Frequency-Resolved Purcell Effect

• Frequency-resolved Purcell effect is the frequency-dependent modification of spontaneous decay by structured environments, accounting for detuning, polarization, and modal resonances.
• It is characterized by spectral selectivity, where interference, local density of states, and input impedance define distinct emission pathways across photonic and microwave systems.
• Applications span nanophotonics, metamaterials, circuit QED, and acoustic systems, enabling tailored emission control without relying solely on high local-field hot spots.

Searching arXiv for the provided topic papers to verify identifiers and publication context. The frequency-resolved Purcell effect is the spectrally selective modification of spontaneous decay or radiated power by a structured electromagnetic or acoustic environment. In the literature surveyed here, it is not treated as a single scalar enhancement factor, but as a function of transition frequency, detuning, polarization, emitter position, symmetry, linewidth, and modal structure. This viewpoint appears in Green-tensor and input-impedance formulations, in periodic photonic systems with band edges and Van Hove singularities, in microwave and superconducting circuits with engineered dissipation, and in spin, magnetic-dipole, many-body, chiral, and phononic platforms (Krasnok et al., 2015, Krasnok et al., 2016, 2209.13670, Joe et al., 13 Mar 2025).

1. Definitions and spectral formulations

A general formulation is to regard the Purcell effect as the frequency dependence of the additional resistive loading or local density of states seen by a small emitter. In the antenna model, the basic weak-coupling relations are

$$F(\omega)=\frac{\Gamma(\omega)}{\Gamma_0(\omega)} =\frac{P(\omega)}{P_0(\omega)} =\frac{R_{\rm rad}(\omega)}{R_{0,\rm rad}(\omega)} =\frac{\Re Z_{\rm in}(\omega)}{\Re Z_{0,\rm in}(\omega)} =\frac{\Im G(\mathbf r_d,\mathbf r_d,\omega)}{\Im G_0(\mathbf r_d,\mathbf r_d,\omega)},$$

with the scattered-field form

$$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$

In lossless environments, $$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$; in general,

$$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$

These equivalences are the core reason the same topic appears in nanophotonics, antenna engineering, and microwave experiments (Krasnok et al., 2015).

A complementary reciprocity-based view writes the enhancement as a radiated-power ratio and introduces an overlap function $f(r)=E_{d\ast}(r)\cdot E(r)$, where $E_{d\ast}(r)$ is the field of the time-reversed dipole in free space and $E(r)$ is the field of the dipole embedded in the dielectric structure. In that formulation, constructive interference of the back-scattered field raises the LDOS and enhances coupling, while destructive interference suppresses it. A practical design rule then follows from the phase of this overlap: regions with negative phase are removed because “a substrate would suppress the local density of states (LDOS) at the emitter” there (2209.13670).

This spectral perspective also makes the Purcell effect independent of any single $Q/V$ cavity estimate. In open, lossy, or multimode systems, the relevant quantity is the full $\omega$-dependent Green tensor or the equivalent input impedance. That is why later work can discuss band-edge resonances, detuning-dependent spin relaxation, or frequency-selective impedance matching within a single conceptual framework (Krasnok et al., 2015).

2. Collective photonic resonances, band edges, and modal spectra

A particularly clear photonic realization is the all-dielectric nanoparticle chain. For a finite chain of $\varepsilon=16$ spheres with radius $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$0 nm and period $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$1 nm, an electric dipole placed at the chain center produces a sharply frequency-dependent Purcell spectrum. The peak grows from about $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$2 for a dimer to about $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$3 for a $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$4-particle chain, and the strongest narrow resonance occurs near $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$5 THz for a transverse dipole. The infinite-chain analysis shows that the large enhancement is associated with the TM band edge, where the group velocity tends to zero and the 1D density of states develops a Van Hove singularity. The effect is strongly symmetry selective: the transverse dipole couples to the dark antiphase band-edge mode, whereas the longitudinal dipole does not show a comparable resonance (Krasnok et al., 2016).

A crucial implication of that work is that a large Purcell factor need not be tied to plasmonic hot spots. The dielectric chain does not provide strong local electric-field enhancement, yet it supports a strong Purcell effect because the decisive ingredients are the density of states and symmetry-allowed coupling to a weakly radiating collective mode. The narrow linewidth is therefore better interpreted as finite-size broadening of a slow-light band-edge resonance than as the $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$6-factor of a localized cavity (Krasnok et al., 2016).

Disordered photonic crystals show a different kind of spectral restructuring. In a 1D disordered stack, moderate disorder preserves enhancement near the photonic band-gap edge through a modified edge state, whereas stronger disorder produces localized in-gap states with $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$7. For the parameters studied, the threshold disorder is approximately $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$8. The main spectral maximum thus migrates from the band edge into the band-gap interior as disorder increases, and the strongest enhancement becomes associated with narrow, high-$$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$9 localized resonances rather than the original edge state (Morozov et al., 2018).

