David Vahey was having a bad day in the lab. The PhD researcher at St John's College, Cambridge, was testing a photocatalyst when he decided to remove it as part of a routine control experiment. The reaction should have flopped. Instead, it worked just as well — and in some cases better — without the catalyst at all.

At first, the unusual result looked like a mistake. Most researchers might have binned it and moved on. Vahey didn't.

"Failure after failure, then we found something we weren't expecting in the mess — a real diamond in the rough," Vahey said. "And it is all thanks to a failed control experiment."

Light replaces toxins

What Vahey and his supervisor, Professor Erwin Reisner, had stumbled onto was a fundamentally new way to modify complex drug molecules. Published this month in Nature Synthesis, the technique uses a simple LED lamp to trigger a self-sustaining chain reaction that forges new carbon–carbon bonds — the essential links that underpin everything from fuels to the medicines in your bathroom cabinet.

Traditionally, forming these bonds requires a process called the Friedel–Crafts reaction: harsh chemicals, heavy metal catalysts, and aggressive conditions. That means the reaction can only happen early in drug manufacturing, followed by many additional steps to reach the final product.

The Cambridge team's "anti-Friedel–Crafts" approach flips the script. It lets chemists make precise, targeted changes to drug molecules at the final stages of production — under mild conditions, at room temperature, and without toxic or expensive reagents.

Why it matters for medicine

In drug development, even a tiny structural tweak can transform how a medicine works in the body — affecting its potency, its behaviour, or its side effects. But making that tweak has traditionally meant dismantling and rebuilding complex molecules from scratch, a process that can eat up months.

"Scientists can spend months rebuilding large parts of a molecule just to test one small change," Vahey explained. "Now, instead of doing a multistep process for hundreds of molecules, scientists can start with their hit and make small modifications later on."

Fewer steps also mean fewer chemicals, less energy consumption, and a smaller environmental footprint — an increasing priority for an industry trying to clean up its act.

"Transitioning the chemical industry to a sustainable industry is arguably one of the most difficult parts of the whole energy transition," said Reisner, Professor of Energy and Sustainability in Cambridge's Yusuf Hamied Department of Chemistry.

From serendipity to industry

The team demonstrated the reaction across a wide range of drug-like molecules and showed it could be adapted to continuous-flow systems used in industrial manufacturing. A collaboration with pharmaceutical giant AstraZeneca tested whether the method could meet the practical and environmental demands of large-scale drug production.

Machine-learning models, developed with Trinity College Dublin, were then trained on the reaction's patterns to predict where it would occur on entirely new, untested molecules — helping researchers identify promising drug candidates faster.

But for all the power of AI, Reisner is clear about what made this discovery possible: human curiosity.

"David could have dismissed it as a failed control," Reisner said. "Instead, he stopped and thought about what he was seeing. That moment, choosing to investigate rather than ignore it, is where discovery happens."

The good days

The story echoes some of science's most celebrated accidents — from penicillin to X-rays to modern weight-loss drugs. Reisner summed it up with a line that every scientist will recognise:

"As a chemist, you only need one or two good days a year — and those can come from a failed experiment."

For Vahey, the work is just the beginning. "What industry and other researchers do with it next — that's where the future impact lies."