The core of the innovation is a photon–phonon co-driven strategy, abbreviated PPCC by the team. Conventional photocatalysis uses light alone to drive chemical reactions, while thermocatalysis relies purely on heat. Each approach has inherent limits. Photocatalysis suffers from a quantum efficiency ceiling, because a single photon can only do so much work, and thermocatalysis often forces a trade-off between yield and selectivity. Methanol conversion to hydrogen and formaldehyde is particularly demanding: it is an endothermic process that doubles the gas-phase volume, and it proceeds through two sequential dehydrogenation steps, each with a different activation energy. No single energy input can efficiently satisfy both. By coupling 365-nanometre ultraviolet light from an LED with moderate heating at 150 degrees Celsius in a continuous-flow reactor, the Tsinghua team matched each elementary step with the energy source best suited to it, photons for the hardest activation step and phonons, the lattice vibrations that carry thermal energy, for the rest.

The catalyst itself is a masterclass in atomic-scale engineering. Platinum single atoms were anchored to a mixed anatase-rutile titanium dioxide support and surrounded by lithium, forming what the evidence suggests are PtLi2 clusters. The optimal formulation, containing roughly 0.3 weight percent platinum and 0.2 weight percent nominal lithium input, delivered a hydrogen production rate of 4.36 moles per gram of catalyst per hour at 150 degrees Celsius, a figure 30.3 times higher than unmodified TiO2 and roughly an order of magnitude above the best low-temperature thermal catalytic benchmarks. The apparent turnover frequency reached a record 275,000 per hour. Even more striking is the overall quantum yield of 184 percent, a value exceeding 100 percent because thermal energy contributes additional driving force beyond the incident photons alone. In the dark, the same catalyst at the same temperature managed only 0.11 moles per gram per hour, confirming that light is the primary driver and heat the essential promoter.

Crucially, the carbon does not end up as carbon dioxide. Instead, the dominant carbon-containing product is formaldehyde, produced with 94 percent selectivity at a synthesis rate of 3.5 moles per gram per hour. In batch experiments, the team accumulated a concentrated formaldehyde solution of 21.2 weight percent after just 140 minutes, a strength approaching commercial grades used in medicine and in the manufacture of industrial resins and plastics. Formaldehyde is currently made by oxidizing methanol over silver catalysts at around 600 degrees Celsius or iron-based catalysts at around 350 degrees, a process that consumes more than 30 percent of global methanol output. Co-generating hydrogen and formaldehyde from a single mild-temperature process therefore offers two saleable products with net-zero carbon emissions and complete atomic utilization of the methanol feedstock.

Lithium turned out to be the quiet hero of the story, performing no fewer than five distinct jobs. First, it retards the recombination of photogenerated charge carriers in the catalyst bulk, extending the lifetime of the reactive electrons and holes. Second, it facilitates electron transfer from the TiO2 bulk to the platinum reaction sites, with density functional theory calculations showing that the intrinsic electron transfer energy becomes more favourable as lithium is added, and that lithium lowers the quantum tunnelling barrier for directional electron injection. Third, it weakens the binding of hydrogen atoms on the platinum surface; computed adsorption free energies shift from a strongly binding minus 0.32 electronvolts on platinum nanoparticles to a nearly ideal minus 0.04 electronvolts on the Pt1Li2 site, easing hydrogen desorption and avoiding catalyst poisoning. Fourth, it maximizes the fraction of platinum present as Pt2+ rather than metallic Pt0, which the team’s poisoning experiments identified as the true active sites. Fifth, and perhaps most importantly, it stabilizes those single atoms against aggregation.

