Synergy of Single Atoms Pd and Oxygen Vacancies on In2O3 for Highly Selective C1 Oxygenates Production from Methane under Visible Light

Methane (CH 4 ) oxidation to high value chemicals under mild conditions through photocatalysis is a sustainable and appealing pathway, nevertheless confronting the critical issues on both conversion and selectivity. Herein, under visible irradiation (420 nm), the synergy of palladium (Pd) atom cocatalyst and oxygen vacancies (OVs) on In 2 O 3 nanorods enabled superior photocatalytic CH 4 activation by O 2 . The optimised catalyst reached ca. 100 µmol·h − 1 of C1 oxygenates, with a selectivity of primary products (CH 3 OH and CH 3 OOH) up to 82.5 %. Mechanism investigation elucidated that such superior photocatalysis was induced by the dedicated function of Pd single atoms and oxygen vacancies on boosting hole and electron transfer pathway, respectively. O 2 was proven to be the only oxygen source for CH 3 OH production, while H 2 O acted as the promoter for ecient CH 4 activation through ·OH production and facilitated product desorption as indicated by DFT modelling. This work thus provides new understandings on simultaneous regulation of activity and selectivity by the signicant synergy of single atom cocatalysts and oxygen vacancies.


Introduction
As the predominant constituent of natural gas, methane hydrate and shale gas resources, selective methane (CH 4 ) oxidation to value-added chemicals holds considerable nancial and environmental prospective [1][2][3][4][5]. However, the inert symmetrical tetrahedral structure of CH 4 makes it rather di cult for the dissociation of the rst C-H bond, which is the most important step for activation of methane. [6][7][8] Industrial multistep route via steam reforming and subsequent Fischer-Tropsch synthesis could e ciently activate CH 4 , while it requires harsh experimental conditions (eg. > 700 o C temperature and/or high pressure), causing huge energy-consumption and safety issues [9][10][11][12][13]. In parallel, it is relatively di cult to achieve high selectivity due to the more reactive characteristics of the desired oxygenates against both the reactant CH 4 and stable product CO 2 . [14][15][16][17] Therefore, selective CH 4 conversion to value-added chemicals under mild conditions other than CO 2 is highly attractive, while confronting considerable challenges.
Photocatalysis offers an appealing alternative to drive many tough redox reactions under mild conditions including CO 2 conversion [18,19], N 2 reduction [20] and selective CH 4 oxidation [8]. Recently, various value-added chemicals such as methanol [1,[21][22][23], formaldehyde [24,25], ethanol [26,27], ethane and ethylene [28][29][30][31][32][33] were produced by photocatalysis. For example, we found that up to 90 % selectivity with a yield of 3.5 µmol·h − 1 methanol could be achieved over the optimized FeO x /TiO 2 photocatalyst under ambient condition using H 2 O 2 as an oxidant [22]. Recently a high yield of liquid oxygenates including CH 3 OH, CH 3 OOH and HCHO were produced under full arc irradiation over Au supported ZnO, together with the good selectivity of primary products (CH 3 OH and CH 3 OOH) (< 70 %) [1]. Very recent, the yields of 18.7 µmol·h − 1 HCHO and 3.7 µmol·h − 1 CH 3 OH were reported on quantum BiVO 4 with an excellent selectivity toward HCHO (87 %) and CH 3 OH (99 %) under 300-400 nm or 400-780nm irradiation [25]. Given these signi cant advances in photocatalytic methane conversion, the yield and/or selectivity to high value chemicals are still quite moderate, in particular it is very challenging to achieve methane activation under visible light irradiation instead of a full arc spectrum due to a narrowed bandgap with mitigated reduction or oxidation potentials To realize visible driven methane oxidation by O 2 gas on narrow bandgap photocatalysts, cocatalyst is crucial that does not only promote charge separation, more importantly manipulates the activation energy of the methane conversion and the selectivity [34][35][36][37][38][39]. Furthermore rationally regulating the production of reactive oxygen species (ROS) through cocatalyst modi cation is necessary as ·OH radicals have been widely regarded as the main species that induced CH 4 activation and over-oxidation [40,41]. When CH 3 OH served as the desired products, over-oxidation to HCHO or CO 2 would be suppressed by lowering the oxidative potential of photogenerated hole through cocatalyst modi cation, thus improving the selectivity. Stimulated by molecular catalysis, single atom cocatalysts promise an extremely high e ciency, where atomic dispersed species with unsaturated coordination environment could improve the catalytic performances based on the unique electronic structure [42][43][44]. Meanwhile, high atom utilization e ciency could be achieved [45,46]. On the other hand, since CH 4 exhibited low electron and proton a nity, moderate decoration of defective sites could enhance the chemicaladsorption of non-polar molecular, then promoting the activation of CH 4 [47]. Therefore, the integration of both defects and single atom cocatalyst decoration could boost charge separation, weaken oxidative potential and enhance CH 4 activation on a photocatalyst.
Herein, atomically dispersed palladium (Pd) supported on defective In 2 O 3 was prepared and served as the visible-light responsive photocatalyst for CH 4 conversion to high value chemicals. Under 420 nm irradiation, the optimized production of oxygenates reached up to ca. 300 µmol in 3 h, with a very high selectivity of 82.5 % of the primary products. In-situ XPS and EPR spectra were conducted to investigate the charge transfer dynamics. The results indicated the dedicated roles of Pd atoms and oxygen vacancies (OVs) in promoting the transfer of photo-induced holes and electrons, respectively. DFT calculation results indicated H 2 O could also promoted the desorption of the oxygenate products, thus suppressing over-oxidation and facilitate high selectivity of primary products. The introduction of atomic Pd and oxygen vacancies further enhanced this effect on suppressing over-oxidation. Isotopic labelled experiments further proved the methane conversion pathway.  (Figure 1a). Under 420 nm irradiation, the products including CH 3 OH, CH 3 OOH and HCHO over Pd-In 2 O 3 reaches 13.4, 32.3 and 27.5 μmol in 3 h reaction, respectively.

