have suggested that holes (h⁺) generated through band-gap excitation of semiconductor photocatalysts can be trapped by metal cocatalysts. This hypothesis is based on the simple assumption that h⁺-accumulating metal cocatalysts act as recombination centers for photogenerated e⁻ and h⁺, without investigating the actual photocatalytic performance. By contrast, our recent study combining operando spectroscopy with real-time mass spectrometry demonstrated that e⁻ and h⁺ can be separately captured by different metal cocatalysts, and the metal cocatalysts play key roles not only in promoting reduction reactions but also in accelerating oxidation reactions, while minimizing charge recombination.¹³ Furthermore, metal cocatalysts not only enhance overall photocatalytic activity but also influence the oxidation selectivity.¹³ These recent findings highlight the need for a deeper understanding of the microscopic roles of metal cocatalysts, beyond the conventional view that they merely promote reduction reactions by accumulating only photogenerated e⁻.^7,8

In this study, we provide further experimental evidence that metal cocatalysts directly impact the photocatalytic oxidation processes. The oxidation process is recognized as the rate-determining step of photocatalysis,^14–16 and the oxidation kinetics should be affected by modulation of this step. To examine whether variations in metal cocatalysts alter the dynamics and kinetics of the rate-determining oxidation step, we investigated the effects of loading amounts of metal cocatalysts on photocatalytic methane conversion with water. Methane is the main component of natural gas and a ubiquitous natural resource. Due to its robust C–H bonds, achieving methane conversion under ambient temperatures and pressures poses a significant challenge in advancing sustainable technologies.¹⁷ Conventional thermocatalytic methane conversion requires harsh conditions (>1000 K and >20 atm);¹⁸ in contrast, photocatalysis offers a more sustainable approach by enabling methane activation and oxidation under ambient conditions (∼300 K and ∼1 atm).¹⁷ Herein, we focused on the Pt- or Pd-loaded gallium oxide (Ga₂O₃) particles as robust model systems for photocatalytic methane reforming.^13,15,19 Through controlled experiments under varying methane pressures (P_CH4), we observed that the amount of metal cocatalysts significantly influences the P_CH4 dependence of the photocatalytic performance, resulting in pronounced modulation of the optimal P_CH4 conditions below ambient pressures. This variation suggests that the loading of metal cocatalysts affects the congestion of intermediate species involved in methane oxidation processes. These results highlight the crucial role of metal cocatalysts as active sites also for the oxidation reaction.

β-Ga₂O₃ is a well-known d¹⁰ photocatalyst² with stable activity and robustness.²⁰ The four β-Ga₂O₃ samples containing either 0.01 wt% or 1 wt% of metal Pt or Pd cocatalysts (Pt(0.01 wt%)/Ga₂O₃, Pd(0.01 wt%)/Ga₂O₃, Pt(1 wt%)/Ga₂O₃, Pd(1 wt%)/Ga₂O₃) were prepared by the impregnation method (see Note S1-1, ESI,† for details). As described in detail in Note S1-2 (ESI†), the photocatalytic performance of these metal-loaded Ga₂O₃ samples was evaluated by irradiating a deep ultraviolet (UV) lamp (∼90 mW cm⁻² at 260 ± 15 nm) at various P_CH4 under a fixed water vapor pressure (P_H2O) of 2 kPa. Notably, the photocatalytic activity under LED light irradiation at 390 nm was low, although loading the Pt and Pd cocatalysts enhanced the light absorption in this wavelength region (Fig. S1-5, ESI†).¹³ This indicates that e⁻ and h⁺ species generated by the excitation of the metal cocatalysts²¹ have little contribution, whereas those generated by the band-gap excitation of Ga₂O₃ dominantly induce photocatalytic methane conversion.¹³ Under the irradiation of a deep UV lamp, the sample temperature was ∼318 K and a single layer of adsorbed water molecules covered the sample surfaces at P_H2O = 2 kPa (∼20% relative humidity for 318 K samples)^22–24 as described in Note S1-3 (ESI†).

