Photon Catalysis: Quantum and Photocatalytic Functions

• Photon catalysis is a dual-concept technique that employs controlled photon interactions to induce non-Gaussian state transformations and optimize photocatalytic processes.
• Quantum optical implementations use interference with ancillary Fock states and beam splitters to generate nonclassical states for quantum information and metrology.
• In heterogeneous catalysis, engineered photon management enables spectral matching and near-field control to enhance reaction rates and selectivity on nanostructured catalysts.

Searching arXiv for recent and foundational papers on “photon catalysis” across quantum optics and photocatalysis. Photon catalysis is a technically overloaded term spanning two research traditions. In quantum optics and continuous-variable quantum information, it denotes a heralded non-Gaussian operation in which a target mode interferes with an ancillary Fock state on a beam splitter and the ancilla output is projected back onto the same photon number, so that the ancilla conditions the transformation without being consumed in the subtraction/addition sense (Kumar et al., 2024). In heterogeneous catalysis and photocatalysis, the term is used more broadly for catalyst architectures in which photons are deliberately harvested, localized, spectrally matched, or dynamically scheduled so that light management itself becomes a catalytic design variable, as in plasmonic metal–semiconductor heterostructures, photonic crystals, metasurfaces, and light-promoted kinetic resonance (Collins et al., 2019). These usages are not interchangeable, but both treat photons as structured resources rather than as undifferentiated energy input.

1. Quantum-optical definition and operator structure

In the quantum-optical literature, photon catalysis is a conditional linear-optical map. A target state interferes with an ancilla $|m\rangle$ on a beam splitter of transmissivity $T_C$, and a photon-number-resolving detector heralds success when exactly $m$ photons are detected in the ancilla output. The standard interpretation is that the ancilla starts and ends in the same photon number, while the signal mode undergoes a non-Gaussian transformation (Kumar et al., 2024).

This operation admits compact operator forms. For $n$-photon quantum catalysis on a target mode $b$, one widely used expression is

$$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$

which makes explicit that catalysis is a Laguerre-polynomial function of the photon-number operator followed by attenuation (Guo et al., 2018). A closely related single-mode derivation shows that multiphoton catalysis acts as

$$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$

again highlighting a number-dependent filter rather than a simple creation or annihilation monomial (Hu et al., 2015).

A recurrent misconception is that photon catalysis is merely photon addition or subtraction in another guise. The cited operator identities show otherwise. Photon subtraction implements an operator proportional to $b$; photon addition implements $b^\dagger$; photon catalysis implements a conditional polynomial filter in $b^\dagger b$, realized by interference plus heralding (Hu et al., 2015). This distinction underlies its role in non-Gaussian state engineering.

2. State engineering and non-Gaussian resource generation

The earliest state-engineering applications used photon catalysis to reshape coherent and squeezed inputs into analytically tractable non-Gaussian states. For coherent-state input, single-photon catalysis produces the single-photon catalyzed coherent state, and beam-splitter and parametric-amplifier implementations were shown to generate exactly the same conditional state after the parameter identification $T_C$0 (Xu et al., 2015). The resulting state interpolates continuously between $T_C$1 at $T_C$2 and $T_C$3 at $T_C$4, and exhibits sub-Poissonian statistics, antibunching, quadrature squeezing, and Wigner-function negativity (Xu et al., 2015).

For coherent-state input with multiphoton catalysis, the output becomes a Laguerre polynomial excited coherent state,

$T_C$5

with nonclassicality controlled by the coherent amplitude $T_C$6, catalysis number $T_C$7, and beam-splitter imbalance (Hu et al., 2015). The same work showed that the optimal squeezing for single-photon catalysis is about $T_C$8 dB below shot noise, while higher $T_C$9 enhances squeezing and Wigner negativity over suitable parameter windows (Hu et al., 2015).