Results And Discussion
The selectivity of the primary products (CH 3 OH and CH 3 OOH) was 62.1 % and the selectivity to the overoxidation products (HCHO and CO 2 ) was 37.9 %. In comparison, Au-In 2 O 3 and Pt-In 2 O 3 performed almost 100 % over-oxidized products (HCHO), with the trace yields of 1.4 and 0.9 μmol HCHO, respectively. Such differences suggested Pd cocatalyst was more suitable than Pt and Au for CH 4 activation to produce these primary products. The yield of oxygenates for Pd-In 2 O 3 was improved further to 179.7 μmol by the introduction of defective sites to form Pd-def-In 2 O 3 , 2.5 times higher than that of Pd- dosage increasing from 25 to 100 mL, which could be attributed to the enhanced desorption of the products from the surface of the photocatalyst when more water was used as discussed later. Notably, in the absence of H 2 O dosage, CO 2 (8.5 μmol) was produced as the only product, suggesting the critical role of H 2 O in promoting CH 4 activation as well suppressing over-oxidation, probably ascribed to the production of ·OH radical and promotion desorption of oxygenates by H 2 O [48]. While increasing the total pressure of the gaseous reactants, CH 4 dissolved increased and the oxygenate production gradually  [49]. For Pd-In 2 O 3 and Pd-def-In 2 O 3 , the dominant peak exhibited a slight left-shift from 130.6 to 129.9 cm -1 , attributed to the surface stain effect induced by the Pd cocatalyst deposition [50].
Electron paramagnetic resonance (EPR) spectra were conducted to evaluate the spin-electrons including oxygen vacancies (Figure 2b). For the pristine In 2 O 3 and Pd-In 2 O 3 , a single Lorentz peak at g = 1.882 was observed, ascribed to the electrons on the conduction band (CB) [51,52]. In the case of Pd-def-In 2 O 3 , the signal of this peak exhibited much stronger intensity than the others, suggesting the higher electron density on CB. Meanwhile, an additional Lorentz peak was observed at g = 2.001, which could be attributed to the free-electrons trapped by the oxygen vacancies [52], thus suggesting the existence of oxygen vacancies in Pd-def-In 2 O 3 . The introduction of oxygen vacancies might contribute to the stronger EPR peak at g = 1.882.
High-resolution transmission electron microscope (HRTEM) images further proved the defective structure of Pd-def-In 2 O 3 (Figure 2c In-situ high-resolution Pd 3d X-ray photoelectric spectra (XPS) in dark and under light were conducted to study the charge transfer direction of Pd-def-In 2 O 3 (Figure 3b and Figure S3). In dark, the Pd 3d5/2 XPS peak could be deconvoluted into two binding peaks at 336.55 and 335.38 eV, which were assigned to the Pd 2+ and Pd 0 species, respectively [53]. Under light irradiation, the peak exhibited a left-shift to higher binding energy ( Figure S3). Further deconvoluted results (Figure 3b) suggested that Pd 2+ content increased to 26.3 %, much higher than 6.9 % in dark. Such increased Pd 2+ content suggested Pd served as the hole acceptors upon excitation. In-situ EPR spectra under light were conducted to evaluate the role of OVs. As shown in Figure 3c, the signal at g = 2.0009 was attributed to the electrons trapped by the oxygen vacancies, which performed gradually increasing intensity from 100 % to 226 % with the prolonged irradiation to 360 seconds. This stronger EPR intensity suggested a higher concentration of spin-electrons and thus demonstrated OVs served as the electron acceptor [54]. Therefore, single atom Pd and OVs separately acted as the hole and electron acceptors under light irradiation, which would greatly contribute to the enhanced charge separation.
Photocurrent responses ( Figure S4) were tested to evaluate the charge separation e ciency. Pristine  Figure 3d, a relatively strong PL emission peak was observed for the pristine In 2 O 3 , attributed to the severe charge recombination. In comparison, the PL intensity for Pd-In 2 O 3 was greatly weakened, indicating the suppressed charge recombination. For Pd-def-In 2 O 3 photocatalyst, the most weakened PL peak were observed, ascribed to the most enhanced charge separation e ciency, which was corresponding with the photocurrent analysis. Time-decay PL spectra were conducted to evaluate the PL lifetime. As shown in Figure S5, Pd-def-In 2 O 3 photocatalyst exhibited the slowest PL decay kinetics. The tting results (Table S1) showed that Pd-def-In 2 O 3 exhibited the average PL lifetime at 4.99 ns, longer than that of In 2 O 3 (3.60 ns) and Pd-In 2 O 3 (4.28 ns), which would be bene cial to the e cient utilization of separated charge carriers.
