High-performance photocatalytic nonoxidative conversion of methane to ethane and hydrogen by heteroatoms-engineered TiO2

Nonoxidative coupling of methane (NOCM) is a highly important process to simultaneously produce multicarbons and hydrogen. Although oxide-based photocatalysis opens opportunities for NOCM at mild condition, it suffers from unsatisfying selectivity and durability, due to overoxidation of CH4 with lattice oxygen. Here, we propose a heteroatom engineering strategy for highly active, selective and durable photocatalytic NOCM. Demonstrated by commonly used TiO2 photocatalyst, construction of Pd–O4 in surface reduces contribution of O sites to valence band, overcoming the limitations. In contrast to state of the art, 94.3% selectivity is achieved for C2H6 production at 0.91 mmol g–1 h–1 along with stoichiometric H2 production, approaching the level of thermocatalysis at relatively mild condition. As a benchmark, apparent quantum efficiency reaches 3.05% at 350 nm. Further elemental doping can elevate durability over 24 h by stabilizing lattice oxygen. This work provides new insights for high-performance photocatalytic NOCM by atomic engineering.

We prepare the TiO2 nanocrystals dominated with (100) and (101) planes according to the previous report and further loaded them with Pd SAs respectively for photocatalytic NOCM measurements ( Supplementary Fig. 4a) 3 . The catalytic results show that TiO2 (001) facet with Pd SAs loading exhibits the highest activity and selectivity of C2H6 production. To further validate the correlation between the lattice plane and its electron structure of the catalyst, we calculate the density of states (DOS) of Pd1/TiO2 with (100) or (101) facet of the supports (Supplementary Fig. 4b and 4c). The Pd single atoms only show 21% and 0.8% occupied states in VBM of their Pd1/TiO2 (100) and Pd1/TiO2 (101) structures, respectively, lower than the 47% occupied states in VBM of Pd1/TiO2 (001). The computational results indicate that the photogenerated holes prefer to be accumulated at O sites in the cases of (100) and (101)  We examine whether molecular oxygen exists in the reactor to induce overoxidation of CH4 to CO2.
Our control experiment shows that adding gaseous oxygen into the reactor results in producing a large amount of CO2 as well as forming H2O instead of H2 ( Supplementary Fig. 14), consistent with the previous reports for photocatalytic oxidative coupling of methane 4 . This demonstrates that the oxidant for formation of CO2 during NOCM is provided by the lattice oxygen of catalyst. Supplementary Fig. 15 Isotope labelling measurements. (a) Mass spectra of C2H6 produced from NOCM process over Pd1/TiO2 using 12 CH4, 13 CH4 or CD4 as the reactant. Mass spectra of (b) C2D4 and (c) 13 CO2 produced from NOCM process over Pd1/TiO2 using CD4 or 13 CH4 as the reactant, respectively.
The origin of C2+ compounds is verified by using 13 CH4, 12 CH4 and deuterated methane (CD4) as carbon source. The m/z values of C2H6 are shift to higher mass/charge ratios with constant relative intensity when CH4 is replaced by 13 CH4 or CD4 while the mass spectra of generated C2H4 and CO2 show similar trends ( Supplementary Fig. 15), confirming that C2H6 is transformed from CH4. Electron paramagnetic resonance (EPR) measurement is performed to investigate possible radicals by using 5,5′-dimethyl-1-pyrroline-N-oxide (DMPO) as a radical scavenger 6,7 . In order to detect the radicals, the reaction is required to carry out in water with CH4 as filling gas under light irradiation. Fig. 19 In situ DRIFTS measurements in the dark. In situ DRIFTS spectra for CH4 adsorption on Pd1/TiO2 at room temperature in the dark. Fig. 20 In situ DRIFTS measurements of TiO2. In situ DRIFTS spectra for photocatalytic NOCM over pure TiO2. The time-dependent NOCM on Pd1/TiO2 shows that the performance decays when the reaction time reaches 6 h ( Supplementary Fig. 25a), which should be ascribed to the consumption of lattice oxygen in TiO2 for CO2 production. The catalytic performance for the first and second cycles is displayed in Supplementary Fig. 25b. Without catalyst regeneration treatment, the yields of C2H6 and CO2 in the second cycle are 3.5 and 2.1 times lower than that in the first cycle, respectively. Supplementary Fig. 26 Catalyst recycling. Reaction-regeneration cycles in photocatalytic NOCM over Pd1/TiO2. The regeneration treatment exposes catalyst to air with 80 °C heating.

Supplementary
The consumed lattice oxygen can be regenerated by heating the catalyst in air, recovering the photocatalytic activity. As shown in Supplementary Fig. 26, the yields of carbonaceous products and H2 are recovered after the heating treatment over Pd1/TiO2 in the air. As shown in Supplementary Fig. 34, when the temperature is increased during temperatureprogrammed oxidation (TPO), the peak starting around 348 K indicates that sample oxidation can occur at low temperature by O2. Furthermore, the change of O content (NO) can be determined using the relationship NO = 32  CO/mTPO, where CO is the O2 consumption measured by TPO and mTPO is the mass of TPO sample with 32 being the molar mass of O2. The CO is 13.2 μmol measured by TPO, and as such, the NO is determined to be 0.7 wt.%.  Supplementary Fig. 35a. The peaks at 562 and 678 ppm correspond to the 17 O resonance of OTi3 and OTi2 11 . The peak intensity at 678 ppm decreases after NOCM measurements, indicating that the content of oxygen with 2-fold (OTi2) oxide coordination on TiO2 surface has been reduced. However, the signal-to-noise ratio of 17 O MAS NMR spectra is unsatisfactory as the abundance of 17 O is extremely low in the nature.
Moreover, the long collection time for 17 O MAS NMR spectra (more than 20 h) may lead to the unexpected oxidation of catalyst by the air. For this reason, the EPR is employed to further verify the location of O consumption during the NOCM process. As shown in Supplementary Fig. 35b,