Microsphere cavities provide a mode-resolved Green-function version of the same idea. There the full Green function is expanded in complex-frequency poles $$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$0, so each whispering-gallery mode contributes its own resonance frequency, linewidth, leakage, and quality factor. The spontaneous-emission enhancement is written as a frequency-dependent Purcell factor $$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$1, and the observed photoluminescence is decomposed into leaky-mode background, Purcell-enhanced spontaneous emission, and stimulated emission. In ZnO microspheres, the main WGM peaks are attributed to Purcell-enhanced spontaneous and stimulated emission, whereas Mie-theory spontaneous emission without Purcell enhancement is dominated by leaky modes; for sphere diameters larger than $$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$2, higher-order modes dominate the photoluminescence spectrum (Chien et al., 2019).

3. Metamaterials, ENZ regimes, percolation, and chirality

Frequency-resolved Purcell physics in metamaterials is not exhausted by a single broadband-LDOS narrative. In nanorod hyperbolic metamaterial resonators, the dominant enhancement arises from discrete Fabry–Perot-like TM cavity modes of the finite array rather than from a featureless continuum. For a square $$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$3 resonator with $$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$4 nm, the strongest peak is associated with TM$$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$5 near $$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$6 nm; Purcell factors reach several hundred and are $$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$7 times larger than those at the epsilon-near-zero transition frequencies. By contrast, the ENZ regime near $$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$8 nm gives only moderate enhancement, typically below $$F(\omega)=P_{\rm rad}(\omega)/P_{0,\rm rad}(\omega)$$9, and is almost insensitive to geometry (Slobozhanyuk et al., 2015).

A different broadband regime appears at metal–insulator percolation. For a two-level emitter above a composite medium, the maximum Purcell enhancement occurs at the percolation threshold $$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$0 over a broad range of transition frequencies. In the near field, around $$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$1 nm, total decay rates can reach $$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$2 times the free-space value, and the enhancement can exceed the homogeneous-metal case by more than two orders of magnitude. The spectral behavior is broad rather than sharply resonant: the key control parameter is filling fraction, while the physical channel is predominantly evanescent and absorption dominated (Szilard et al., 2016).

Near ENZ, nonlinearity can make the Purcell effect explicitly switchable in frequency. In an Ag/TiO$$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$3 multilayer HMM, a Kerr-induced topological transition changes propagating and evanescent transmission, and with a $$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$4-polarized control pulse at $$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$5 nm the Purcell factor changes from $$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$6 to around $$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$7. Away from ENZ, at $$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$8 nm, the change is less than $$F(\omega)=\frac{P_{\rm rad}(\omega)+P_{\rm nonrad}(\omega)}{P_{0,\rm rad}(\omega)} \equiv F_{\rm rad}(\omega)+F_{\rm nonrad}(\omega).$$9. The effect is therefore spectrally localized around the ENZ/topological-transition region and can be switched on sub-picosecond timescales, with $f(r)=E_{d\ast}(r)\cdot E(r)$0 fs sufficient for the nonlinear response to approach steady state (Jahani et al., 2018).

Chiral environments add a distinct spectral contribution. In macroscopic QED, the total decay rate is $f(r)=E_{d\ast}(r)\cdot E(r)$1, with

$f(r)=E_{d\ast}(r)\cdot E(r)$2

and

$f(r)=E_{d\ast}(r)\cdot E(r)$3

For a homogeneous chiral bulk,

$f(r)=E_{d\ast}(r)\cdot E(r)$4

so the chiral correction scales with transition frequency, medium chirality $f(r)=E_{d\ast}(r)\cdot E(r)$5, and optical rotatory strength $f(r)=E_{d\ast}(r)\cdot E(r)$6. The paper concludes that the chiral effect is greatest for large transition frequencies, molecules with large optical rotatory strength, media with strong cross-susceptibility, and, for a half space, short molecule–interface distances (Rapp et al., 2024).

4. Antenna, microwave, circuit-QED, and hybrid resonator implementations

In microwave systems, the frequency-resolved Purcell effect often appears directly as impedance matching and radiative loading. A striking example is a $f(r)=E_{d\ast}(r)\cdot E(r)$7 cm grounded quarter-wavelength emitter surrounded by a structured dielectric hemisphere of radius $f(r)=E_{d\ast}(r)\cdot E(r)$8 cm with $f(r)=E_{d\ast}(r)\cdot E(r)$9 and $E_{d\ast}(r)$0. The optimized structure supports two omnidirectional radiation modes at $E_{d\ast}(r)$1 GHz and $E_{d\ast}(r)$2 GHz. At $E_{d\ast}(r)$3 GHz the radiation efficiency is $E_{d\ast}(r)$4 and the quoted Purcell factor is $E_{d\ast}(r)$5; at $E_{d\ast}(r)$6 GHz the efficiency is $E_{d\ast}(r)$7 and the Purcell factor is $E_{d\ast}(r)$8. The measured $E_{d\ast}(r)$9-factors are $E(r)$0 and $E(r)$1, respectively. Here the enhancement is explicitly frequency selective and is interpreted as Purcell-enhanced radiative loading that simultaneously improves impedance matching and reduces reflected power (2209.13670).

