Wait, What? A carbonyl group can absorb light and then break a carbon–carbon bond that was perfectly stable in the dark—or reach several atoms along its own chain and remove a hydrogen atom from the γ-position.
Those two families of excited-state carbonyl chemistry are called Norrish Type I and Norrish Type II reactions. They are valuable because they force us to stop treating an organic molecule as if its ground-state reaction map remains unchanged after photon absorption. Excitation changes the electronic state, the accessible barriers and the reaction pathways.
The direct answer
Norrish Type I is α-cleavage of an excited carbonyl compound: a C–C bond next to the carbonyl carbon breaks homolytically, producing an acyl radical and an alkyl radical.
Norrish Type II begins with intramolecular abstraction of a γ-hydrogen by the excited carbonyl oxygen, commonly through a favourable six-membered geometric arrangement. That creates a 1,4-biradical. The biradical can then fragment, cyclise to a cyclobutanol in a Norrish–Yang pathway, or return toward starting material.
The labels describe mechanistic families, not guaranteed outcomes. Which path dominates depends on excited-state character, molecular conformation, bond strengths, substituents, solvent, phase, spin dynamics and the rates of competing processes.
First change the state of the molecule
In ordinary ground-state organic chemistry, we often ask which bond is polarised, which nucleophile attacks and which transition state is lowest. Photochemistry adds an earlier question: what electronic state has been prepared?
Many ketones and aldehydes can absorb light into an electronically excited state involving promotion from a non-bonding oxygen orbital into a carbonyl π* orbital, commonly described as n→π*. Depending on the molecule and environment, the initially formed singlet excited state may react, relax, fluoresce, undergo internal conversion, or cross to a triplet state through intersystem crossing. Aromatic ketones often show substantial triplet chemistry.
The chemical consequence is profound. Occupying an antibonding orbital changes electron distribution around C=O and can make hydrogen abstraction or radical cleavage accessible on timescales that are irrelevant in the ground state.
Norrish Type I: α-cleavage beside the carbonyl
Consider a ketone written generically as R–C(=O)–R′. In a Type I step, one bond between the carbonyl carbon and an α-carbon undergoes homolytic cleavage:
R–C(=O)–R′* → R–C•=O + •R′
The asterisk indicates an electronically excited precursor, not a conventional ground-state reagent. The immediate products are radical fragments: an acyl radical and a carbon-centred radical. From there, the network can branch. Acyl radicals may lose carbon monoxide under suitable energetic and structural conditions; radical pairs can separate, recombine or undergo secondary reactions.
A common oversimplification is “Type I gives radicals, therefore any radical-derived product proves Type I.” That is not enough. Radicals can arise by multiple photochemical pathways. Mechanistic attribution is stronger when product mapping, isotope effects, time-resolved spectroscopy and energetic analysis all point to the same cleavage sequence.
Which α-bond breaks?
Unsymmetrical ketones can cleave on either side. A useful first hypothesis is that the route giving the more stabilised carbon-centred radical may be favoured, but real selectivity also depends on excited-state potential-energy surfaces, conformational access, cage effects and the fate of the paired acyl radical. “Most stable radical wins” is therefore a starting model, not a universal law.
Norrish Type II: a γ-hydrogen becomes chemically close
Type II chemistry is a beautiful reminder that proximity is three-dimensional rather than simply a matter of bond counting. The excited carbonyl oxygen can abstract a hydrogen from a γ-carbon when the molecule can adopt a suitable geometry. The net elementary description is a 1,5-hydrogen-atom transfer:
excited C=O + γ-C–H → C–OH radical centre + γ-carbon radical centre
The resulting species is a 1,4-biradical: two radical centres separated through the molecular skeleton. The favoured abstraction geometry is often represented by a six-membered arrangement because that allows the excited oxygen and γ-hydrogen to approach with useful orbital alignment.
But “has a γ-hydrogen” does not mean “must undergo Type II”. The required conformation may be rare, a competing excited-state process may be faster, or the relevant C–H geometry may be poor. Conformational probability can therefore become a kinetic variable.
The 1,4-biradical is a branching point, not the final answer
Once the 1,4-biradical forms, several chemically distinct outcomes are possible.
- β-scission / Type II fragmentation: cleavage can produce an alkene plus an enol-derived carbonyl product after tautomerisation.
- Norrish–Yang cyclisation: radical centres combine intramolecularly to form a four-membered ring, giving a cyclobutanol framework.
- Return: reverse hydrogen transfer or other recombination can regenerate starting material.
- Alternative radical chemistry: depending on structure and environment, rearrangement or interception may compete.
This is why a mechanism should not stop at “1,4-biradical formed”. Product selectivity is decided by the rates and conformations of the steps that follow.
Spin matters
If a triplet excited carbonyl produces a triplet 1,4-biradical, immediate formation of a conventional closed-shell σ bond is spin-restricted. The system may need intersystem crossing to a singlet biradical surface before efficient ring closure or some other closed-shell products can form. That does not mean every triplet biradical behaves identically; spin–orbit coupling, radical separation, conformation and substituents all influence the competition.
At undergraduate level this is where photochemistry becomes more than arrow-pushing. The electronic surface on which the nuclei are moving helps determine what products are dynamically accessible.
Thermodynamic permission is not the same as photochemical speed
A reaction product can be thermodynamically lower in energy yet form poorly because the excited molecule rarely reaches the required geometry before deactivation. Conversely, a pathway can dominate because it has fast access from the populated excited state even if another product would be more stable after full relaxation.