Fig. 1 illustrates the P_CH4 dependence of the formation rates of H₂ (R_H2) and CO₂ (R_CO2) for the four photocatalyst samples. The rate of H₂ evolution was approximately four times higher than that of CO₂ formation (R_CO2 ≈ 0.25R_H2), indicating nearly stoichiometric CO₂ production in the steam methane reforming reaction (CH₄ + 2H₂O → CO₂ + 4H₂). In addition, C₂H₆ derived from the side reaction (2CH₄ → C₂H₆ + H₂)²⁵ was also detected, as will be discussed later (Fig. 3).

Fig. 1 P _CH4 profiles of photocatalytic steam methane reforming. Formation rates of H₂ and CO₂ for (a) Pt(1 wt%)/and Pd(1 wt%)/Ga₂O₃ photocatalysts, and (b) Pt(0.01 wt%)/and Pd(0.01 wt%)/Ga₂O₃ photocatalysts under UV irradiation as a function of P_CH4 at P_H2O = 2 kPa. Curve fitting results based on eqn (3) with U = 40 kJ mol⁻¹ and eqn (4) with U = 37 kJ mol⁻¹ are also shown in (a) and (b), respectively.

As reported previously for the Pt(1 wt%)/Ga₂O₃ and Pd(1 wt%)/Ga₂O₃ photocatalysts,^13,15R_H2 and R_CO2 showed a strong dependence on P_CH4; the rates increased sharply with P_CH4 below ∼30 kPa (0.3 atm) and nearly reached saturation value at ∼100 kPa (1 atm). In contrast, we found that the P_CH4 profiles for the Pt(0.01 wt%)/Ga₂O₃ and Pd(0.01 wt%)/Ga₂O₃ photocatalysts differ markedly from those for the 1 wt% metal-loaded samples (Fig. 1b); while R_H2 and R_CO2 increased with P_CH4 below ∼20 kPa, they began to decline at P_CH4 above ∼20 kPa. Thus, for the 1 wt% metal-loaded samples, the photocatalytic performance for methane conversion is almost maximized at ambient pressure (∼1 atm), whereas the optimal P_CH4 is around 20 kPa, well below 1 atm, for the 0.01 wt% metal-loaded samples. The observed variation in the P_CH4 profile (Fig. 1) indicates that the microscopic redox reaction kinetics of photocatalytic steam methane reforming is affected by the loading amount of metal cocatalysts. Beyond the conventional assumption that the role of metal cocatalysts is limited solely to reduction reactions in the semiconductor bandgap photoexcitation scheme,^7,8 our experimental results clearly show the impact of metal cocatalysts not only on reduction (H₂ evolution) but also on oxidation (CO₂ formation) processes in this widely used photoexcitation scheme.

To provide microscopic insights on the influence of metal cocatalyst loadings on the reaction kinetics, detailed analysis was carried out on the observed P_CH4 dependence of the photocatalytic performance (Fig. 1). In general, the CH₄ oxidation process toward CO₂ can be divided into three steps (Fig. 2a and b):^13,15 (i) dissociative adsorption and desorption of CH₄ at vacant active sites, (ii) sequential reactions involving surface intermediate species,²⁶ and (iii) desorption of the final products, primarily CO₂ (see also Note S2, ESI†). The C–H cleavage of CH₄ in the first step is initiated by hydrogen abstraction from photoactivated interfacial water species,¹⁵ resulting in the formation of activated methane species (X₁ = ˙CH₃) that subsequently chemisorb onto the catalyst surfaces (Fig. 2a and b). This process leads to the P_CH4 dependence of the coverage of ˙CH₃ (θ_CH3) similar to the Langmuir adsorption isotherm as follows:^13,15

θ_CH3 = KP_CH4/(1 + KP_CH4), (1)

where K is the equilibrium constant for the first step (dissociative adsorption of methane and desorption of methyl radical; CH_4(gas) + ˙OH_(ad) ↔ ˙CH_3(ad) + H₂O_(ad)), expressed as K ≡ k_ad/k_de ∝  exp(U/k_BT_s). Here, k_B and T_s represent the Boltzmann constant and surface temperature, respectively, and U denotes the stabilization energy of the ˙CH_3(ad) intermediate (see Note S2-1 for details of the kinetic analysis, ESI†).