Reactive oxygen species including ·OOH and ·OH radicals were widely regarded as the main active species for CH 4 activation [55] and monitored by in-situ EPR spectra with 5, 5-dimethyl-1-pyrroline N-oxide (DMPO) as the spin-electron trapping agents. As shown in Figure 4a, the DMPO-OOH adduct was detected under light over different photocatalysts and ascribed to the presence of ·OOH, which came from the reduction of O 2 molecule with photo-induced electrons and H + . A stronger intensity of DMPO-OOH was observed for Pd-def-In 2 O 3 , suggesting the production of ·OOH radical was enhanced by the integration of single atom Pd and OVs. On the other hand, in-situ EPR spectra under light was used to monitor the generation of ·OH radical with DMPO as the trapping agent in H 2 O. The 1:2:2:1 quartet signals were observed and assigned to the DMPO-OH adduct, suggesting the generation of ·OH radical (Figure 4b). It was obvious that Pd-def-In 2 O 3 produced much more ·OH under identical conditions than Pd-In 2 O 3 and In 2 O 3 was the worse. It is believed that ·OH initially activates CH 4 to methyl radical (·CH 3 ), thus Pd-def-In 2 O 3 performed CH 4 activation best followed by Pd-In 2 O 3 , which is consistent with the step by step enhanced photocatalytic performances by Pd and then both Pd and oxygen vacancies, indicating that oxygen vacancies could promote charge separation and also facilitate water oxidation reaction on Pd. Coumarin was used as the probe for ·OH radical detection due to the easy reaction between coumarin and ·OH to produce 7-hydroxycoumain (7-HC) that could be detected by UV-vis spectra at 412 nm ( Figure  4c). The results further supported that Pd-def-In 2 O 3 held the strongest ability for ·OH production, which facilitated CH 4 activation. Therefore, single atom Pd worked as the hole acceptor, which then catalysed ·OH radical production from water oxidation. Simultaneously, OVs acted as the electron acceptor, which then catalysed O 2 reduction to generate ·OOH radical. the signals at m/z = 34 and 33 were attributed to the isotopic labelled CH 3 18 OH and its fragment ( Figure   4d), suggesting O 2 was the only oxygen source for CH 3 OH formation. Carbon source for methanol production were also studied in the presence of 5 bar isotopic labelled 13 CH 4 (Figure 4e), where the signal of mass spectra (MS) at m/z = 33 was ascribed to 13 CH 3 OH, demonstrating that CH 4 was the carbon source for oxygenates production.
DFT calculations ( Figure 5) were conducted to explain the improved selectivity of primary products. It should be noted that timely desorption of the primary products on the active sites could e ciently avoid its deep-oxidation to HCHO and CO 2 . As ·OH radical was regarded as the main species that induced oxidation on single atom Pd cocatalyst, it was accordingly considered that the e cient desorption of primary products like CH 3 Figure 5a  In order to suppress over-oxidation, it was also critical to enhance the desorption of primary oxygenate products by H 2 O, as supported by the DFT calculation.
In summary, visible-light-driven CH 4 conversion at ambient temperature was reported over the In 2 O 3 nanorod photocatalyst with loading of Pd single atoms cocatalysts and oxygen vacancies. Under 420 nm irradiation, superior yield (99.7 μmol·h -1 ) and selectivity (82.5 %) of the primary products were achieved on Pd-def-In 2 O 3 photocatalyst under optimized reaction conditions. In-situ XPS and EPR spectra under visible light irradiation indicated that Pd and oxygen vacancies acted as the hole and electron acceptors, respectively, thus synergistically boosted charge separation and transfer. Isotopic labelled experiment proved that O 2 was the only oxygen source for oxygenates production, while H 2 O was the promoter of CH 4 activation through the production of ·OH radical as monitored by the in-situ EPR spectra with DMPO as the spin-trapping agent. DFT calculation results suggested that H 2 O performed much larger adsorption energies than CH 3 OH on either In 2 O 3 , def-In 2 O 3 or Pd-def-In 2 O 3 , suggesting the stronger adsorption of H 2 O than CH 3 OH, which was bene cial to the timely desorption of the produced CH 3 OH, thus avoiding further over-oxidation. The introduction of Pd and oxygen vacancies could further improve the selectivity of primary oxygenates mainly through the greatly enhanced adsorption of H 2 O and the reduced oxidation potential of photoinduced holes. This work provided an useful avenue on co-modi cation by oxidative single atom cocatalyst and oxygen vacancies for simultaneous regulation of both activity and selectivity through enhancing charge separation, moderated photohole oxidation ability and timely promoted desorption of primary products by a solvent.

