Photoredox Catalysis

How a photon becomes a single electron

The catalyst absorbs a visible photon and is promoted to a long-lived triplet excited state — for Ru(bpy)₃²⁺ a metal-to-ligand charge-transfer (MLCT) state with a lifetime near 1.1 microseconds, long enough to meet a substrate before relaxing. That excited state, written [Ru(bpy)₃²⁺]*, is simultaneously a better oxidant and a better reductant than the ground state, because promoting an electron leaves both a low-lying hole (easy to fill = oxidizing) and a high-lying electron (easy to donate = reducing).

From there, two closed catalytic cycles are possible:

• Oxidative quenching: the excited catalyst gives its high-energy electron to an acceptor (e.g. an aryl halide or a diazonium salt), generating a substrate radical and the oxidized catalyst P⁺. A sacrificial reductant then returns P⁺ to the ground state.
• Reductive quenching: the excited catalyst takes an electron from a donor (an amine, a Hantzsch ester, or a carboxylate), producing a radical cation and the reduced catalyst P⁻, which later hands its electron to the substrate.

The distinction matters mechanistically but the net result is the same: a stable, bench-storable molecule is converted into a reactive open-shell radical at room temperature, using nothing but light. Because both quenching pathways regenerate the ground-state catalyst, turnover numbers can reach into the thousands.

Reading redox potentials: the whole ballgame

Choosing a photocatalyst is really about matching excited-state redox potentials to the substrate. The relevant numbers are reported versus the saturated calomel electrode (SCE) in acetonitrile. Ru(bpy)₃²⁺ is a mild reagent: its excited state can only reduce substrates down to about −0.81 V. To reduce a tough aryl bromide you instead reach for fac-Ir(ppy)₃, whose excited state is a powerful reductant at roughly −1.73 V (vs. SCE).

On the oxidizing side, the metal-free organic catalyst 4CzIPN (a carbazolyl dicyanobenzene) reaches about +1.35 V, enough to oxidize many carboxylates and amines while costing a fraction of an iridium complex. A reaction only proceeds if single-electron transfer is thermodynamically feasible — you compare the catalyst's excited-state potential with the substrate's reduction or oxidation potential and look for a favorable (negative) ΔG. This is why the same transformation can succeed with one catalyst and fail entirely with another that differs by only a few hundred millivolts.

Conditions, reagents, and a representative reaction

A typical photoredox reaction needs remarkably little hardware: the substrate, catalyst at 0.5–2 mol%, a degassed polar solvent (acetonitrile, DMF, DMSO, or DMSO/water), a stoichiometric electron shuttle, and a strip of blue LEDs. Reactions run at room temperature over hours, and oxygen is usually excluded because triplet O₂ quenches the excited catalyst.

Common auxiliary reagents include:

• Hantzsch ester and tertiary amines (e.g. Et₃N, iPr₂NEt) as terminal reductants that close reductive-quenching cycles.
• Persulfate (S₂O₈²⁻) or aryl diazonium salts as oxidative-quenching acceptors.
• Radical precursors that fragment after SET: alkyl halides, N-(acyloxy)phthalimide redox-active esters, trifluoroborates, and carboxylic acids.

The canonical example is MacMillan's 2008 organocatalytic–photoredox α-alkylation. A chiral imidazolidinone condenses an aldehyde into an enamine; meanwhile Ru(bpy)₃²⁺ reduces an alkyl bromide to an electron-poor radical. That radical adds to the enamine's β-carbon, the resulting α-amino radical is oxidized back to an iminium, and hydrolysis releases the enantioenriched product — frequently in >90% ee and good yield, with two catalytic cycles turning over in concert.

Scope, selectivity, and limitations

The scope of visible-light photoredox is now enormous: decarboxylative couplings, C–H functionalization, [2+2] and radical cyclizations, aminations, trifluoromethylations, reductive dehalogenations, and photocatalytic polymerizations. Because radicals are only weakly affected by steric bulk and ignore the polar-matching rules that govern ionic chemistry, photoredox often tolerates unprotected alcohols, amines, and heterocycles that would derail a Grignard or a strong base.