Photon catalysis was later extended to exact and approximate preparation of more structured bosonic resources. Interference of coherent states with single-photon ancillae followed by photon-number-resolved detection was shown to generate exact displaced single-photon states, $m$0-symmetric superpositions of squeezed vacuum, squeezed cat states as the $m$1 case, and, through breeding, states approximating square and hexagonal Gottesman–Kitaev–Preskill codewords (Eaton et al., 2019). In a more recent resource-theoretic formulation, two-mode photon catalysis between low-number Fock states and squeezed vacuum was analyzed with stellar rank; the output has stellar rank $m$2, and low-resource cases such as $m$3, $m$4, and $m$5 were identified as instances where photon catalysis can be provably optimal for approximating squeezed cat states at fixed non-Gaussian complexity (Nauth et al., 2 Jul 2026).

The same conceptual line culminated in a multimode synthesis theorem: any core state containing at most $m$6 photons in $m$7 modes, whose homogenization has Waring rank $m$8, can be prepared exactly from vacuum using multiport interferometers, photon additions, one projection conditioning on $m$9 photons, one displacement, and one vacuum projection (Aralov et al., 25 Jul 2025). In that construction, extra photons are injected into a larger interferometric network and later retrieved by measurement; the authors explicitly identify this injected-then-retrieved mechanism as photon catalysis (Aralov et al., 25 Jul 2025).

3. Entanglement, quantum communication, and metrology

Photon catalysis has also been studied as an entanglement-enhancing operation on squeezed resources. Local quantum-optical catalysis on both arms of a two-mode squeezed vacuum state yields a locally quantum-catalyzed TMSVS of the form

$n$0

and can improve entanglement entropy, second-order Einstein–Podolsky–Rosen correlation, and coherent-state teleportation fidelity, but only in the regime of small input squeezing and low beam-splitter transmissivities (Xu, 2015). The same restricted-regime character recurs in later applications.

In continuous-variable quantum key distribution, one influential proposal treated zero- and single-photon quantum catalysis as a practical non-Gaussian improvement over both Gaussian TMSV protocols and photon subtraction, with bilateral symmetric catalysis outperforming single-sided catalysis and tolerable excess noise reaching $n$1 at $n$2 km for zero-photon catalysis (Guo et al., 2018). A later re-examination derived the Wigner characteristic function of the $n$3-photon catalysed two-mode squeezed coherent state and showed that the $n$4 case is exactly Gaussian, with covariance matrix independent of displacement $n$5 (Kumar et al., 2024). After global optimization over variance, transmissivity, and displacement, that study found that zero- and single-photon catalysis improve the maximum transmission distance by less than $n$6 km and that the optimal displacement is $n$7, so displacement offers no benefit in improving CV-MDI-QKD (Kumar et al., 2024). The tension between these papers is substantive rather than terminological: later analysis attributes earlier apparent gains largely to non-optimal high-variance operating points.

In quantum metrology, multiphoton catalysis has been embedded into multimode entangled squeezed-vacuum architectures. The resulting multi-mode entangled catalysis squeezed vacuum state was proposed as a probe for simultaneous multiphase estimation, with the reported quantum Cramér–Rao bound improving as the catalytic photon number increases or the catalysis beam-splitter transmissivity is varied, and remaining lower than that of the ideal entangled squeezed vacuum state even under photon loss (Zhang et al., 2022). This suggests that photon catalysis can act as a tunable non-Gaussian deformation of Gaussian probe families, although the preparation remains probabilistic and experimentally demanding.

4. Photon catalysis in plasmonic and photonic heterogeneous catalysis

In heterogeneous catalysis, photon catalysis refers to architectures in which optical absorption, charge separation, and photon transport are co-engineered. A representative example is the inverse-opal Au–semiconducting-metal-oxide platform, where Au nanoparticle-functionalized V$n$8O$n$9 and TiO$b$0 inverse opals integrate semiconductor band-gap absorption, Au localized surface plasmon resonance, and photonic-crystal control of photon propagation (Collins et al., 2019). The central mechanistic lesson is conditional spectral matching: optimal catalysis occurs when the excitation wavelength overlaps the semiconductor band edge, the Au LSPR, and the slow-light region near the photonic band edge. In Au–V$b$1O$b$2 inverse opals, $b$3 increases from $b$4 in the dark to $b$5 under $b$6 nm light, whereas Au–TiO$b$7 benefits more from broadband illumination because TiO$b$8 is not visible-band-gap active (Collins et al., 2019).

