Solar energy produces hydrogen fuel from waste

Researchers at the University of Cambridge have scaled up a sunlight-driven panel that turns plastic bottles into hydrogen. They tested a 1 m² panel outdoors for the first time (Nat. Chem. Eng. 2026, DOI: 10.1038/s44286-026-00406-y).

Hydrogen demand has nearly doubled over the past 2 decades, and more than 95% of it still comes from fossil fuels. The classical alternative is splitting water with sunlight to release hydrogen, but that process consumes a lot of energy. Waste organics, like plastics, take less energy to break down than water, and the process also yields by-products useful to the chemical industry. “It’s energetically easier, and we create more value than the classical processes,” says Erwin Reisner, who led the research.

The process, called photoreforming, begins with solid plastic waste that has been converted to a solution. That liquid flows into a reactor, which holds two key materials: a semiconductor that absorbs sunlight and a catalyst that helps the reaction produce hydrogen gas.

Sunlight striking the semiconductor frees electrons, each one leaving behind a positively charged gap called a hole. The holes pull electrons away from the dissolved plastic, oxidizing it, and releasing protons along with a few other useful chemical species. The catalyst then helps those protons combine with the freed electrons to form hydrogen gas.

Photoreforming isn’t new. The novel aspect of the Cambridge study is how the team built the cobalt catalyst that drives hydrogen formation. Earlier methods for attaching that catalyst to a panel required high heat or polymer binders, both impractical at large scale. Instead, the team used a single-source precursor: one molecule that contains cobalt and zirconium in a fixed ratio. “If you have two different compounds, it’s much more difficult to control the composition of your catalyst film,” Reisner says.

Dissolved in a solvent, the precursor can be sprayed onto glass and dried at room temperature. “The films are also transparent,” Reisner adds, “unlike many precious-metal catalysts that block light from reaching the semiconductor underneath.”

Spray coating performed as well as the slower deposition methods it replaced, the team found. Outdoors in Cambridge under sunlight about half as strong as their usual lab simulators, a 1 m² array ran for 6 h; hydrogen output reached 5.24 mmol per square meter from glucose and 1.51 mmol per square meter from pretreated cellulose. Cellulose also produced formate and acetate. Those numbers are modest. “But if it works in the UK, it works anywhere,” Reisner jokes, referring to the country’s weak sunlight.

In smaller, lab-scale tests, the team also tried real PET water bottles, cutting one and dissolving it in a hot alkaline solution for several days to break down the plastic. Fed into the same kind of reactor, it produced less hydrogen than was produced from cellulose at that scale, but the reaction worked and made useful by-products as well. Reisner says that hydrogen may not even be the most valuable product. The by-products left behind could be more useful to the chemical industry.

Alberto Puga, a chemical engineer at Rovira i Virgili University who wasn’t involved in the work, calls the single-source precursor “novel and judiciously designed.” Avoiding precious metals should help it scale, he says, since cobalt is “much more abundant and cheaper than state-of-the-art platinum co-catalysts.”

But leaching remains a concern: about 60% of the cobalt dissolved after 22 h of use. Respraying fresh catalyst restores performance, but doing that repeatedly isn’t practical, says Ariffin Bin Mohamad Annuar, a graduate student who ran the experiments.

Earlier cost analyses for this kind of hydrogen were largely guesswork, based on how well an idealized system might perform, since few research groups had built one at large scale. Reisner’s team could plug in real, measured numbers instead. They found that hydrogen made in this way costs more than the conventional, fossil fuel–based process.

Another shortcoming is that the semiconductor captures only UV light, which is roughly 5% of the solar spectrum. Next, the researchers want a semiconductor that captures visible light in addition to UV. They also want a continuous-flow setup instead of having to refill the reactor between runs, and a catalyst that produces fewer by-products to avoid the need for chemical separations.