So do not reason from product stability alone. Photochemical selectivity is often kinetic and state-specific.
How do chemists know the biradicals are real?
Mechanisms are not accepted because an arrow diagram looks plausible. Norrish chemistry has been investigated through converging evidence:
- Product analysis identifies fragmentation and cyclisation patterns expected from particular intermediates.
- Isotopic substitution can test whether breaking a particular C–H bond is kinetically important.
- Time-resolved absorption and related spectroscopies can observe short-lived excited states and radical intermediates.
- CIDNP and EPR-type evidence can reveal radical-pair behaviour in suitable systems.
- Quantum yields and excited-state lifetimes connect product formation with competing decay channels.
- X-ray crystallography plus solid-state photochemistry can test whether a fixed molecular conformation predicts which γ-hydrogen is geometrically reachable.
- Quantum-chemical and molecular-dynamics calculations can compare potential-energy surfaces and conformational populations, but remain models that must be checked against experiment.
A JACS study of a crystalline ketone, for example, combined transient spectroscopy, solid-state NMR and X-ray diffraction to connect biradical kinetics with conformational motion. The lesson is methodological: a mechanistic claim becomes stronger when independent observables converge.
Observation versus inference
If transient spectroscopy shows a signal that decays in 50 ns, the decay is an observation. Assigning that signal to a particular triplet 1,4-biradical is an inference supported by spectral shape, kinetics, isotope effects, calculations and product correlations. Good mechanistic chemistry keeps those layers separate.
Conformation can rewrite selectivity
Norrish Type II is particularly sensitive to geometry because γ-hydrogen abstraction requires spatial access. The same constitutional molecule can therefore behave differently in a fluid solution and in a crystal. In solution, conformers interconvert; in a crystal, packing can favour one reactive geometry and suppress another. Published Norrish–Yang studies have shown striking changes in regio- and stereoselectivity between solution and different crystal polymorphs.
This is a useful transfer lesson for all chemistry: molecular formula does not specify molecular motion. Reactivity can depend on which conformations are populated on the timescale of reaction.
A learning progression from Secondary to professional chemistry
- Foundation: light can transfer energy to matter, and absorbed energy can change chemical behaviour.
- Secondary: distinguish bond breaking by heat from a photochemical pathway initiated by electronic excitation.
- JC: use homolytic bond cleavage, radicals and carbonyl structure to describe Type I and Type II outcomes qualitatively.
- Undergraduate: track singlet/triplet states, nπ* character, 1,5-H transfer, 1,4-biradicals and product branching.
- Advanced: treat conformation, spin conversion, excited-state surfaces, solvent and non-adiabatic dynamics as mechanistic variables.
- Professional/research: distinguish a proposed pathway from one supported by time-resolved, isotopic, structural and kinetic evidence.
Common misconceptions
- “Type I means any photochemical bond cleavage.” It specifically refers to α-cleavage of an excited carbonyl compound.
- “Type II only needs a γ-hydrogen on the structural formula.” Productive three-dimensional geometry and competing rates matter.
- “The 1,4-biradical always cyclises.” Fragmentation, return and other pathways can compete.
- “The most stable final product must dominate.” Excited-state kinetics and conformational access often control branching.
- “A radical product proves one unique mechanism.” Radical evidence must be connected to the particular precursor state and sequence.
Counterexamples that improve the model
A ketone may contain an abstractable-looking γ-hydrogen yet react weakly by Type II because the required conformation is inaccessible. Another carbonyl may show efficient Type II fragmentation but little Yang cyclisation because its biradical conformations favour cleavage. A crystal can invert or sharpen selectivity relative to solution because packing limits motion. These are not failures of chemistry; they are reminders that the named reaction is a mechanistic framework, not a deterministic keyword.
Transfer checks
Check 1. A ketone has no γ-hydrogen but absorbs into a reactive excited state. Which of the two classical Norrish families is structurally excluded at its first defining step?
Check 2. Two conformers have identical bonds but only one brings a γ-C–H close to the carbonyl oxygen. Why can their Type II rates differ?
Check 3. A product mixture contains an alkene and a cyclobutanol. Propose the common intermediate that can connect those branches, then name one independent experiment that could strengthen the assignment.
Delayed reasoning check. Without notes, draw a generic ketone and mark α, β and γ positions. Then explain Type I in one sentence and Type II in two steps. Finally say why “product stability” alone is insufficient.
How we know: selected evidence
- J. Am. Chem. Soc. 1971, DOI 10.1021/ja00737a023: classic evidence on 1,4-diradical intermediates in Type II photocleavage.
- J. Am. Chem. Soc. 2012, DOI 10.1021/ja2090004: transient kinetics, solid-state NMR and X-ray diffraction linking conformation with Norrish Type II biradical reactivity.
- J. Phys. Chem. A 2004, DOI 10.1021/jp037735l: electronic-structure analysis of competing Type I/II pathways and intersystem crossing in butyrophenone.
- J. Org. Chem. 2023, DOI 10.1021/acs.joc.2c01855: solution-versus-solid-state Norrish–Yang regio- and stereoselectivity.
The quiet return
Norrish chemistry begins with a simple event—light is absorbed—but the useful lesson is deeper. Once the electronic state changes, the molecule inherits a new map of possible motions and bond changes. Type I follows radical cleavage beside the carbonyl. Type II reaches through conformation to a γ-hydrogen and creates a branching biradical. Learning the names is easy; learning to ask which state, which geometry and which evidence is the chemistry.
