On the mechanism of photocatalytic reactions with eosin Y

A reaction running in a flask under green light, with or without metals and with or without rare-earth catalysts — just a pink dye borrowed from biology labs. Chemists have been calling this photocatalysis. But is it? That single question is the engine of a careful, methodical paper by Majek, Filace, and Jacobi von Wangelin, and the tool they use to answer it is a number called quantum yield — how many molecules of product you get per photon absorbed. It sounds simple. What it revealed was not. For most of the history of photoredox chemistry, the catalysts of choice were metal complexes — ruthenium and iridium compounds that absorb visible light and drive electron transfer with impressive efficiency. They work. The problem is that ruthenium and iridium are expensive, toxic to handle, and difficult to remove from products. So the field began looking at organic dyes, and eosin Y became a favorite almost immediately. It is commercially available, it is cheap, and it is the same pink stain biologists use to color tissue sections. Chemically, it is the tetrabromo derivative of fluorescein, and its excited-state redox potential — estimated at about 1.1 volts versus the standard calomel electrode — is strong enough to reduce arenediazonium salts, the reaction partners that matter here. Arenediazonium salts are compounds where an aromatic ring carries a nitrogen-nitrogen unit with a positive charge. They are easy to make from anilines, they release nitrogen gas when the carbon-nitrogen bond breaks, and the redox potentials of most substituted versions cluster around 0.0 volts versus the standard calomel electrode. That makes them unusually easy to reduce. The aryl radical you get when that bond cleaves is a versatile intermediate — it can build biaryls, stilbenes, benzothiophenes, and arylboron compounds. On paper, eosin Y absorbs green light, transfers an electron to the diazonium salt, and the cycle closes. Clean. Elegant. Photocatalytic. Except Majek and colleagues found that the picture underneath was considerably messier. Three complications, operating simultaneously, had been mostly ignored by the field. The first is eosin Y itself. In solution, it does not exist as a single species. It cycles through four acid-base forms: a spirocyclic form, a neutral form, a monoanion, and a dianion. The two acidic protons have pKa values of about 2.0 and 3.8 in water, so deprotonation happens easily. The spirocyclic form is photocatalytically inactive — its conjugation is interrupted. The neutral form shows only weak, very short-lived fluorescence, which makes it poorly suited for driving redox chemistry. The active species are the mono- and dianionic forms, which absorb strongly at 535 nanometers and produce the long-lived excited states needed for electron transfer. Here is why that matters in practice. When eosin Y is used at catalytic concentrations — small amounts in organic solvents — the equilibrium between these forms is exquisitely sensitive to conditions. In acetonitrile, the dye sits predominantly in its yellow, weakly fluorescent, inactive forms. Add a trace of base, and the solution shifts: it turns strongly fluorescent, absorption at 535 nanometers appears, and catalytic activity is restored. The solvent dimethyl sulfoxide, being a stronger base than acetonitrile, pushes the equilibrium toward the active anionic forms on its own. Majek and colleagues showed this concretely: phenanthrene synthesis and photoborylation worked in dimethyl sulfoxide without added base, but the same reactions in acetonitrile required explicit base addition. Trace impurities, solvent purity, and even the specific commercial form of eosin Y — the "spirit soluble" version sold by Sigma-Aldrich may be the spirocyclic or neutral form — can determine whether the catalyst is doing anything at all. The second complication is the lamp. Much of the published eosin Y photocatalysis literature used broad-spectrum compact fluorescent lamps or white light-emitting diodes. Those sources emit substantial power between 400 and 500 nanometers. And arenediazonium salts absorb in that region. A photon at 400 nanometers carries about 3.1 electron volts of energy. The thermal activation energy for breaking the carbon-nitrogen bond in a diazonium salt is approximately 115 kilojoules per mole — about 1.19 electron volts. The photon carries more than twice the energy needed. No catalyst required. The team measured this directly. For a mixture of a para-bromobenzenediazonium salt with a boron reagent, the absorbance exceeded 0.1 at 400 nanometers, meaning roughly twenty-one percent of incident light at that wavelength was being absorbed by the substrate. For an ortho-biphenyl diazonium salt, essentially all light at 400 nanometers was absorbed before the dye saw any of it. When they ran the borylation reaction under broad-spectrum irradiation with no eosin Y present, they got a fifty-four percent yield. The dye was not the catalyst. The lamp was doing the work. That brings us to quantum yield — the decisive measurement. Quantum yield is defined as the rate of substrate conversion divided by the absorbed photon flux. In plain terms: how many substrate molecules are converted for each photon absorbed. A simple photocatalytic cycle, where one photon drives one catalytic turnover, gives a value between zero and one. A value greater than one means something else is happening — a radical chain, where a single initiation event triggers a cascade of subsequent reactions, each propagating without another photon. Majek and colleagues measured quantum yields for a suite of eosin Y reactions using potassium Reineckate actinometry — a calibrated chemical system for measuring photon flux — with a 3.8-watt green light-emitting diode at 535 nanometers. The values spanned nearly two orders of magnitude, from 0.075 to 4.7. Let that range sit for a moment. The lowest value, 0.075, means fewer than one in ten photons results in a product molecule — an inefficient but genuinely photocatalytic process. The highest value, 4.7, means nearly five product molecules form for every photon absorbed. That reaction is not being driven by light. Light is striking the match; the chain reaction does the rest. The reactions that exceeded quantum yields of one were the heterobiaryl coupling and the Heck-type olefination with styrene. Both are sustained primarily by radical-chain propagation. The phenanthrene synthesis came in at 0.35, and the photoborylation at 0.60 — both consistent with a genuine photocatalytic pathway, though the borylation had already revealed its susceptibility to direct photolysis under broad-spectrum sources. The benzothiophene formation and photothiolation returned low values attributable to non-productive pathways. What the paper also showed is that optimizing conditions around the correct mechanism can improve yields. In dimethyl sulfoxide with 535-nanometer irradiation — conditions that ensure the active anionic forms of eosin Y are present and that exclude direct photolysis — the phenanthrene synthesis gave a seventy-one percent yield, compared to fifty-three percent under the unspecified literature conditions. Knowing what is actually happening lets you design the experiment correctly. The implications extend well past eosin Y. Majek, Filace, and Jacobi von Wangelin argue that the field has a reproducibility problem rooted in omitted experimental detail. The lamp spectrum, the solvent identity and purity, the concentration of catalyst and substrate, the pH or acid-base additives, and the exact commercial form of the dye — these are not minor footnotes. They can individually switch a reaction between three entirely different mechanisms: true photocatalysis, radical-chain propagation, and direct substrate photolysis. A reaction reported as "eosin Y-catalyzed" in one lab may not be reproducible in another simply because the lamp is different or the solvent came from a different supplier. Their prescription is direct: report all of these details, and measure the quantum yield. Not as an optional validation, but as a diagnostic. A quantum yield above one tells you a chain is running. A quantum yield well below one flags inefficiency and potential non-productive pathways. And a control experiment without the dye, under the same irradiation, tells you whether the substrate is doing the work itself. The paper is, at its core, a methodological argument delivered through careful experiment. The spectroscopic evidence established which forms of eosin Y are active and under what conditions. The synthetic evidence showed that direct photolysis can account for substantial yield. The quantum yield measurements distinguished chain from catalytic mechanisms across a range of reactions. Together, they make a case that is hard to argue with: before you call something photocatalysis, you need to measure what the light is actually doing. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.