Pd composition and dispersion control selectivity in photocatalytic methane oxidation

Abstract

Selective methane oxidation to methanol under mild conditions presented a significant challenge due to the high bond dissociation energy of methane and the difficulty in achieving high selectivity. This study investigated how Pd composition and dispersion influenced photocatalytic oxidation of methane over brookite TiO₂ (BTO) nanorods, using molecular O₂ and water as oxidants. In situ time-resolved transient absorption (TA) spectroscopy in the visible to mid-infrared (IR) range, along with X-ray photoelectron spectroscopy (XPS) analysis before and after methane oxidation, revealed that BTO nanorods with high Pd loading and lower Pd dispersion exhibited significant electron trapping in the Pd/PdO nanoparticles under light irradiation. This electron trapping promoted the self-reduction of PdO to metallic Pd (Pd⁰), which in turn drove the selective production of CH₃OH with 98% selectivity. In contrast, a moderate Pd loading and dispersion on the BTO nanorods enhanced electron transfer to O₂, reducing electron trapping, and resulted in a higher concentrations of mixed Pd⁰/PdO nanoparticles, favoring the formation of primary oxygenates – CH₃OOH and CH₃OH. DFT calculations and experimental findings showed that Pd⁰ facilitates the direct three-electron reduction of O₂ to *OH or ˙OH, which subsequently coupled with *CH₃ or ˙CH₃ to form CH₃OH. Meanwhile PdO promotes the one-electron reduction of O₂ to *OOH or ˙OOH, leading to CH₃OOH formation. This work provides valuable insights into the design of efficient photocatalysts for selective methane oxidation and underscores the critical role of Pd composition and dispersion in modulating charge dynamics and steering product selectivity in methane.

---

Introduction

Methane (CH₄), a potent greenhouse gas and key natural gas component, holds promise as a feedstock for methanol (CH₃OH), a clean transportable fuel and valuable chemical building block. However, its strong C–H bond (∼439 kJ mol⁻¹) complicates selective oxidation.^1–4 Currently, industrial CH₃OH synthesis relies on high-temperature steam reforming (>700 °C) to produce syngas.^1–4 Direct methane oxidation is a more energy-efficient alternative^5,6 but faces challenges, such as high reaction temperatures,^1,2 or the use of costly or environmentally harmful oxidants (e.g., H₂O₂ or N₂O).^1–4

Photocatalytic methane oxidation under mild conditions using molecular O₂ and water as oxidants presents a promising solution. Titanium dioxide (TiO₂), particularly in its mixed anatase/rutile^7–14 and pure anatase forms,^15–22 is widely studied for methane oxidation due to its high stability, efficient light-to-energy conversion, and cost-effectiveness. To enhance charge separation and surface reactions, metal cocatalysts (e.g., Au,^7,12,19,21 Ag,^7,15 Pd,^7,12,14,18 Ni¹¹) and oxygen vacancy^10,16,17 modifications are often employed. However, over-oxidation of products to formaldehyde (HCHO) and carbon dioxide (CO₂) diminishes CH₃OH selectivity.^{8,11–14,16,17,19} Although the anatase–rutile interface promotes charge separation and photocatalytic performance, excessive hydroxyl radical (˙OH) generation during water photo-oxidation contributes to over-oxidation, limiting CH₃OH yield.⁷

Surface modifications with hole acceptors^7,10,14,20 like CoO_X,⁷ CuO_X,¹⁴ and N–Fe₂Ti₃O₉,²⁰ and W-doping⁹ (to adjust the valence band potential), help control ˙OH formation and improve oxygenate selectivity and overall efficiency. Despite these improvements, over-oxidation to HCHO and CO₂ persists.^7,9,14 Although alternative catalysts such as ZnO,^23,24 BiVO₄,^25,26 WO₃,^27,28 and In₂O₃ (ref. 29 and 30) show potential, they face long-term stability issues, highlighting the need for further development of TiO₂-based photocatalysts with better selectivity and reduced over-oxidation.

Brookite TiO₂, a less-studied polymorph, shows significant potential as a photocatalyst.^31–39 Previous studies have demonstrated that brookite TiO₂ nanorods with well-developed {210} and {212} facets can be synthesized through hydrothermal reactions using TALH (dihydroxybis(ammonium lactato) titanium(IV)) and urea as precursors,^34,35,37,38 with sodium lactate enhancing crystallinity.^32,39 These {210} and {212} facets serve as reduction and oxidation sites, respectively, improving electron–hole separation and photocatalytic performance.^33–35,38 Additionally, brookite TiO₂ possesses a moderate electron trap depth, which could facilitate the regulation of ˙OH radical production during oxygen reduction, thereby helping to address the issue of over-oxidation.³⁶ Photo-deposition of metal cocatalysts is a promising approach for selectively loading them onto reduction-active sites, thereby enhancing photocatalytic performance. Compared to non-site-selective chemical reduction methods, photo-deposition allows for more precise placement of cocatalyst particles, maximizing their effectiveness.^40,41 While Au, Ag, and Pd are active cocatalysts for generating reactive oxygen species in CH₄ oxidation, Au and Ag tend to exhibit weak interactions with the TiO₂ surface, leading to particle aggregation at high metal loadings and reduced active sites.^37,42–44 In contrast, Pd forms a strong metal-support interaction with TiO₂, facilitating better dispersion and smaller, more effective Pd nanoparticles for photocatalytic reaction.⁴⁴

Herein, we uncover how the composition and dispersion of Pd critically govern the photocatalytic oxidation of methane to primary oxygenates – methanol (CH₃OH) and methyl hydroperoxide (CH₃OOH) – over brookite-phase TiO₂ nanorods. By engineering Pd-modified BTO (Pd/BTO) nanorods with controlled Pd⁰/PdO ratios and dispersion profiles, we systematically evaluated their activity using molecular O₂ and water as oxidants. In situ time-resolved transient absorption (TA) spectroscopy, spanning the visible to mid-infrared (IR) range, alongside X-ray photoelectron spectroscopy (XPS) analysis conducted before and after methane oxidation, revealed that catalysts with higher Pd loadings and lower dispersion exhibited pronounced electron trapping under light irradiation. This trapping effect facilitates the self-reduction of PdO to metallic Pd⁰, thereby enabling highly selective CH₃OH production with 98% selectivity. In contrast, lower Pd loadings and higher dispersion favor electron transfer to O₂ and reduce electron trapping, leading to a mixed Pd⁰/PdO state that promotes the formation of both CH₃OOH and CH₃OH. Density functional theory (DFT) calculations, supported by experimental evidence, show that Pd⁰ mediates the three-electron reduction of O₂ to *OH or ˙OH species, which couple with methyl intermediates (*CH₃ or ˙CH₃) to form CH₃OH. Meanwhile, PdO catalyzes the one-electron reduction of O₂ to *OOH or ˙OOH, steering the reaction toward CH₃OOH. These results highlight the fundamental role of Pd composition and dispersion in directing charge carrier behavior and controlling product selectivity during methane oxidation.

Results and discussion

Morphological and physicochemical characterization

The BTO nanorods were synthesized using hydrothermal reactions with aqueous solutions containing TALH, sodium lactate and urea. Pd/BTO samples with different amount of Pd nanoparticles (0.05, 0.2, 0.4, 1 and 3 wt%) were photo-deposited on the BTO nanorod surface by a conventional photo-reduction technique. Inductively coupled plasma optical emission spectrometry (ICP-OES) showed that in all the Pd/BTO samples, the Pd precursor was loaded onto the BTO nanorod surface, with the nominal Pd loading corresponding to the actual Pd content in the samples (Table S1). These Pd/TiO₂ samples are hereafter named as x%Pd/BTO, where x% is the weight percentage of Pd photo-deposited on the BTO nanorods. X-ray diffraction (XRD) patterns (Fig. S1) showed the as-prepared BTO and Pd/BTO nanorods exhibited only the diffraction peaks of brookite phase TiO₂ (JCPDS no. 29-1360). No characteristic diffraction peaks associated with anatase or rutile phase TiO₂ was observed indicating the purity of the BTO nanorods. The Raman spectrum of BTO also indicated the formation of a pure-phase brookite TiO₂ (Fig. S2).³⁸ The scanning electron microscopy (SEM) image of the BTO sample (Fig. S3) showed the BTO sample exhibited a nanorod morphology with an average length and width of 112 ± 17 nm and 31.5 ± 3.8 nm, respectively.