Synthesis of In 2 O 3 nanorods
In 2 O 3 nanorods were prepared according to the previous study [56]. Typically, 12.0 g urea and 1.

Analysis of hydroxyl radical (·OH)
Coumarin was used as the probe for the quanti cation of ·OH via the production of 7-HC [40]. Typically, 20 mg photocatalyst was dispersed in 100 mL aqueous coumarin solution (5×10 -4 M). After stirring for 30 min in dark, the suspension was irradiated with the LED light source (420 nm, PLS-LED100C, Beijing Perfectlight Technology Co., Ltd). Certain amount of suspension was sampled and ltered in the 10 min intervals. PL intensity of the produced 7-HC in the solution was then measured on the F4500 spectro uorometer.

DFT calculation of adsorption energies
The rst-principles were employed to perform all the density functional theory (DFT) calculations within the generalized gradient approximation (GGA) using the PBE formulation. The projected augmented wave (PAW) potentials have been chosen to describe the ionic cores and take valence electrons into account using a plane wave basis set with a kinetic energy cutoff of 400 eV. Partial occupancies of the Kohn−Sham orbitals were allowed using the Gaussian smearing method and a width of 0.05 eV. The electronic energy was considered self-consistent when the energy change was smaller than 10 −4 eV. A geometry optimization was considered convergent when the force change was smaller than 0.05 eV/Å. Grimme's DFT-D3 methodology was applied to describe the dispersion interactions. Three models including In 2 O 3 with (111) facet, def-In 2 O 3 with one oxygen vacancy and Pd-def-In 2 O 3 with both one oxygen vacancy and single atom Pd modi cation were conducted. During structural optimizations, the 2×2×1 Monkhorst-Pack k-point grid for Brillouin zone was used for k-point sampling for structures. Finally, the adsorption energies (E ads ) were calculated as E ads = E ad/sub -E ad -E sub , where E ad/sub , E ad , and E sub were the total energies of the optimized adsorbate/substrate system, the adsorbate in the structure, and the clean substrate, respectively.

Data availability
The data that support the ndings of this study are available from the corresponding author upon reasonable request.