A closely related plasmonic–photonic strategy hybridizes Ag nanoparticle plasmons with a guided-mode resonance of a photonic-crystal slab, creating narrowband, tunable absorption that is much stronger than the uncoupled nanoparticle response (Huang et al., 2019). For Ag nanocuboids coupled to a photonic-crystal slab and illuminated at $b$9 nm, the simulated nanoparticle absorption cross section increases by about $$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$0, the measured $$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$1 peak is about $$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$2 larger than the uncoupled LSPR signal, and the reduction of 4-nitrothiophenol to 4-aminothiophenol is strongly accelerated while thermal modeling predicts only about $$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$3 temperature rise (Huang et al., 2019). In this context, photon catalysis is inseparable from nanophotonic impedance matching: the photonic resonance concentrates free-space light into the plasmonic absorber, and the plasmonic absorber converts that energy into hot carriers at the catalytic surface.

These studies reframe photocatalysis from materials composition alone toward photon management. Pore periodicity, photonic band gaps, localized near fields, and interfacial band alignment all become catalytic variables because they determine where, at what energy, and for how long photons interact with reactive matter (Collins et al., 2019).

5. Selectivity control and dynamic light modulation

Recent work has extended photon catalysis in chemistry from rate enhancement to selectivity control. In visible-light oxidative coupling of methane with nitrous oxide, AuPd$$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$4/TiO$$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$5 alloys separate optical and chemical functions: Au provides a broad LSPR at $$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$6 nm, while Pd supplies sites for C–H activation and C–C coupling (Lee et al., 20 Apr 2026). The optimal AuPd$$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$7/TiO$$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$8 catalyst, under $$\hat{O}_n \equiv {}_C\!\langle n \rvert B(T_2)\lvert n\rangle_C = :L_n\!\left(\frac{1-T_2}{T_2}b^\dagger b\right):\left(\sqrt{T_2}\right)^{\,b^\dagger b+n},$$9 nm illumination at $$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$0 and feed CH$$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$1:N$$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$2O:Ar $$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$3 sccm, produces C$$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$4H$$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$5, C$$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$6H$$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$7, C$$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$8H$$\hat B_m=\cos^m\theta\, :\!L_m(b^\dagger b\tan^2\theta)\!:\,e^{b^\dagger b\ln\cos\theta},$$9, and C$b$0H$b$1 with approximately $b$2 selectivity to C$b$3–C$b$4 hydrocarbons, while excited-state calculations reduce the decisive C–C coupling barrier from $b$5 eV to about $b$6 eV (Lee et al., 20 Apr 2026). The mechanistic emphasis is on illumination-driven redistribution of hydroxyl intermediates at the AuPd–TiO$b$7 interface, shifting the hydrophilic center and suppressing overoxidation.

A different route to dynamic control uses anisotropic plasmonic metasurfaces rather than catalyst reformulation. Elliptical Au–TiO$b$8 nanopillars support polarization-dependent resonances, so the same sample can be tuned in situ by rotating the incident linear polarization (Lyu et al., 26 Sep 2025). At $b$9 nm, measured absorption changes from $b^\dagger$0 under TM to $b^\dagger$1 under TE polarization, and the methylene-blue product-yield metric derived from the $b^\dagger$2 Raman peak changes from $b^\dagger$3 to $b^\dagger$4 after $b^\dagger$5 s, a factor of about $b^\dagger$6 (Lyu et al., 26 Sep 2025). This establishes a direct absorption–yield correlation in a single device and suggests that resonance tuning can become a real-time control knob for resonance-driven reactions.