Fig. 1a–f and S4 show the high-resolution transmission electron microscopy (HR-TEM) images of BTO and Pd/BTO with varying Pd loadings. HR-TEM imaging revealed that the BTO nanorods exhibited exposed {210} crystalline body facets, which were surrounded by {212} facets (Fig. 1a). Additionally, the HR-TEM images of Pd/BTO samples demonstrated that the Pd nanoparticles were selectively photo-deposited on the {210} facet of BTO. As the Pd concentration was increased from 0.05% to 3%, more Pd nanoparticles appeared on the {210} facet of BTO (Fig. 1b–d), indicating that the {210} facet of BTO acted as a reduction site. In contrast, previous research by Ohno et al. demonstrated that the {212} facets of BTO served as oxidation sites through the photo-deposition of PbO₂.³³ Lattice spacing imaging of the Pd nanoparticles on the BTO nanorods revealed that the deposition of three distinct types of Pd nanoparticles on the {210} facet of BTO. The first type of Pd nanoparticles consisted of pure metallic Pd⁰ nanoparticles, which were highly crystalline and exhibited a single lattice spacing of 0.224 nm, corresponding to the {111} planes of metallic Pd⁰ (Fig. 1e).⁴⁵ The second type of Pd nanoparticles observed were metallic Pd⁰/PdO nanoparticles, which displayed two distinct sets of lattice spacings: one corresponding to metallic Pd⁰ with a {111} lattice spacing, and the other to PdO with a lattice spacing of 0.215 nm, aligning with the {110} planes of PdO (Fig. 1f).⁴⁶ The third type of Pd particles identified were PdO {110} nanoparticles, as confirmed by the high-angle annular dark-field scanning transmission electron microscope (HAADF-STEM) image (Fig. 1g). Fig. 1h illustrates a schematic representation of the oxidation and reduction sites corresponding to the {212} and {210} facets of a BTO nanorod. In the typical mechanism of photo-deposition, photogenerated electrons are produced on the surface of a semiconductor (e.g., BTO) upon photoexcitation. Metal precursors such as Pd²⁺, are initially reduced on this surface, followed by further reduction of the metal-deposited layer. As a result, the core of the cocatalyst nanoparticles consists of highly reduced metal species (e.g., Pd⁰), while the exposed surface forms an oxide layer (e.g., PdO) due to electron transfer from the metal surface to the aqueous environment (water) during the photo-deposition process.⁴⁷

Fig. 1 HRTEM images of (a) BTO nanorods, (b) 0.05%Pd/BTO, (c) 0.4%Pd/BTO, (d) 3%Pd/BTO, and (e) and (f) Pd⁰ and PdO/Pd⁰ cocatalyst nanoparticles, respectively, (g) HAADF-STEM image of a PdO nanoparticle. (h) Schematic representation of the oxidation and reduction sites of BTO nanorods. (i) Band gap structure of BTO, along with the redox potentials of proposed WOR, ORR, and MOR reactions.

Fig. S5 shows the Pd particle size and distribution for Pd/BTO with different Pd loadings. It is observed, on average, the size of the Pd nanoparticles increased marginally from 2.0 ± 0.8 nm to 3.0 ± 0.8 nm with increasing Pd loading. Specifically, the average diameters of the Pd nanoparticles for 0.05%Pd/BTO, 0.2%Pd/BTO, 0.4%Pd/BTO, 1%Pd/BTO and 3%Pd/BTO are 2.0 ± 0.8, 2.4 ± 0.6, 2.6 ± 0.7, 3.1 ± 0.7 and 3.0 ± 0.8 nm, respectively. HRTEM imaging revealed no evidence of particle agglomeration, even at higher Pd loadings. The Pd nanoparticles remained well-dispersed on the BTO nanorods with narrow size distributions, based on measurements of over 100 individual particles per sample. No large clusters or fused particles were observed, indicating that agglomeration did not occur during the photo-deposition process. Additional representative TEM images are provided in the SI to support this observation (Fig. S6–S8). Thus, as the Pd loading increased, the average particle size increased slightly, leading to a decrease in Pd dispersion due to a lower proportion of surface-exposed atoms relative to the total Pd content.

XPS was used to investigate the surface composition of BTO and Pd/BTO samples, as shown in Fig. S9. The fitted Ti 2p spectra revealed a Ti 2p_3/2 at 458.5 eV, corresponding to Ti⁴⁺ (Fig. S9a).^14,15 The fitted O 1s spectra primary exhibited the lattice oxygen (O_lat) peak at 529.9 eV, with a minor concentration of O vacancies at 531.7 eV, suggesting that the BTO nanorods possessed high crystallinity (Fig. S9b).⁴⁸ The fitted Pd 3d spectra displayed the two peaks for the Pd 3d_5/2 band: one at 335.0 eV, attributed to metallic Pd⁰, and another at 336.5 eV, corresponding to PdO (Fig. S9c).^49,50 The Pd 3d spectra did not show any evidence of PdCl₂, the precursor used in the photo-deposition of Pd, which would exhibit a peak around 337.9 eV if present.⁵¹ The fractional compositions of Pd⁰ and PdO in the Pd/BTO samples were estimated based on the area ratios of Pd⁰ or PdO peaks to the total area of both the metallic Pd⁰ and PdO peaks. As shown in Fig. S9d, the concentration of metallic Pd⁰ in the 0.05%Pd/BTO sample was 29.6%, which increased to 54.9% and 66.7% for the 0.4%Pd/BTO and 3%Pd/BTO samples, respectively. The increase in concentration of metallic Pd⁰ suggests that the Pd nanoparticles on the BTO nanorod surface undergo greater reduction as the Pd dispersion increases.

Fig. S10 shows the UV-vis diffused reflectance spectra (UV-vis DRS) of BTO and Pd/BTO. The minimum absorbance at λ = ∼380 nm was attributed to the band gap charge transition from the O 2p state (valence band) to Ti 3d state (conduction band) of BTO. The Tauc plot of BTO showed an optical band gap (E_g) of 3.2 eV (Fig. S11a). Mott–Schottky analysis of the BTO sample at different measurement frequencies showed positive slopes with increasing potential, confirming an n-type characteristic (Fig. S11b). The flatband potential was estimated to be approximately −0.68 V. Assuming that the flatband potential corresponds to the conduction band potential for n-type semiconductors, the valence band potential of BTO with E_g of 3.20 eV was determined to be +2.52 V, which also coincided with the measured valence band potential obtained from XPS (Fig. S11c). The proposed band gap structure of BTO, along with the water oxidation reaction (WOR), oxygen reduction reaction (ORR), and methane oxidation reaction (MOR) redox potentials, is shown in Fig. 1i. The BTO nanorods displayed appropriate conduction and valence band potentials to effectively drive these reactions.

MOR performance

The as-prepared BTO and Pd/BTO were tested for photocatalytic oxidation of CH₄ with O₂. The photocatalytic MOR were conducted at room temperature under 19 bar CH₄ and 1 bar O₂ in a batch reactor with 50 mL of water and a catalyst concentration of 0.4 g L⁻¹. The reactions were performed for 3 hours with a 300 W Xe lamp (wavelength λ = 320–780 nm) as the light source at a light intensity of 200 mW cm⁻². In this study, CH₃OH and CH₃OOH are considered the primary desired oxygenates of MOR. Fig. 2a shows the MOR product yield and the selectivity for primary oxygenates on BTO and Pd/BTO. The MOR oxygenates detected were primarily CH₃OOH and CH₃OH with a trace amount of HCHO. Gaseous CO₂ was also detected. Bare BTO exhibited the lowest MOR product yield. However, in the presence of Pd cocatalysts, the MOR product yield increased significantly, following a volcanic trend with higher Pd loading. 0.4%Pd/BTO exhibited the highest MOR productivity among the tested samples, achieving a rate of 1.95 mmol g⁻¹ h⁻¹. It also exhibited a high primary oxygenate selectivity of 87.8%, indicating excellent selectivity towards desired products. Furthermore, 0.4%Pd/BTO showed remarkable stability over repeated use, with negligible decreases in both productivity and selectivity after four consecutive reaction cycles (Fig. 2b), underscoring its potential practical and sustainable catalytic applications.