Selectivity comes from a mix of factors: the thermodynamics of SET (potential matching, above), radical stability and polarity (nucleophilic vs. electrophilic radicals pair with complementary partners, à la Giese addition), and, for asymmetric variants, a separate chiral catalyst — an organocatalyst, a Lewis-acid, or a chiral nickel complex — because the photocatalyst itself confers no stereocontrol.

The main limitations are practical. Light penetration is poor in concentrated or scattering mixtures, so scale-up historically stalled until continuous-flow photoreactors with thin channels solved the "photon transport" problem. Precious-metal Ir and Ru catalysts are costly and must sometimes be removed to ppm levels for pharmaceutical use, which is a driver behind metal-free 4CzIPN and acridinium catalysts. Oxygen sensitivity, competing back-electron transfer, and over-reduction of products are recurring nuisances.

Metallaphotoredox and drug synthesis

The most industrially consequential advance was metallaphotoredox dual catalysis, in which a photocatalyst feeds radicals or modulates oxidation states of a second transition-metal catalyst — most often nickel. In the 2014 MacMillan–Doyle and Molander decarboxylative and C–H arylations, the photocatalyst generates an alkyl radical (from a carboxylic acid or a C–H bond) that is captured by a Ni center already bearing an aryl group from oxidative addition into an aryl halide; reductive elimination then forges a C(sp³)–C(sp²) bond that is notoriously hard to make by classical cross-coupling.

These reactions have moved into process chemistry. Merck's team used metallaphotoredox and, later, a decarboxylative C–N coupling, to streamline routes to drug candidates, and the approach has been demonstrated on kilogram scale in flow. Photoredox trifluoromethylation and late-stage C–H functionalization also let medicinal chemists diversify complex drug scaffolds at the final stage rather than rebuilding a molecule from scratch. Beyond pharma, the same single-electron logic powers photoinduced ATRP and RAFT polymerizations, controlled radical processes that give designer polymers with light as the on/off switch.

A short history

The photophysics predates the synthetic boom by decades. Ru(bpy)₃²⁺ was studied through the 1970s and 1980s as a light-driven electron-transfer agent, largely aimed at solar water splitting and artificial photosynthesis rather than making bonds. Scattered organic examples appeared — Kellogg used a related idea in the 1970s, and Deronzier and others explored radical reactions — but the chemistry stayed niche.

The turning point was 2008, when David MacMillan, Tehshik Yoon, and Corey Stephenson published back-to-back demonstrations that visible-light photoredox could do practical synthesis. The following decade saw an explosion: Stephenson's radical fragment couplings, Yoon's dual Lewis-acid/photoredox asymmetric cycloadditions, the introduction of cheap organic photocatalysts like Nicewicz's acridiniums and 4CzIPN, and finally the metallaphotoredox merger with nickel catalysis around 2014. In under fifteen years, photoredox catalysis went from a curiosity to a standard chapter in the synthetic toolbox.

Common photocatalysts and their excited-state redox potentials (vs. SCE in MeCN)

Catalyst	Type	E(P*/P−) reduction	E(P+/P*) oxidation	Typical role
Ru(bpy)<sub>3</sub><sup>2+</sup>	Ru polypyridyl	+0.77 V	&minus;0.81 V	Mild oxidant/reductant
Ir(ppy)<sub>3</sub>	Ir cyclometalated	+0.31 V	&minus;1.73 V	Strong excited-state reductant
Ir(ppy)<sub>2</sub>(dtbbpy)<sup>+</sup>	Ir cyclometalated	+0.66 V	&minus;0.96 V	Balanced oxidant/reductant
4CzIPN	Organic (metal-free)	+1.35 V	&minus;1.04 V	Cheap, strong oxidant
Eosin Y	Organic dye	+0.83 V	&minus;1.11 V	Green-light metal-free