A more abstract kinetic formulation appears in catalytic resonance theory, which models light as a discrete stream of photon events promoting a specific elementary step such as product desorption,

$b^\dagger$7

Microkinetic and kinetic Monte Carlo simulations identify three regimes—thermal desorption control, an intermediate photon-arrival-controlled regime with near-zero Arrhenius slope, and surface-reaction control—and locate the maximum photocatalytic rate at the resonance condition $b^\dagger$8, where the per-site photon arrival frequency matches the intrinsic surface reaction rate constant (Dauenhauer, 18 Nov 2025). This usage broadens photon catalysis from optical materials design to a general theory of dynamic, discrete photonic forcing.

6. Morphology, measurement, and contested mechanisms

A recurring theme in photocatalytic photon catalysis is that the catalytically relevant photon field is not identical to the nominal illumination field. For bicontinuous TiO$b^\dagger$9–silica spinodal aerogels, GPU Monte Carlo transport on 3D Cahn–Hilliard masks shows that the catalytic solid phase receives $b^\dagger b$0 more photons than the volume average at porosity $b^\dagger b$1, rising to $b^\dagger b$2 at $b^\dagger b$3, because connected pore channels support quasi-ballistic “photon channelling” (Vallée, 15 Apr 2026). The extracted intrinsic kinetic descriptor differs by $b^\dagger b$4 between the 3D Monte Carlo solid-phase estimate and a diffusion-model estimate, and roughly $b^\dagger b$5 of the total $b^\dagger b$6 discrepancy is intrinsic to the bicontinuous structure rather than to diffusion-model error alone (Vallée, 15 Apr 2026). In this literature, photon catalysis includes morphology-controlled redistribution of fluence toward the active phase.

The same concern with misassigned mechanisms appears in experimental methodology. The open-source Catalight reactor package was demonstrated on plasmonic acetylene hydrogenation over Au$b^\dagger b$7Pd$b^\dagger b$8/Al$b^\dagger b$9O$T_C$00, where automated sweeps over temperature, laser power, and wavelength showed an increased apparent activation barrier upon light excitation, unchanged reaction orders of about $T_C$01 in H$T_C$02 and $T_C$03 in C$T_C$04H$T_C$05, and no noticeable increase in ethylene selectivity (Bourgeois et al., 10 Apr 2025). Analysis of laser-induced thermal inhomogeneity indicated that a partial photothermal effect combined with a photochemical or hot-electron-driven mechanism best explained the observations (Bourgeois et al., 10 Apr 2025). This suggests that automated multidimensional control is not merely a convenience; it is necessary for separating rate enhancement, selectivity shifts, and thermal artifacts.

The strongest critique of nonthermal interpretation comes from a thermal reinterpretation of plasmonic photocatalysis, which argues that many reported hot-electron effects are explainable by illumination-induced heating. The paper combines the Arrhenius law,

$T_C$06

with a light-induced temperature rise,

$T_C$07

and argues that the efficiency of nonthermal hot-electron generation under typical conditions is only about $T_C$08, so almost all absorbed energy becomes heat (Sivan et al., 2019). Whether one accepts this conclusion universally, the paper establishes a lasting caution: because catalytic rates are exponentially sensitive to temperature, modest temperature mismatch or thermal inhomogeneity can be misread as photon-induced barrier lowering.

Photon catalysis therefore remains a plural concept. In quantum optics it is a precise heralded non-Gaussian operation with Laguerre-polynomial structure, central to state engineering, entanglement manipulation, and finite-resource bosonic computation. In chemical catalysis it is a family of strategies for spectral matching, near-field localization, interfacial charge routing, morphology-controlled fluence redistribution, and dynamic light scheduling. What unifies these otherwise distinct domains is the insistence that photons can be conditioned, filtered, localized, and timed so that they alter transition pathways, accessible states, or effective rates in ways that uniform heating or uncoupled illumination do not automatically provide.