Fig. 2 (a) MOR product yields and primary oxygenate selectivities (sel.) for BTO and Pd/BTO with different Pd loadings evaluated under identical conditions: 19 bar CH₄ and 1 bar O₂, in a batch reactor containing 50 mL of water and a catalyst concentration of 0.4 g L⁻¹. Each reaction was conducted for 3 hours under illumination from 300 W Xe lamp (λ = 320–780 nm) with a light intensity of 200 mW cm⁻². (b) Cycling tests of MOR over 0.4%Pd/BTO evaluated under identical conditions as (a). (c) MOR product yields and liquid oxygenate selectivities (sel.) for BTO and Pd/BTO with different Pd loadings derived from (a). (d) Yield of products under different catalyst concentrations of 0.4%Pd/BTO. (e) MOR product yields, primary oxygenate selectivities and liquid oxygenate selectivities over different reaction times for 0.4%Pd/BTO with a catalyst loading of 0.1 g L⁻¹. (f) Performance comparison of 0.4%Pd/BTO with reported TiO₂ catalysts under similar experimental conditions.

Notably, the MOR experiments revealed a shift in the liquid oxygenates from CH₃OOH to CH₃OH with increased Pd loading. The CH₃OH selectivity was 98% for 3%Pd/BTO, compared to 52%, 41%, 58%, 67% and 93% for bare BTO, 0.05%Pd/BTO, 0.2%Pd/BTO, 0.4%Pd/BTO and 1%Pd/BTO, respectively (Fig. 2c). Furthermore, control experiments in the absence of light, under similar CH₄/O₂ atmosphere and reaction conditions did not yield any products, confirming that the formation of MOR products is driven by a photocatalytic process (Fig. S12a and b). Control experiments in Ar/O₂ atmosphere (19 bar Ar and 1 bar O₂) did not yield any liquid oxygenates (Fig. S12c).

The catalyst concentration and reaction time of 0.4%Pd/BTO catalyst were further optimized to enhance MOR performance. The MOR product productivity of the catalyst increased threefold, from 1.95 mmol g⁻¹ h⁻¹ to 6.01 mmol g⁻¹ h⁻¹, when the catalyst concentration was reduced from 0.4 g L⁻¹ to 0.1 g L⁻¹, as shown in Fig. 2d. Additionally, the selectivity towards primary oxygenates improved from 87.8% to 92.1%, indicating a reduced extent of over oxidation to CO₂. Moreover, the decrease in catalyst concentration led to a shift in the distribution of liquid oxygenates from CH₃OH to CH₃OOH. At a catalyst concentration of 0.4 g L⁻¹, the selectivities for CH₃OH and CH₃OOH were 67.1% and 32.2%, respectively. When the catalyst concentration was reduced to 0.1 g L⁻¹, the selectivities shifted to 48.8% for CH₃OH and 50.5% for CH₃OOH.

Subsequently, the reaction time was optimized. As shown in Fig. 2e, at an optimal reaction time of 40 min and catalyst concentration of 0.1 g L⁻¹, the 0.4%Pd/BTO achieved an MOR productivity of 7.22 mmol g⁻¹ h⁻¹, along with a high primary oxygenate selectivity of 98.1%. Under these reaction conditions, the selectivities for CH₃OOH and CH₃OH were 72.8% and 26.5%, respectively. Furthermore, the apparent quantum efficiency (AQY) for primary oxygenates was 4.6% under photo-irradiation with a 365 nm LED at a light intensity of 12 mW cm⁻² (Table S2). To better assess the photocatalytic performance of 0.4%Pd/BTO, the findings were compared with those of mixed anatase/rutile and pure anatase TiO₂ catalysts reported in the literature. With a primary oxygenate selectivity of 98.1% and productivity of 7.22 mmol g⁻¹ h⁻¹, 0.4%Pd/BTO demonstrated superior selectivity and productivity for CH₃OOH and CH₃OH, outperforming most TiO₂ catalysts (Fig. 2f and Table S3).

Elucidating the photocatalytic MOR mechanism on Pd/BTO nanorods

Time-dependent MOR experiments. To elucidate the photocatalytic MOR mechanism on Pd/BTO, time-dependent MOR experiments were conducted at 20, 45, 90 and 180 min using BTO samples with low (0.05%Pd/BTO), intermediate (0.4%Pd/BTO) and high (3%Pd/BTO) Pd loadings (Fig. 3a–f). At low Pd loading (0.05%, Fig. 3a and d), the initial productivity and selectivity of liquid oxygenate followed the trend CH₃OOH > CH₃OH. As the reaction progressed, the distribution gradually shifted towards CH₃OH. By 180 min, the selectivities for CH₃OOH and CH₃OH were 57% and 43%, respectively. At intermediate Pd loading (0.4%, Fig. 3b and e), CH₃OH productivity increased rapidly over time, whereas CH₃OOH productivity rose more slowly. Consequently, the distribution of liquid oxygenates shifted more prominently towards CH₃OH, with final selectivities of 32% for CH₃OOH and 67% for CH₃OH at 180 min. At high Pd loading (3%, Fig. 3c and f), the initial selectivities at 20 min were 52% for CH₃OOH and 47% for CH₃OH. Remarkably, as the reaction proceeded, the oxygenate distribution shifted predominantly towards CH₃OH. By 180 min, 3%Pd/BTO exhibited 98% selectivity for CH₃OH. For all Pd loadings, CO₂ was detected only after 90 min or 180 min, suggesting that the over-oxidation of CH₃OOH and/or CH₃OH to CO₂ occurred when the concentrations of these intermediates reached a certain threshold. However, the observed CO₂ levels remained low at 180 min, indicating minimal overoxidation of the products.

Fig. 3 Time-dependent MOR experiments at 20 min, 45 min, 90 min and 180 min for 0.05%Pd/BTO (a and d), 0.4%Pd/BTO (b and e), and 3%Pd/BTO (c and f). Reaction conditions: catalyst loading of 0.4 g L⁻¹, 50 mL of water, 1 bar O₂, 19 bar CH₄ and 300 W Xe lamp (λ = 320–780 nm) at a light intensity of 200 mW cm⁻².

Earlier, TEM analysis revealed that increasing Pd loading on BTO nanorods decreased the dispersion of Pd particles. XPS analysis showed that Pd/BTO samples with higher Pd loading contained a higher fractional composition of metallic Pd⁰ compared to PdO. These observations, along with the results from the time-dependent MOR experiments, suggest that MOR activity and product selectivity are influenced by Pd composition and dispersion on the BTO nanorods.

Potential radical and photochemical pathways for CH₃OOH and CH₃OH formation. To investigate how changes in Pd composition and dispersion on BTO nanorod surface affect MOR activity and product selectivity, we first examine the potential radical and photochemical processes likely involved in the formation of CH₃OOH and CH₃OH (eqn (1)–(11)). Photoexcitation of BTO or Pd/BTO generates photoexcited electrons (e⁻) and holes (h⁺). Methyl radicals (˙CH₃) can be formed either by direct hole oxidation of methane (eqn (1)) or indirectly through the reaction of methane with hydroxyl radicals (˙OH) (eqn (2)). The ˙OH can be produced via a one-hole WOR (eqn (3)) or through an ORR pathway.^52–54 There are two possible routes for ˙OH formation via the ORR pathway.⁵⁴ The first is a direct, three-electron stepwise proton-coupled electron transfer (PCET) that proceeds through surface-bound(*) intermediates: *OOH, *O (via the breaking of O–O bond), and *OH as intermediates (eqn (4)–(6)).^53,54 The second is an indirect route, where ˙OH is generated by the reduction of in situ produced H₂O₂ by an electron (eqn (9)).⁵⁴ H₂O₂ can be produced from a second electron reduction of the *OOH intermediate (eqn (7)) or from a direct two-electron reduction of an oxygen (eqn (8)).⁵³ For CH₃OOH, the coupling of ˙CH₃ with a superoxide radical (˙OOH) forms a CH₃OOH molecule (eqn (10)), with ˙OOH originating from the first electron reduction of oxygen (eqn (4)). For the CH₃OH formation, ˙CH₃ couple with ˙OH to form a CH₃OH molecule (eqn (11)).

