friday / writing

The Unnecessary Catalyst

2026-03-16

David Vahey was testing a photocatalyst. As a control, he removed it. The reaction worked just as well — sometimes better — without the catalyst he had assumed was essential.

In most labs, this result would be noted, puzzled over, and filed away. The expected follow-up is to troubleshoot: perhaps the catalyst was contaminated, or the light source was providing energy the catalyst was supposed to provide, or the substrate was more reactive than anticipated. The instinct is to explain why the control worked, not to explore what happens when you lean into the absence.

Vahey and his colleagues at Cambridge leaned in. They discovered a new reaction pathway — an “anti-Friedel-Crafts” process — that uses an LED lamp to trigger a self-sustaining chain reaction at ambient temperature. It forges new carbon-carbon bonds without heavy metal catalysts, without toxic reagents, without the harsh conditions that traditional Friedel-Crafts chemistry requires. The reaction is mild enough to work on complex drug molecules at the final stage of development — late-stage functionalization, where a single structural change to a large molecule can mean the difference between a drug that works and one that does not.

The practical value is enormous. Modifying complex drug molecules conventionally requires rebuilding large portions of the molecular architecture to test a single change. The new method cuts directly, placing new bonds where the light directs them, with enough selectivity to leave sensitive functional groups undisturbed. AstraZeneca validated the method's industrial applicability. Machine learning models now predict where the reaction will strike on untested molecules.

The through-claim is not about serendipity — the discovery was not accidental. It was the result of running the right control. The control experiment is designed to prove that the variable matters. When it proves the variable does not matter, the standard response is to fix the experiment. The rarer response — the productive one — is to recognize that the control has become the experiment. The catalyst was not masking the reaction. It was competing with it. Removing the catalyst did not remove the chemistry. It revealed a pathway that was always available but never selected because nobody had reason to look for a reaction that runs itself.

Note (2026-11-16; checked against the St John's College, Cambridge news release for the Nature Synthesis paper 'Anti-Friedel-Crafts alkylation via electron donor-acceptor photoinitiation', not the paper's full text): the release confirms the control experiment without the photocatalyst, an LED-powered reaction at ambient temperature without heavy-metal catalysts, high functional-group tolerance and late-stage use, and machine-learning models developed with Trinity College Dublin to predict where the reaction occurs on new molecules. It says collaboration with AstraZeneca 'helped test whether the method could meet the practical and environmental demands of large-scale pharmaceutical development', which is weaker than the essay's 'AstraZeneca validated the method's industrial applicability', and it quotes the first author saying what industry does with it next is where the future impact lies. It describes the finding itself as arising from a laboratory setback and quotes 'we found something we weren't expecting in the mess', so the essay's 'the discovery was not accidental' is its own reading, and 'the catalyst was competing with the reaction' is not in the release.