Fig. 2 Reaction scheme and kinetic analysis of photocatalytic steam methane reforming. (a) Surface reaction pathway for the conversion of CH₄ to CO₂, where X_i (i = 1, 2,…,8) represents reaction intermediates; specifically, X₁ and X₈ denote methyl radical (˙CH₃) and adsorbed CO₂, respectively. For simplicity, the contribution of OH radical species (˙OH) generated via water oxidation (H₂O + h⁺ → ˙OH + h⁺)¹⁵ is omitted in the diagram. (b) Possible reaction intermediate species from CH₄ to CO₂ in Fig. 2a through hydrogen abstraction or hydroxylation by photoactivated water species.²⁶ (c) and (d) P_CH4 dependence of R_CO2 given by (c) eqn (3) and (d) eqn (4) at different values of U (35, 40, and 45 kJ mol⁻¹).

Notably, the product yields linearly increased with irradiation time,^13,15 indicating that the reactions proceeded under steady-state conditions. In the case where the reaction of adsorbed ˙CH₃ is rate-determining,^13,15R_CO2 under steady-state conditions is dominated by the reaction of ˙CH₃ as follows:

where is the rate constant factor for the rate-determining reaction process, and θ_v ≡ 1 − ∑θ_Xi (θ_Xi is the coverage of the i-th intermediate species (i = 1, 2,…,8, see Fig. 2a and b) that block the reaction sites) represents the fraction of vacant surface sites available for the forward reaction of ˙CH₃ species toward CO₂ (see also Note S2 for details, ESI†). When the density of surface reaction sites is high enough that few active sites are occupied (θ_v ≈ 1) in steady reaction conditions, eqn (2) is approximated as

The P_CH4 profiles of R_CO2 at typical values of U are displayed in Fig. 2c. As U increases, the reaction activity tends to start increasing at the lower P_CH4. Then, with increasing P_CH4, reaction activity monotonically increases as an upward convex curve and then saturates in response to the abundance of the rate-determining intermediate species.

In contrast, when the density of active sites is not sufficiently high, the coverage of intermediate species competitively affects the fraction of vacant surface sites. Since the majority of the surface intermediates are ˙CH₃ (∑θ_Xi ≈ θ_CH3) in the case where the reaction of adsorbed ˙CH₃ intermediates is rate-determining,^13,15 the fraction of vacant sites is approximated as θ_v ≈ 1 – θ_CH3, yielding the following form of eqn (2):

In this case, although the reaction activity starts to increase at the lower P_CH4 region, it begins to decrease with increasing P_CH4 after reaching a maximum value (Fig. 2d). This characteristic P_CH4 profile arises from a balance between two competing effects (Fig. S2-3, ESI†): the positive impact of increasing the abundance of the rate-determining intermediate species and the negative impact of molecular-level congestion due to the reduced availability of vacant active sites for further reactions of the rate-determining intermediate species (X₁ → X₂ in Fig. 2a and b). It should also be noted that, even if the rate-determining step is not the forward reaction of ˙CH₃ (X₁) but rather another surface reaction involving different intermediate species (X_i → X_i+1 in Fig. 2a and b), the P_CH4 dependencies similar to eqn (3) and (4) are derived as discussed in Note S3 (ESI†).