˙CH₃ formation:

CH₄ + h⁺ → ˙CH₃ + H⁺ (1)

CH₄ + ˙OH → ˙CH₃ + H⁺ (2)

Water oxidation reaction (WOR):

H₂O + h⁺ → ˙OH + H⁺ (3)

O₂ reduction reaction (ORR):

O₂ + e⁻ + H⁺ → *OOH (or ˙OOH) (4)

*OOH + e⁻ + H⁺ → *O + H₂O (5)

*O+ e⁻ + H⁺ → *OH (or ˙OH) (6)

*OOH + e⁻ + H⁺ → H₂O₂ (7)

O₂ + 2 (e⁻ + H⁺) → H₂O₂ (8)

H₂O₂ + e⁻ + H⁺→ *OH (or ˙OH) + H₂O (9)

CH₃OOH & CH₃OH formation:

˙CH₃ + ˙OOH → CH₃OOH (10)

˙CH₃ + ˙OH → CH₃OH (11)

Investigation of dominant MOR pathways via reactive species scavenging experiments. To elucidate the dominant MOR mechanism, electrons (e⁻), holes (h⁺) and reactive oxygen species (ROS) scavenging experiments were undertaken on the 0.05%, 0.4% and 3%Pd/BTO catalysts. Salicylic acid, Na₂C₂O₄, p-benzoquinone and K₂Cr₂O₇ were adopted as the scavengers of ˙OH, h⁺, ˙OOH and e⁻, respectively.^19,55 We only consider the liquid oxygenates produced (i.e., CH₃OOH, CH₃OH and HCHO) since CO₂ can be a by-product of the scavenger tests. No liquid oxygenates were observed after the addition of salicylic acid, p-benzoquinone, and K₂Cr₂O₇, indicating that ˙OH, ˙OOH and electrons played active roles in the MOR (Fig. 4a and S13a–c). Conversely, the addition of hole scavenger Na₂C₂O₄ had a lesser effect on MOR (Fig. 4b), indicating that photogenerated h⁺ is a less dominant species in the MOR. Therefore, the abstraction of ˙H from CH₄ to form ˙CH₃ is likely occurring indirectly through the reaction of CH₄ with ˙OH (eqn (2)), rather than through the direct oxidation of CH₄ by photogenerated h⁺. Consequently, scavenging ˙OH negatively impacts MOR activity, highlighting the important role of ˙OH. This is reasonable because gaseous CH₄ is hydrophobic, while the TiO₂ surface is hydrophilic, making it easier for water to adsorb onto the TiO₂ surface compared to CH_4. As a result, photogenerated h⁺ is more likely consumed by WOR to generate ˙OH. Photogenerated e⁻ play a crucial role in the formation of ˙OOH, H₂O₂ and ˙OH for MOR through the one-electron, two-electron, and three-electron ORR pathways, respectively. Therefore, MOR cannot occur in the absence of photogenerated electrons.

Fig. 4 (a) Photocatalytic MOR with 0.4%Pd/BTO performed in the absence and presence of salicylic acid, p-benzoquinone and K₂Cr₂O₇, which acted as scavengers for ˙OH, ˙OOH and e⁻, respectively. (b) Photocatalytic MOR with BTO, 0.05%Pd/BTO, 0.4%Pd/BTO, and 3%Pd/BTO, in the absence and presence of Na₂C₂O₄ as a hole scavenger. (c) Photocatalytic MOR performed with 3%Pd/BTO in the presence of ¹⁶O₂ and H₂¹⁸O. The reaction products were analyzed using GC-MS to trace the incorporation of oxygen atoms from both ¹⁶O₂ and H₂¹⁸O into the CH₃OH product. (d) Metallic Pd⁰ and PdO surface concentrations estimated from the XPS Pd 3d spectra for both fresh and used samples.

Tracing the oxygen origin in CH₃OH via H₂¹⁸O isotope labeling and GC-MS. We observed that a high Pd loading, which corresponds to a decreased in Pd dispersion on BTO surface, results in a product selectivity shift from CH₃OOH to CH₃OH. Earlier, we deduced that CH₃OH is produced by the coupling of ˙CH₃ and ˙OH, with ˙OH being generated through three possible pathways: (i) one-hole oxidation of water (eqn (3)), (ii) one-electron reduction of H₂O₂ (eqn (9)) or, (iii) direct three-electron reduction of O₂ (eqn (4)–(6)). However, from the trapping experiment alone, we could not determine the dominant source of ˙OH for CH₃OH formation. Therefore, we conducted isotope experiments using ≤97% ¹⁸O-labeled water (H₂¹⁸O) in the reaction with 3%Pd/BTO to identify the oxygen source of CH₃OH, using gas chromatography-mass spectroscopy (GC-MS). The results show that CH₃¹⁸OH accounted for ∼6% of the produced CH₃OH, while CH₃¹⁶OH was the main product comprising about 94% (Fig. 4c). This suggests that H₂O was not the predominant oxygen source for the generation of ˙OH in CH₃OH formation. Instead, the dominant source of ˙OH originated from the ORR pathway. However, it remained unclear whether ˙OH was primarily sourced from the H₂O₂ pathway via one-electron reduction of H₂O₂ or from the direct three-electron ORR pathway. Therefore, further mechanistic studies were conducted.

Photocatalytic O₂ reduction tests for H₂O₂ formation. From the time-dependent MOR experiments, we observed an accumulation of H₂O₂ over time on BTO with low and intermediate Pd loading (0.05%Pd/BTO and 0.4%Pd/BTO), although the formation of H₂O₂ was low (Tables S4 and S5). In contrast, there was little formation of H₂O₂ when Pd loading was high (1%Pd/BTO and 3%Pd/BTO) (Tables S4 and S5). To investigate, if two-electron reduction of O₂ to H₂O₂ is favorable with Pd/BTO samples, we conducted photocatalytic O₂ reduction tests. Fig. S14 shows the time course of photocatalytic H₂O₂ production over BTO and Pd/BTO in the presence of electron donor (2-propanol). The photocatalytic ORR tests were performed in the presence of electron donor to ensure that oxygen reduction on the Pd cocatalysts was not limited by the rate of water oxidation via photogenerated holes. The results showed that photocatalytic H₂O₂ production decreased with increased Pd particle loading. Therefore, Pd/BTO are unfavorable for the two-electron reduction of O₂ to H₂O₂.

Evaluation of photocatalytic H₂O₂ reduction to ˙OH using coumarin as a photoluminescent probe. Additionally, to investigate the ability of BTO and Pd/BTO to reduce H₂O₂ to ˙OH, we conducted photocatalytic H₂O₂ reduction tests. We used coumarin as a probe for photoluminescent (PL) determination of ˙OH formation rates. Fig. S15 shows the time-dependent PL spectra of the produced 7-hydroxycoumarin (7-HC) for determination of ˙OH formation rates for BTO and Pd/BTO in the presence of H₂O₂. The photocatalytic production of ˙OH from H₂O₂ decreased with increasing Pd particle loading on the BTO nanorods, indicating that Pd nanoparticles are unfavorable for H₂O₂ reduction to ˙OH. Notably, under non-illuminated (dark) conditions, 7-HC was not detected for any of the BTO and Pd/BTO samples, indicating that the production of ˙OH from H₂O₂ is a photocatalytic process rather than a thermal one. Overall, the results showed that Pd nanoparticles are unfavorable for both the two-electron reduction of O₂ to H₂O₂ and the one-electron reduction of H₂O₂ to ˙OH. Thus, the dominant source of ˙OH for CH₃OH formation in Pd/BTO is likely the direct three-electron O₂ reduction pathway.

Post-reaction XPS analysis of catalysts. To further elucidate the MOR mechanism with Pd/BTO, XPS analysis was performed on the recovered Pd/BTO catalysts after the reaction. No obvious changes in the Ti 2p XPS spectra were observed, suggesting that BTO was not reduced during the MOR (Fig. S16). In addition, the O 1s spectrum showed no significant changes in either the lattice oxygen (O_lat) peak at 529.9 eV or the oxygen vacancy peak at 531.7 eV, further indicating the stability of the BTO under reaction conditions (Fig. S17). However, the Pd 3d spectra of the Pd/BTO showed that the surface concentration of metallic Pd⁰ increased in the Pd/BTO samples after the MOR (Fig. S18 and 4d). The surface concentration of metallic Pd⁰ in the recovered samples was 61.3% for 0.05%Pd/BTO; 57.8% for 0.4%Pd/BTO; and 90.9% for 3%Pd/BTO, respectively (Fig. 4d). The results indicated that the metallic Pd⁰ content in the Pd nanoparticles increased by approximately 31.7% and 24.2% during the MOR with 0.05%Pd/BTO and 3%Pd/BTO catalysts, respectively. This suggests significant electron trapping occurred in the Pd nanoparticles of both catalysts, leading to the in situ reduction of Pd nanoparticles during the MOR. However, the reduction of Pd nanoparticles in 0.4%Pd/BTO was less significant, the metallic Pd⁰ content only increased by 2.9% during MOR. The low electron trapping in 0.4%Pd/BTO suggests that photogenerated electrons were consumed more readily by adsorbed O₂ in this catalyst compared to 0.05%Pd/BTO and 3%Pd/BTO, resulting in the highest MOR productivity observed for 0.4%Pd/BTO.