That stabilizing role was demonstrated with unusual rigour. X-ray photoelectron spectroscopy showed that in a lithium-free Pt–TiO2 catalyst the Pt2+ fraction collapsed from 70 percent to 29 percent within one hour of reaction, while in the lithium-modified material it barely moved, from 100 percent to 96 percent. Extended X-ray absorption fine structure revealed growing platinum-platinum coordination in the unmodified catalyst, a signature of clustering, but no such change in the lithium version. Identical-location electron microscopy, in which the very same catalyst particles were imaged before and after one hour under realistic reaction conditions, directly visualized new platinum nanoparticles forming and existing particles growing on Pt–TiO2, while the lithium-stabilized sample remained atomically dispersed. The practical consequence was dramatic: the catalyst ran stably for 19 days under a simulated solar cycle of eight hours illumination and sixteen hours darkness, maintaining hydrogen output and formaldehyde selectivity above 94 percent throughout.

The mechanistic picture that emerges from transient absorption spectroscopy, isotope labelling, in situ electron paramagnetic resonance and infrared measurements is an elegant division of labour between light and heat. Photons excite charge carriers in TiO2, which couple to the lattice to form polarons; thermal phonons then help these polarons hop to the surface rather than remaining trapped. On the surface, water adsorbs more strongly than methanol, so photogenerated holes preferentially oxidize water to hydroxyl radicals. These radicals, rather than the holes themselves, attack methanol, abstracting a hydrogen atom to form the hydroxymethyl radical, a step calculated to cost only 0.37 electronvolts compared with roughly 0.83 electronvolts for direct hole-driven cleavage. Kinetic isotope experiments confirmed that carbon-hydrogen bond cleavage in methanol is the rate-determining step, and it is precisely this demanding step that the energetic photogenerated carriers are positioned to drive. The subsequent dehydrogenation of the hydroxymethyl intermediate to formaldehyde, by contrast, is a nearly barrierless thermal transformation that phonons alone can push forward.

The team also showed the concept scales in a practical sense: connecting two flow reactors in series boosted hydrogen production by 1.7 times, and the system tolerated variations in light intensity, pressure, flow rate and catalyst loading that were systematically optimized. The 40 percent methanol feed concentration proved optimal, with control experiments showing that any carbon dioxide formed arises secondarily from formaldehyde reforming with water rather than from direct over-oxidation. Isotope labelling with deuterated methanol confirmed that hydrogen atoms from the methyl group of methanol end up in the hydrogen gas, while hydroxyl hydrogens exchange with water, consistent with the proposed cascade.

Beyond the headline numbers, the significance of this work lies in its conceptual generality. Photon–phonon co-driven catalysis offers a framework for any transformation hobbled by an activity-selectivity trade-off, by assigning each elementary step to the energy input that matches its activation barrier. The authors argue that future progress will depend on reactor engineering tailored to dual energy fields, explicitly managing photon transport and heat flow to balance the energy budget. If those engineering challenges can be met, a methanol economy in which the liquid fuel doubles as a hydrogen battery and a chemical feedstock, releasing no carbon dioxide in the process, moves a decisive step closer to industrial reality.

Subject of Research: Photon–phonon co-driven catalytic conversion of methanol to hydrogen and formaldehyde using lithium-modified platinum single atoms on titanium dioxide

Article Title: Green synthesis of hydrogen and formaldehyde from methanol with enhanced quantum efficiency

Article References: Zhao, J., Xiong, L., Zheng, Y., Nie, Q., Yu, Z., Guo, J., Li, X., Liu, Q., Sun, X., Zheng, X., Wang, Z., Selloni, A., Gong, X., Huang, W., & Tang, J. (2026). Green synthesis of hydrogen and formaldehyde from methanol with enhanced quantum efficiency. Nature Sustainability. https://doi.org/10.1038/s41893-026-01959-9

Image Credits: AI Generated

DOI: 10.1038/s41893-026-01959-9

Keywords: hydrogen production, methanol, formaldehyde, single-atom catalysis, photocatalysis, photon–phonon co-driven catalysis, platinum catalyst, titanium dioxide, lithium modification, quantum yield, green chemistry, Nature Sustainability