For the Pt(1 wt%)/Ga₂O₃ and Pd(1 wt%)/Ga₂O₃ photocatalysts, P_CH4 profiles of R_CO2 were well-fitted to eqn (3) with a stabilization energy U of 40 kJ mol⁻¹ (Fig. 1a), as reported in our previous studies.^13,15 On the other hand, the reaction activity for the Pt(0.01 wt%)/Ga₂O₃ and Pd(0.01 wt%)/Ga₂O₃ photocatalysts behaved differently, showing a better fit to eqn (4) compared to eqn (3) with a similar U value of 37 kJ mol⁻¹ (Fig. 1b). These curve-fitting analyses indicate that a sufficient number of active sites are provided for the 1 wt% loading of metal cocatalysts, avoiding competitive occupation of active sites by intermediate species. In contrast, the 0.01 wt% loading provides a limited number of active sites for intermediate species, resulting in the competition of positive and negative effects of active site occupation by intermediate species. Therefore, the observed modulation of reaction kinetics upon loading amount of metal cocatalysts provides compelling evidence that the metal cocatalysts themselves function as reaction fields for the photocatalytic oxidation.¹³ These results further suggest that the photogenerated e⁻ and h⁺ migrate separately from Ga₂O₃ to the metal cocatalyst particles,¹³ enabling certain metal cocatalysts to act as reduction sites by capturing electrons, while others function as oxidation sites by trapping holes, as discussed in detail in Note S4 (ESI†).

Additional important evidence supporting that metal cocatalysts are deeply involved in molecular-level dynamics and congestion of photocatalytic oxidation processes has been obtained. This is demonstrated by focusing on the simultaneously induced side-reaction of methane into ethane (2CH₄ → C₂H₆ + H₂). Fig. 3 illustrates the P_CH4 dependence of ethane formation rates (R_C2H6) for the four photocatalyst samples. In contrast to R_CO2 (Fig. 1), the P_CH4 profiles of R_C2H6 not only vary depending on the loading amount of metal cocatalysts but also vary depending on the metal element of cocatalysts (Fig. 3). As detailed in our previous study,¹³ the P_CH4 profiles for the Pt(1 wt%)/Ga₂O₃ and Pd(1 wt%)/Ga₂O₃ photocatalysts (Fig. 3a) were described by the second-order and first-order reactions of the surface adsorbed ˙CH₃ intermediate species, respectively, as follows:

where is a rate constant. The apparent difference of the reaction order indicates that the coupling of ˙CH₃ intermediates (2˙CH₃ → C₂H₆) occurs on the catalyst surfaces (eqn (5) and Fig. S5-1a, S5-2, ESI†) for the Pt(1 wt%)/Ga₂O₃, while it proceeds in the gas phase for the Pd(1 wt%)/Ga₂O₃ (eqn (6) and Fig. S5-1b, S5-5, ESI†).¹³

Fig. 3 P _CH4 profiles of photocatalytic ethane formation from methane. Formation rates of C₂H₆ for (a) Pt(1 wt%)/and Pd(1 wt%)/Ga₂O₃ photocatalysts, and (b) Pt(0.01 wt%)/and Pd(0.01 wt%)/Ga₂O₃ photocatalysts under UV irradiation at P_H2O = 2 kPa. Curve fitting results based on eqn (5) and (6) with U = 36 kJ mol⁻¹, eqn (7) with U = 39 kJ mol⁻¹, and eqn (8) with U = 37 kJ mol⁻¹ are also shown in (a) and (b).

As the main reaction (CH₄ + 2H₂O → CO₂ + 4H₂), we found that the limited 0.01 wt% loading of metal cocatalysts also has a significant impact on the P_CH4 profiles of R_C2H6 (Fig. 3b). For the Pt(0.01 wt%)/Ga₂O₃ photocatalyst, R_C2H6 steeply declined at P_CH4 > 10 kPa after the initial increase at P_CH4 < 10 kPa. A similar trend was observed for the Pd(0.01 wt%)/Ga₂O₃ photocatalyst, although the decline was more gradual compared to the Pt(0.01 wt%)/Ga₂O₃ photocatalyst. These non-monotonic P_CH4 profiles are derived from the competition between the positive impact of increasing the abundance of the rate-determining ˙CH₃ intermediate species and the negative impact of their molecular-level congestion due to the reduced availability of vacant active sites for further reactions. Then, these observed P_CH4 profiles were well-fitted with the following equations explicitly incorporating the effect of competing occupation of limited active sites by intermediate species (θ_v ≈ 1 – θ_CH3).