Probing photogenerated electron dynamics via in situ time-resolved transient absorption spectroscopy. To elucidate the behaviors of photogenerated electrons in BTO, 0.05%Pd/BTO, 0.4%Pd/BTO and 3%Pd/BTO under light irradiation, in situ time-resolved transient absorption (TA) spectroscopy in the visible to mid-infrared (IR) region was performed. Fig. 5a shows the TA spectrum of BTO after UV-laser pulse irradiation. Bandgap photoexcitation resulted in broad absorption peaks at ∼24 000 cm⁻¹ and ∼12 000 cm⁻¹.^36,56 The peak at ∼24 000 cm⁻¹ is attributed to photogenerated holes, while the peak at ∼12 000 cm⁻¹ corresponds to trapped electrons.^36,56 The effects of Pd dispersion on trapped electrons were further investigated by monitoring the decay of TA signal at 12 000 cm⁻¹.

Fig. 5 (a) In situ time-resolved transient absorption (TA) spectrum of BTO after UV-laser pulse irradiation. The TA spectrum shows broad absorption peaks at approximately 24 000 cm⁻¹ and 12 000 cm⁻¹, corresponding to photogenerated holes and electrons, respectively. (b) TA spectra of BTO and Pd/BTO after UV-laser pulse irradiation measured at 12 000 cm⁻¹ in vacuum. Among the Pd/BTO samples, 3%Pd/BTO shows the fastest electron trapping (∼1 μs), with a sharp increase in trapped electrons within 0.5 μs, followed by the quickest decay rate. TA spectra of (c) BTO, (d) 0.05%Pd/BTO, (e) 0.4%Pd/BTO, and (f) 3%Pd/BTO after UV-laser pulse irradiation in the presence of reactant molecules (O₂ and H₂O), with N₂ as a reference. The TA spectra of 0.05%Pd/BTO and 3%Pd/BTO showed less significant reductions in TA upon exposure to O₂, suggesting that O₂ adsorption is less efficient, and electron trapping occurs in the Pd cocatalysts of these samples. TA measurements under H₂O vapor show that bare BTO exhibits limited hole oxidation, while Pd-loaded BTO shows a significant increase in TA intensity, indicating enhanced hole oxidation and improved charge separation due to efficient electron capture by Pd cocatalysts.

Fig. 5b shows the TA spectra of BTO and Pd/BTO measured at 12 000 cm⁻¹ in vacuum. In the presence of Pd nanoparticles, the intensities of the TA signals from trapped electrons are lower than those of BTO, indicating that the population of trapped electrons in Pd-loaded BTO is reduced due to electron capture by the Pd cocatalysts. Moreover, the loading of Pd cocatalysts also led to accelerated electron decay. The delay in detecting trapped electrons in Pd-loaded BTO suggests rapid electron transfer from BTO to Pd (within a few μs). Among the Pd-loaded samples, 3%Pd/BTO exhibited the fastest electron trapping (∼1 μs), with a sharp increase in trapped electrons within 0.5 μs, followed by the fastest decay rate.

TA spectra of BTO and Pd/BTO were also investigated in the presence and absence of reactant molecules (O₂ or H₂O) (Fig. 5c–f), with the TA measurement in N₂, used as a reference. For BTO, the intensity of TA decreased upon exposure to O₂, and electron decay occurred with a fast rate, indicating that adsorbed O₂ on BTO surface rapidly consumed electrons and accelerated electron decay. For 0.4%Pd/BTO, the TA spectrum also showed reduced intensity and accelerated electron decay, indicating that electrons were readily consumed by O₂ on its Pd cocatalysts. However, the TA spectra of 0.05%Pd/BTO and 3%Pd/BTO showed less significant reductions in TA upon exposure to O₂, suggesting that O₂ adsorption is less efficient, and electron trapping occurs in the Pd cocatalysts of these samples.

On the other hand, TA measurements with photocatalysts in the presence of H₂O vapor reflect hole oxidation kinetics, as the consumption of photogenerated holes by H₂O vapor prevents electron recombination and extends electron lifetime.⁵⁶ For bare BTO, a relatively low enhancement in TA intensity was observed, indicating that charge separation efficiency in BTO is limited by poor hole oxidation kinetics. In contrast, for the Pd-loaded BTO samples, a significant increase in TA intensity was observed upon exposure to H₂O vapor, suggesting that efficient electron capture by Pd cocatalysts simultaneously boosts hole oxidation kinetics.

In addition, we investigated the population of surviving electrons at a specific delay time – 10 μs after photoexcitation – as an indicator of the number of electrons that successfully evade recombination and remain available for redox reactions. To assess the reactivity of electrons and holes, we compared the relative electron populations under different atmospheric conditions: N₂ (inert), O₂ (electron scavenger), and H₂O vapor (hole scavenger). The results are summarized in Tables S6 and S7. Under an O₂ atmosphere, 0.05%Pd/BTO and 3%Pd/BTO exhibited higher relative electron population than BTO and 0.4%Pd/BTO, which is attributed to increased electron trapping. In the presence of H₂O vapor, BTO catalysts with Pd cocatalysts exhibited significantly higher relative electron populations compared to bare BTO, indicating enhanced hole scavenging and improved charge separation dynamics due to the presence of Pd cocatalysts. Notably, 0.05%Pd/BTO and 3%Pd/BTO exhibited the highest relative electron population because electron trapping increase the lifetime of photogenerated electrons. Overall, the results from in situ time-resolved TA spectroscopy are in good agreement with the XPS findings, indicating more efficient photogenerated electron transfer between 0.4%Pd/BTO and O₂, while significant electron trapping occurred in 0.05%Pd/BTO and 3%Pd/BTO.

Probing MOR mechanism on Pd⁰/BTO and PdO catalysts using DFT calculations. To further elucidate the O₂ reduction reaction and the mechanism of CH₃OOH and CH₃OH formation under different Pd loadings, DFT calculations were performed. Experimental results indicate that at low and moderate loadings, Pd exists as a mixture of PdO and Pd⁰ nanoparticles, whereas at high loadings, Pd predominantly forms Pd⁰ nanoparticles. Based on these findings, we constructed Pd₈/BTO (210) (hereafter, denoted as Pd⁰/BTO) and PdO(110) catalysts (Fig. S19–S25). The Pd⁰/BTO catalyst shows binding energy and adhesion energy of −6.30 eV and −1.26 eV, respectively, demonstrating there is a strong binding energy and a strong adhesion between the Pd₈ cluster and BTO (210) surface, the Pd₈ cluster can adhere stably on the BTO (210) surface. Fig. 6a and b show the free energy diagram for the four- and two-electron oxygen reduction on Pd⁰/BTO and PdO, respectively. Both Pd⁰/BTO and PdO catalysts demonstrate O₂ adsorption in a bridge configuration across two adjacent Pd atoms, with adsorption free energies (ΔG_ads) of −1.08 eV and −1.16 eV. This suggests that both metallic Pd⁰/BTO and PdO exhibited strong O₂ adsorption ability. In addition, *OOH can be formed from a PCET step with *O₂. The generated *OOH can continue to react via three potential pathways. In the first pathway, in an endothermic reaction, *OOH desorb to form a ˙OOH (ΔG = 1.25 eV for Pd/BTO and 1.67 eV for PdO; see Fig. 6c). In the second or third pathway, *OOH undergoes a second PCET to form either a *H₂O₂ or *O. For both Pd⁰/BTO and PdO, the formation of H₂O₂ is also endothermic (ΔG = 0.12 eV for Pd/BTO and 0.86 eV for PdO). The formation of *O (through O–O bond breaking) in contrast is strongly exothermic (ΔG = −1.55 eV for Pd⁰/BTO and −2.09 eV for PdO). Further PCET with *O to form *OH also proceeds exothermically (ΔG = −1.23 eV for Pd⁰/BTO and −1.38 eV for PdO). However, the PCET with *OH to H₂O happens exothermically on Pd⁰/BTO (ΔG = −0.80 eV) but endothermically on PdO (ΔG = 0.31 eV).