The decline of R_C2H6 is more significant for eqn (7) than eqn (8) (Fig. S5-7b, ESI†). As described in detail in Note S5 (ESI†), the apparent difference of the reaction order for θ_CH3θ_v indicates that the homocoupling of methyl radical intermediates (2˙CH₃ → C₂H₆) occurs on the catalyst surfaces (eqn (7) and Fig. S5-1a, S5-3, ESI†) for the Pt(0.01 wt%)/Ga₂O₃ sample, while it proceeds in the gas phase for the Pd(0.01 wt%)/Ga₂O₃ sample (eqn (8) and Fig. S5-1b, S5-6, ESI†).¹³ Even though the homocoupling of ˙CH₃ proceeds in the gas phase for the Pd(0.01 wt%)/Ga₂O₃ sample (Fig. S5-1b, ESI†), manifestation of the negative P_CH4 impact in the apparent first order reaction suggests that the desorption of ˙CH₃ into the gas phase occurs after it migrates from its initial adsorption site to a nearby vacant site that is more preferable for desorption (see Fig. S5-4, ESI† for more precise picture). Note that, in the situation where the competitive occupation of active sites by intermediate species is negligible (θ_v ≈ 1), eqn (7) and (8) correspond to eqn (5) and (6), respectively. Thus, our results clearly demonstrate that the competitive occupation of active surface sites by reaction intermediates and the resulting modulation of reaction kinetics emerges also in the side reaction pathway of methane partial oxidation (CH₄ → ˙CH₃ → C₂H₆). Owing to this effect, the optimal P_CH4 condition for the photocatalytic ethane formation also significantly varied with the amount of metal loading (Fig. 3a and b). Therefore, the observed modulation of ethane formation kinetics also provides compelling evidence that the metal cocatalysts themselves functions as active sites for oxidative photoactivation. Furthermore, the homocoupling pathway for ethane formation is independent of the loading amount of metal cocatalysts (1 wt% or 0.01 wt%) but varies with the elements of metal cocatalysts used (Pt or Pd). This is also a manifestation of the phenomena that the metal cocatalysts fundamentally influence oxidation kinetics and dynamics.

Finally, we emphasize that our concept of metal cocatalysts also provides a rationale for previously reported, yet incomprehensible, results in photocatalytic oxidation. For instance, variations in the activity of photocatalytic oxidation have been reported for semiconductor photocatalysts loaded with metal cocatalysts of different elements and/or loading amount.^27–29 Nevertheless, metal cocatalysts have traditionally been regarded solely as the trapping site of e⁻ and the resultant promotors of reduction reactions.^7,8 This interpretation arises from the seemingly plausible assumption that more efficient oxidation reactions are simply driven by an increased number of h⁺ that avoid annihilative charge recombination with e⁻ due to the preferential trapping/consumption of e⁻ facilitated by metal cocatalysts.^11,12 Beyond this classical picture, our results provide a series of solid evidence that the metal cocatalyst actively takes part in photocatalytic oxidation and modulates oxidation kinetics/dynamics, shedding new light on the complex interplay between cocatalyst loading and interfacial reaction.

This work was supported by JSPS KAKENHI [JP22H00296, JP24H02205, JP22KJ1427, JP22KJ3098], JST CREST [JPMJCR22L2], JST ACT-X [JPMJAX24D7] and Demonstration Project of Innovative Catalyst Technology for Decarbonization through Regional Resource Recycling, the Ministry of the Environment, Government of Japan.

Data availability

The data and analytical details supporting this article have been included as part of the ESI.†

Conflicts of interest

There are no conflicts to declare.