Fig. 6 Free energy diagrams for the four- and two-electron oxygen reduction reaction on (a) Pd⁰/BTO and (b) PdO. Gibbs free energy barriers for the key elementary steps involved in the ORR, specifically the formation of *OOH and *OH intermediates on (c) Pd⁰/BTO and (d) PdO, respectively. The reaction pathways for the coupling *CH₃ or ˙CH₃ and *OOH or ˙OOH coupling to form CH₃OOH on Pd⁰/BTO and PdO are shown in (e). In (f), the reaction pathways for the coupling of *CH₃ or ˙CH₃ and *OH or ˙OH coupling to form CH₃OH on Pd⁰/BTO and PdO are presented.

We also calculated the Gibbs free energy barriers for *OOH and *OH formation on Pd⁰/BTO and PdO, as shown in Fig. 6c and d. The results show that the hydrogenation of *O₂ to form *OOH is more favorable on the PdO catalyst (G_a = 0.70 eV) compared to the Pd⁰/BTO catalyst (G_a = 0.90 eV). Conversely, the hydrogenation of *O to form *OH is more favorable on the Pd⁰/BTO catalyst (G_a = 0.70 eV) than on PdO (G_a = 1.07 eV). These findings suggest that metallic Pd⁰ is more effective than oxidized PdO in facilitating the three-electron reduction of O₂ to *OH while one-electron reduction of O₂ to *OOH is more easily achieved on PdO.

In the experiments, we observed CH₄ can be activated by ˙OH to form ˙CH₃ radicals. Therefore, we further explored via DFT calculations, the reaction pathways for ˙CH₃/*CH₃ and ˙OH/*OH coupling to form CH₃OH, as well as the possible pathway for ˙CH₃/*CH₃ and ˙OOH/*OOH coupling to form CH₃OOH. As shown in Fig. 6e and f, when the Pd⁰ or PdO surface contains *OOH or *OH intermediates, the free ˙CH₃ radical readily adsorbs onto the Pd site forming *CH₃ on the catalyst surface in an exothermic reaction, with *CH₃ being more easily adsorbed on Pd⁰/BTO than on PdO. The simultaneous adsorption of *CH₃ and *OH on Pd sites induces a coupling reaction forming CH₃OH, which is endothermic on Pd⁰/BTO but even more so on PdO (Fig. 6f). This suggests that CH₃OH formation is more favorable on Pd⁰/BTO. On the other hand, the coupling reaction between *CH₃ and *OOH on PdO is less endothermic than on Pd⁰/BTO indicating that CH₃OOH formation is more favorable on PdO (Fig. 6e). The difference in the free energy required for CH₃OOH and CH₃OH formation on Pd⁰/BTO and PdO supports the experimental observations. In particular, 3%Pd/BTO, which exhibited a high surface concentration of metallic Pd⁰ due to in situ Pd reduction during MOR, predominantly produced CH₃OH. In contrast, 0.05%Pd/BTO and 0.4%Pd/BTO, which contained a mixture of Pd⁰ and PdO on their surface, produced both CH₃OH and CH₃OOH.

Photocatalytic oxygen reduction evaluation for ˙OH and ˙OOH generation using coumarin and nitroblue tetrazolium chloride assays. To further confirm the roles of Pd⁰ and PdO in photocatalytic ORR, coumarin and nitroblue tetrazolium chloride (NBT) tests were conducted to detect the presence of ˙OH and ˙OOH radicals, respectively, on BTO and Pd/BTO samples. Fig. 7a shows the time-dependent PL spectra of 7-HC under 365 nm LED photo-irradiation in O₂-saturated aqueous solutions. The PL intensity of 7-HC increases with higher Pd loading on the BTO nanorods, suggesting that Pd⁰-rich samples effectively promote ˙OH radicals generation. These results imply that ORR pathways involving the direct three-electron reduction of O₂ to ˙OH are more active on Pd⁰-rich surfaces. In contrast, the NBT test (Fig. 7b), which detects ˙OOH radicals, shows that the highest photocatalytic production of ˙OOH radicals occurs on the 0.05%Pd/BTO samples, followed by 0.4%, and then 3%Pd/BTO. This trend implies that ˙OOH/*OOH generation via single-electron reduction of O₂ is more favorable on PdO-rich samples. These findings support the proposed mechanism that Pd⁰ facilitates a direct three-electron reduction of O₂, leading to the formation of ˙OH/*OH radicals, which can then react with ˙CH₃/*CH₃ radicals to form CH₃OH. PdO, in contrast, PdO promotes a one-electron O₂ reduction pathway, resulting in ˙OOH/*OOH that couple with ˙CH₃/*CH₃ to form CH₃OOH.

Fig. 7 (a) Time-dependent PL spectra of the produced 7-hydroxycoumarin (7-HC) to determine ˙OH formation rates for BTO and Pd/BTO in O₂-saturated aqueous solutions. (b) Kinetic constants of NBT photodegradation for ˙OOH radical detection using various Pd/BTO catalysts in O₂-saturated aqueous solutions.

Bader charge analysis to elucidate the favorability of ˙OH radical generation. We conducted Bader charge analysis to elucidate why ˙OH radical generation is more favorable on Pd⁰ than on PdO. The results show that surface Pd atoms on PdO carry substantial positive charges, which enhance the adsorption of *OH species and make their desorption more difficult (Fig. S26a). In contrast, the surface Pd atoms on Pd⁰/BTO are in metallic (zero-valent) state, leading to weaker *OH binding and easier desorption (Fig. S26b). This facilitates the release of ˙OH radicals, making Pd⁰/BTO more favorable for ˙OH radical generation.

In situ diffuse reflectance infrared fourier transform spectroscopy for elucidating the MOR mechanism. The MOR mechanism was further investigated using in situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) under both dark and light irradiation conditions. The measurements were conducted using a water-vapor-saturated gas mixture of CH₄ and O₂ (CH₄/O₂ = 19 : 1). Fig. 8a–d presents the in situ DRIFTS spectra of the BTO and Pd/BTO samples obtained under these conditions. The peaks at 3014 cm⁻¹ and 1297 cm⁻¹ are attributed to the C–H deformation vibration of CH₄, while the peak at 1442 cm⁻¹, assigned to adsorbed *CH₃ indicates the dissociation of CH₄.⁴⁶ The intensities of these peaks increased with prolonged light irradiation time, suggesting the progressive activation of CH₄. Similarly, characteristic peaks corresponding to CH₃O* species were observed at 2943 cm⁻¹, 2840 cm⁻¹, and 1069 cm⁻¹,⁵⁷ confirming the formation of CH₃OH. Moreover, the peak at 3690 cm⁻¹ associated with adsorbed *OH, the key reaction intermediate, were more intense on 0.4%Pd/BTO and 3%Pd/BTO compared to 0.05%Pd/BTO, indicating a higher coverage of *OH on 0.4%Pd/BTO and 3%Pd/BTO, which aids the formation of CH₃OH.⁵⁸

Fig. 8 In situ DRIFTS spectra of (a) BTO, (b) 0.05%Pd/BTO, (c) 0.4%Pd/BTO, and (d) 3%Pd/BTO photocatalysts using a water-vapor-saturated gas mixture of CH₄ and O₂ (CH₄/O₂ = 19 : 1) under both dark and light irradiation conditions.

On the other hand, peaks at 1365 cm⁻¹ and 1336 cm⁻¹, attributed to the vibrations of adsorbed CH₃OOH species were observed across all samples.⁵⁹ Notably, 0.05%Pd/BTO exhibited the highest intensity for these peaks, reflecting more favorable formation of CH₃OOH on its surfaces, consistent with its higher CH₃OOH selectivity in MOR. Additionally, peaks at 1585 cm⁻¹ and 1539 cm⁻¹ can be assigned to *HCOO species, indicative of overoxidation byproducts.⁶⁰ The results are consistent with the proposed mechanism that Pd⁰-rich BTO catalyst facilitates a direct three-electron reduction of O₂, leading to the formation of *OH species, which can then react with *CH₃ to produce CH₃OH. In contrast, PdO-rich catalyst, favors a one-electron O₂ reduction pathway, generating *OOH species that couple with *CH₃ to form CH₃OOH. This mechanistic divergence accounts for the different product distributions observed among the catalysts.

Proposed MOR mechanisms for Pd/BTO. Building on the relationship between charge carrier behavior and oxygen adsorption and reduction, we examine the effects of Pd dispersion and composition on BTO nanorods for methane oxidation. As shown in Scheme 1, for Pd/BTO with high Pd loading (e.g., 3%Pd/BTO), the low dispersion of Pd particles on the BTO nanorod surface results in significant electron trapping at the Pd cocatalysts, as indicated by the TA spectroscopy. This leads to substantial in situ reduction of the PdO, as shown by the XPS. Notably, after the MOR, the metallic Pd⁰ composition in 3%Pd/BTO is remarkably high at 90.9%, suggesting that the high surface concentration of metallic Pd⁰ promotes selective CH₃OH formation. In contrast, for Pd/BTO with low Pd loading (e.g., 0.05%Pd/BTO), the Pd nanoparticles are very small with high dispersion, leading to significant oxidation upon exposure to ambient conditions and a high PdO surface concentration (60.4%). Consequently, some of the photoexcited electrons are consumed in the in situ reduction of the PdO to Pd⁰ during the MOR, as indicated by the XPS.

Scheme 1 Pd dispersion and composition on BTO nanorods regulate the activity and selectivity of photocatalytic MOR products.

For 0.4%Pd/BTO, with an optimal Pd particle loading and dispersion, electron transfer to O₂ is more efficient, yielding the highest MOR product output with minimal in situ reduction of PdO, as supported by the XPS and TA spectroscopy. However, due to the co-existence of both Pd⁰ and PdO active sites in 0.05%Pd/BTO and 0.4%Pd/BTO, these catalysts produce both CH₃OOH and CH₃OH. DFT calculations and the experimental observations indicate that Pd⁰ facilitates the direct three-electron reduction of O₂ to form *OH or ˙OH, which then reacts with *CH₃ or ˙CH₃ to produce CH₃OH. In contrast, PdO favors one-electron reduction of oxygen and the formation of CH₃OOH by coupling of *CH₃ or ˙CH₃ with *OOH or ˙OOH.

Conclusion

In conclusion, the study highlights the crucial role of Pd composition and dispersion in enhancing the photocatalytic oxidation of methane to primary oxygenates – CH₃OH and CH₃OOH over BTO nanorods. Pd/BTO nanorods with higher Pd loading and lower Pd dispersion, enhanced electron trapping in the Pd⁰/PdO nanoparticles and promoted the self-reduction of PdO to metallic Pd⁰ upon light irradiation. The higher surface concentration of Pd⁰ resulted in selective CH₃OH production. Conversely, BTO with optimum Pd loading and dispersion promoted electron transfer to O₂; however, the presence of both Pd⁰ and PdO, led to the formation of both CH₃OH and CH₃OOH. DFT calculations together with experimental findings show that Pd⁰ enables the direct three-electron reduction of O₂ to form *OH or ˙OH, which then reacts with *CH₃ or ˙CH₃ to produce CH₃OH. In contrast, PdO favors one-electron reduction of oxygen and the formation of CH₃OOH by coupling of *CH₃ or ˙CH₃ with *OOH or ˙OOH. The distinct reaction pathways provide valuable insights into optimizing Pd-based photocatalysts for selective methane oxidation, highlighting the importance of Pd composition and dispersion in influencing charge carrier dynamics and determining the selectivity of products in methane oxidation.

Author contributions

Jiawei Wu: investigation, formal analysis, methodology, validation, visualization. Hu Ding: methodology, formal analysis, validation, visualization. Yan Wang: formal analysis, validation. Ulfi Muliane: investigation, methodology, validation. Shuji Anabuki: investigation, methodology, validation. Naoya Murakami: validation. Xinli Zhu: resources, methodology, validation. Akira Yamakata: resources, methodology, validation. Teruhisa Ohno: validation. Shi Nee Lou: conceptualization, methodology, validation, funding acquisition, data curation, project administration, resources, supervision, writing – original draft, writing – review & editing.

Conflicts of interest

There are no conflicts to declare.

Data availability

The authors confirm that the data supporting the findings of this study are available within the article and its SI. Supplementary information is available. See DOI: https://doi.org/10.1039/d5ta04478b.

Acknowledgements

This work was supported by National Nature Science Foundation of China (grant no. 22250410262), Tianjin Natural Science and Technology Project (grant no. 22JCYBJC01410) and National Key Research and Development Program of China (grant no. 2022YFB4101702).