Notes and references

1. A. L. Linsebigler, G. Lu and J. T. Yates, Chem. Rev., 1995, 95, 735.
2. A. Kudo and Y. Miseki, Chem. Soc. Rev., 2009, 38, 253.
3. P. Zhang and X. W. D. Lou, Adv. Mater., 2019, 31, 1900281.
4. K. Villa and M. Pumera, Chem. Soc. Rev., 2019, 48, 4966.
5. Q. Wang and K. Domen, Chem. Rev., 2020, 120, 919.
6. M. Yamauchi, H. Saito, T. Sugimoto, S. Mori and S. Saito, Coord. Chem. Rev., 2022, 472, 214773.
7. J. Yang, D. Wang, H. Han and C. Li, Acc. Chem. Res., 2013, 46, 1900.
8. J. Ran, J. Zhang, J. Yu, M. Jaroniec and S. Z. Qiao, Chem. Soc. Rev., 2014, 43, 7787.
9. R. Su, R. Tiruvalam, A. J. Logsdail, Q. He, C. A. Downing, M. T. Jensen, N. Dimitratos, L. Kesavan, P. P. Wells, R. Bechstein, H. H. Jensen, S. Wendt, C. R. A. Catlow, C. J. Kiely, G. J. Hutchings and F. Besenbacher, ACS Nano, 2014, 8, 3490.
10. T. Kotani, K. Ogawa, H. Suzuki, K. Kato, O. Tomita, A. Yamakata and R. Abe, EES Catal., 2023, 1, 255.
11. C. L. Muhich, Y. Zhou, A. M. Holder, A. W. Weimer and C. B. Musgrave, J. Phys. Chem. C, 2012, 116, 10138.
12. M. Yoshida, A. Yamakata, K. Takanabe, J. Kubota, M. Osawa and K. Domen, J. Am. Chem. Soc., 2009, 131, 13218.
13. H. Saito, H. Sato, T. Higashi and T. Sugimoto, Angew. Chem., Int. Ed., 2023, 62, e202306058.
14. X. Ma, Y. Shi, J. Liu, X. Li, X. Cui, S. Tan, J. Zhao and B. Wang, J. Am. Chem. Soc., 2022, 144, 13565.
15. H. Sato, A. Ishikawa, H. Saito, T. Higashi, K. Takeyasu and T. Sugimoto, Commun. Chem., 2023, 6, 8.
16. H. Zou, Y. Qi, S. Du, Y. Bao, X. Xin, W. Fan, Y. Xiao, S. Jin, Z. Feng and F. Zhang, J. Am. Chem. Soc., 2024, 146, 28182.
17. X. Meng, X. Cui, N. P. Rajan, L. Yu, D. Deng and X. Bao, Chem, 2019, 5, 2296.
18. A. I. Olivos-Suarez, À. Szécsényi, E. J. M. Hensen, J. Ruiz-Martinez, E. A. Pidko and J. Gascon, ACS Catal., 2016, 6, 2965.
19. H. Sato and T. Sugimoto, J. Am. Chem. Soc., 2024, 146, 24800.
20. F. Amano, C. Akamoto, M. Ishimaru, S. Inagaki and H. Yoshida, Chem. Commun., 2020, 56, 6348.
21. S. Luo, X. Ren, H. Lin, H. Song and J. Ye, Chem. Sci., 2021, 12, 5701.
22. K. Shirai, T. Sugimoto, K. Watanabe, M. Haruta, H. Kurata and Y. Matsumoto, Nano Lett., 2016, 16, 1323.
23. Z. Lin, H. Saito, H. Sato and T. Sugimoto, J. Am. Chem. Soc., 2024, 146, 22276.
24. F. Amano, A. Ishikawa, H. Sato, C. Akamoto, S. P. Singh, S. Yamazoe and T. Sugimoto, Catal. Today, 2024, 426, 114375.
25. G. Wang, X. Mu, J. Li, Q. Zhan, Y. Qian, X. Mu and L. Li, Angew. Chem., Int. Ed., 2021, 60, 20760.
26. V. S. Bagotzky, Y. B. Vassiliev and O. A. Khazova, J. Electroanal. Chem., 1977, 81, 229.
27. H. Yuzawa, S. Yoneyama, A. Yamamoto, M. Aoki, K. Otake, H. Itoh and H. Yoshida, Catal. Sci. Technol., 2013, 3, 1739.
28. Y. Negishi, Y. Matsuura, R. Tomizawa, W. Kurashige, Y. Niihori, T. Takayama, A. Iwase and A. Kudo, J. Phys. Chem. C, 2015, 119, 11224.
29. S. Mori and S. Saito, Green Chem., 2021, 12, 810.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cc06774f

---