References

1. M. Ravi, M. Ranocchiari and J. A. van Bokhoven, Angew. Chem., Int. Ed., 2017, 56, 16464–16483.
2. P. Schwach, X. Pan and X. Bao, Chem. Rev., 2017, 117, 8497–8520.
3. A. I. Olivos-Suarez, À. Szécsényi, E. J. M. Hensen, J. Ruiz-Martinez, E. A. Pidko and J. Gascon, ACS Catal., 2016, 6, 2965–2981.
4. P. Tang, Q. Zhu, Z. Wu and D. Ma, Energy Environ. Sci., 2014, 7, 2580–2591.
5. N. Agarwal, S. J. Freakley, R. U. McVicker, S. M. Althahban, N. Dimitratos, Q. He, D. J. Morgan, R. L. Jenkins, D. J. Willock, S. H. Taylor, C. J. Kiely and G. J. Hutchings, Science, 2017, 358, 223–227.
6. H. Ceri, M. F. Michael, H. A. R. Mohd, T. Adam, H. Qian, D. Nikolaos, K. Joo and H. Henk, Angew. Chem., Int. Ed., 2012, 51, 5129–5133.
7. H. Song, X. Meng, S. Wang, W. Zhou, S. Song, T. Kako and J. Ye, ACS Catal., 2020, 10, 14318–14326.
8. Y. Jiang, W. Zhao, S. Li, S. Wang, Y. Fan, F. Wang, X. Qiu, Y. Zhu, Y. Zhang, C. Long and Z. Tang, J. Am. Chem. Soc., 2022, 144, 15977–15987.
9. M. Huang, S. Zhang, B. Wu, X. Yu, Y. Gan, T. Lin, F. Yu, Y. Sun and L. Zhong, ChemSusChem, 2022, 15, e202200548.
10. M. Huang, S. Zhang, B. Wu, Y. Wei, X. Yu, Y. Gan, T. Lin, F. Yu, F. Sun, Z. Jiang and L. Zhong, ACS Catal., 2022, 12, 9515–9525.
11. H. Song, H. Huang, X. Meng, Q. Wang, H. Hu, S. Wang, H. Zhang, W. Jewasuwan, N. Fukata, N. Feng and J. Ye, Angew. Chem., Int. Ed., 2023, 62, e202215057.
12. C. Xie, E. Sun, G. Wan, J. Zheng, R. Gupta and A. Majumdar, Nano Lett., 2023, 23, 2039–2045.
13. R. Zhang, J. Shi, L. Fu, Y.-G. Liu, Y. Jia, Z. Han, K. Yuan and H.-Y. Jiang, ACS Nano, 2024, 18, 12994–13005.
14. Y. Li, K. Sun, S. Ning, P. Qiao, S. Wang, Z.-j. Wang, L. Zhu, X. Zhang, K. Peng, X.-s. Wang, D. Wang, L. Liu, H. Song and J. Ye, J. Catal., 2024, 440, 115840.
15. N. Feng, H. Lin, H. Song, L. Yang, D. Tang, F. Deng and J. Ye, Nat. Commun., 2021, 12, 4652.
16. P.-P. Luo, X.-K. Zhou, Y. Li and T.-B. Lu, ACS Appl. Mater. Interfaces, 2022, 14, 21069–21078.
17. X.-K. Zhou, Y. Li, P.-P. Luo and T.-B. Lu, ACS Appl. Mater. Interfaces, 2023, 15, 36280–36288.
18. X. Zhang, Y. Wang, K. Chang, S. Yang, H. Liu, Q. Chen, Z. Xie and Q. Kuang, Appl. Catal., B, 2023, 320, 121961.
19. K. Chang, X. Zhang, Y. Hua, Z. Wang, Z. Song, Y. Wang, Q. Chen, W. Yan, Q. Kuang and Z. Xie, ACS Appl. Nano Mater., 2024, 7, 21453–21462.
20. F. Si, M. Lv, X. Cai, Y. Li, M. Zhao, T. Hou and Y. Li, Appl. Catal., B, 2024, 358, 124417.
21. F. Si, G. Wang, M. Lv, J. Bai, Y. Li, T. Hou and Y. Li, Appl. Catal., B, 2025, 365, 124934.
22. L. Yan, S. Wang, C. Qian and S. Zhou, J. Mater. Chem. A, 2025, 13, 4662–4672.
23. H. Song, X. Meng, S. Wang, W. Zhou, X. Wang, T. Kako and J. Ye, J. Am. Chem. Soc., 2019, 141, 20507–20515.
24. W. Zhou, X. Qiu, Y. Jiang, Y. Fan, S. Wei, D. Han, L. Niu and Z. Tang, J. Mater. Chem. A, 2020, 8, 13277–13284.
25. W. Zhu, M. Shen, G. Fan, A. Yang, J. R. Meyer, Y. Ou, B. Yin, J. Fortner, M. Foston, Z. Li, Z. Zou and B. Sadtler, ACS Appl. Nano Mater., 2018, 1, 6683–6691.
26. Y. Fan, W. Zhou, X. Qiu, H. Li, Y. Jiang, Z. Sun, D. Han, L. Niu and Z. Tang, Nat. Sustain., 2021, 4, 509–515.
27. K. Wang, L. Luo, C. Wang and J. Tang, Chin. J. Catal., 2023, 46, 103–112.
28. Y. Fan, Y. Jiang, H. Lin, J. Li, Y. Xie, A. Chen, S. Li, D. Han, L. Niu and Z. Tang, Nat. Commun., 2024, 15, 4679.
29. L. Luo, L. Fu, H. Liu, Y. Xu, J. Xing, C.-R. Chang, D.-Y. Yang and J. Tang, Nat. Commun., 2022, 13, 2930.
30. Y. Jiang, S. Li, S. Wang, Y. Zhang, C. Long, J. Xie, X. Fan, W. Zhao, P. Xu, Y. Fan, C. Cui and Z. Tang, J. Am. Chem. Soc., 2023, 145, 2698–2707.
31. K.-i. Katsumata, Y. Ohno, K. Tomita, T. Taniguchi, N. Matsushita and K. Okada, ACS Appl. Mater. Interfaces, 2012, 4, 4846–4852.
32. H. Lin, L. Li, M. Zhao, X. Huang, X. Chen, G. Li and R. Yu, J. Am. Chem. Soc., 2012, 134, 8328–8331.
33. T. Ohno, T. Higo, N. Murakami, H. Saito, Q. Zhang, Y. Yang and T. Tsubota, Appl. Catal., B, 2014, 152–153, 309–316.
34. T. Ohno, T. Higo, H. Saito, S. Yuajn, Z. Jin, Y. Yang and T. Tsubota, J. Mol. Catal. A: Chem., 2015, 396, 261–267.
35. A. Naldoni, T. Montini, F. Malara, M. M. Mróz, A. Beltram, T. Virgili, C. L. Boldrini, M. Marelli, I. Romero-Ocaña, J. J. Delgado, V. Dal Santo and P. Fornasiero, ACS Catal., 2017, 7, 1270–1278.
36. J. J. M. Vequizo, H. Matsunaga, T. Ishiku, S. Kamimura, T. Ohno and A. Yamakata, ACS Catal., 2017, 7, 2644–2651.
37. Y. Ma, K. Kobayashi, Y. Cao and T. Ohno, Appl. Catal., B, 2019, 245, 681–690.
38. Y. Cao, S. N. Lou, S. Wang, H. Yang, Q. Zhang, C. Wang, N. Murakami and T. Ohno, Appl. Catal., A, 2022, 634, 118539.
39. M. Kamran, T. A. Kandiel, S. Abdel-Azeim, M. A. Morsy and D. W. Bahnemann, J. Phys. Chem. C, 2023, 127, 7707–7717.
40. R. Li, F. Zhang, D. Wang, J. Yang, M. Li, J. Zhu, X. Zhou, H. Han and C. Li, Nat. Commun., 2013, 4, 1432.
41. K. Fuku, R. Takioka, K. Iwamura, M. Todoroki, K. Sayama and N. Ikenaga, Appl. Catal., B, 2020, 272, 119003.
42. P. D. Cozzoli, R. Comparelli, E. Fanizza, M. L. Curri, A. Agostiano and D. Laub, J. Am. Chem. Soc., 2004, 126, 3868–3879.
43. J. Ohyama, A. Yamamoto, K. Teramura, T. Shishido and T. Tanaka, ACS Catal., 2011, 1, 187–192.
44. Z. H. N. Al-Azri, M. AlOufi, A. Chan, G. I. N. Waterhouse and H. Idriss, ACS Catal., 2019, 9, 3946–3958.
45. J. Yang, H. Zhou, J. Luo and M. Wang, ACS Catal., 2025, 15, 1663–1671.
46. W. Zhang, D. Xi, Y. Chen, A. Chen, Y. Jiang, H. Liu, Z. Zhou, H. Zhang, Z. Liu, R. Long and Y. Xiong, Nat. Commun., 2023, 14, 3047.
47. O. Tomita, T. Otsubo, M. Higashi, B. Ohtani and R. Abe, ACS Catal., 2016, 6, 1134–1144.
48. X. Wu, Q. Zhang, W. Li, B. Qiao, D. Ma and S. L. Wang, ACS Catal., 2021, 11, 14038–14046.
49. T. Liu, Z. Pan, J. J. M. Vequizo, K. Kato, B. Wu, A. Yamakata, K. Katayama, B. Chen, C. Chu and K. Domen, Nat. Commun., 2022, 13, 1034.
50. H. Zhang, P. Sun, X. Fei, X. Wu, Z. Huang, W. Zhong, Q. Gong, Y. Zheng, Q. Zhang, S. Xie, G. Fu and Y. Wang, Nat. Commun., 2024, 15, 4453.
51. Y. Yu, T. He, L. Guo, Y. Yang, L. Guo, Y. Tang and Y. Cao, Sci. Rep., 2015, 5, 9561.
52. Y. Nosaka and A. Y. Nosaka, Chem. Rev., 2017, 117, 11302–11336.
53. A. Kulkarni, S. Siahrostami, A. Patel and J. K. Nørskov, Chem. Rev., 2018, 118, 2302–2312.
54. S. N. Lou, J. Lim, T. H. Jeon and W. Choi, ACS ES&T Eng., 2022, 2, 1116–1129.
55. Q. Zhou, X. Tan, X. Wang, Q. Zhang, C. Qi, H. Yang, Z. He, T. Xing, M. Wang, M. Wu and W. Wu, ACS Catal., 2024, 14, 955–964.
56. A. Yamakata, M. Kawaguchi, N. Nishimura, T. Minegishi, J. Kubota and K. Domen, J. Phys. Chem. C, 2014, 118, 23897–23906.
57. J. Wang, G. Li, Z. Li, C. Tang, Z. Feng, H. An, H. Liu, T. Liu and C. Li, Sci. Adv., 2017, 3, e1701290.
58. H. Gong, L. Zhang, C. Deng, M. Liu, X. Liu, Y. Huang, K. Zhou, P. He, J. Li, Y. Yang, L. Wang, Q. Yang, Z. Bao, Q. Ren, T. Tan, S. Yao and Z. Zhang, J. Am. Chem. Soc., 2025, 147, 9134–9146.
59. Y. Xu, C. Wang, X. Li, L. Xiong, T. Zhang, L. Zhang, Q. Zhang, L. Gu, Y. Lan and J. Tang, Nat. Sustain., 2024, 7, 1171–1181.
60. Z. Xiao, J. Shen, J. Jiang, J. Zhang, S. Liang, S. Han, J. Long, W. Dai, Y. Li, X. Wang, H. Xi, X. Fu and Z. Zhang, Adv. Funct. Mater., 2025, 35, 2422726.

---

Footnote

† These authors contributed equally to this work.

---