The same logic underlies superconducting Purcell filters. In a bandpass-filter architecture, the readout resonator acquires a frequency-dependent effective linewidth

$E(r)$2

so the measurement bandwidth is governed by $E(r)$3 while qubit decay is governed by $E(r)$4. The filtered Purcell rate is then

$E(r)$5

and the suppression factor is $E(r)$6. For $E(r)$7 GHz, $E(r)$8 GHz, $E(r)$9 GHz, and $Q/V$0 MHz, the paper gives $Q/V$1 ns, $Q/V$2, and $Q/V$3, corresponding to a suppression factor of about $Q/V$4, with about $Q/V$5 possible for adjusted parameters (Sete et al., 2015).

Drive can itself reshape the Purcell spectrum in circuit QED. In a driven Jaynes–Cummings ladder, the relevant transitions move to dressed frequencies $Q/V$6, with $Q/V$7. As the cavity photon number increases, these sidebands move farther from the cavity passband and the Purcell rate decreases. In the dispersive regime,

$Q/V$8

and for $Q/V$9 the rate scales roughly as $\omega$0 (Sete et al., 2014).

A more spatially resolved circuit formulation is the “waves-in-space Purcell effect.” In a $\omega$1 chip-in-tube readout geometry, relocating the readout port changes the qubit lifetime from approximately $\omega$2 at a WISPE location to approximately $\omega$3 at an anti-WISPE location, while inferred Purcell decay times through the protected readout port reach $\omega$4 ms and $\omega$5 ms for two qubits. The central claim is that Purcell decay depends not only on detuning and linewidth but also on the real-space overlap of qubit fields, cavity fields, and dissipation ports; for qubit frequencies below the readout mode this is stated to be distinct from the multimode Purcell effect (Patel et al., 14 Mar 2025).

A related hybrid-resonator example appears in photon–magnon systems. In a planar HRR–YIG structure, increasing the YIG damping $\omega$6 from $\omega$7 to $\omega$8 reduces the extracted coupling $\omega$9 from $\varepsilon=16$0 MHz to $\varepsilon=16$1 MHz, increases the magnon linewidth $\varepsilon=16$2 to $\varepsilon=16$3 MHz, and broadens the photon linewidth $\varepsilon=16$4 from $\varepsilon=16$5 MHz to $\varepsilon=16$6 MHz. The split resonances nearly coalesce, and the system is interpreted as entering a Purcell regime in which lossy magnons enhance photon decay (Verma et al., 8 Jan 2025).

5. Spin, many-body, magnetic-dipole, and acoustic variants

Spin-resonator systems make the phrase “frequency-resolved Purcell effect” literal. In pulsed ESR, when Purcell relaxation dominates,

$\varepsilon=16$7

so each spin packet has its own relaxation constant. In the broad-line regime $\varepsilon=16$8, the Fourier component of the echo obeys

$\varepsilon=16$9

which means different spectral components of the same echo recover at different rates. Experimentally, donor spins in silicon measured by superconducting micro-resonators show that the $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$00 component relaxes faster than the $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$01 component and that repetition time reshapes the echo through this same spectral selectivity (Ranjan et al., 2019).

In many-body cavity QED, the frequency-resolved Purcell effect becomes transition selective. For strongly interacting emitters, the condition

$$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$02

allows a narrow cavity to enhance one transition while quenching subsequent off-resonant decays. For $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$03, tuning $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$04 selects $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$05 and suppresses $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$06, producing dissipative stabilization of the symmetric entangled state. For general $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$07, tuning $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$08 selects $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$09. The paper reports fidelities around $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$10 for $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$11 under favorable conditions and fidelities exceeding $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$12 for $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$13 in suitable regions (Vivas-Viaña et al., 2023).

Optical magnetic-dipole Purcell enhancement shows that the same spectral logic applies to magnetic LDOS. A single $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$14 ion in centrosymmetric MgO, coupled to a silicon photonic-crystal nanocavity, exhibits a nearly pure magnetic-dipole transition at $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$15 nm. With $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$16, $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$17 GHz, and a single-ion linewidth as narrow as $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$18 kHz, the bulk lifetime $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$19 ms is shortened to $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$20, yielding a magnetic Purcell factor $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$21. The extracted coupling is $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$22 MHz, corresponding to an inferred local single-photon magnetic field of $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$23 G (Horvath et al., 2023).

The acoustic case extends the same formal structure to phonons. A single SiV center in a diamond optomechanical crystal shows spin relaxation enhancement when its Zeeman-tuned transition crosses a localized $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$24 GHz mechanical breathing mode. The measured spin decay rate rises from approximately $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$25 kHz off resonance to approximately $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$26 kHz on resonance, a ten-fold enhancement, with a best observed Purcell linewidth of about $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$27 MHz after ALD tuning. The single-mode fit is

$$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$28

with extracted $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$29 kHz, and the broadband response is modeled as

$$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$30

allowing the SiV to act as a local probe of the phonon spectrum up to about $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$31 GHz (Joe et al., 13 Mar 2025).

6. Conceptual distinctions, common misconceptions, and limitations

A recurring misconception is that a large Purcell factor requires a large local-field hot spot. The all-dielectric chain shows the opposite: high enhancement can arise “without high local-field enhancement” when the emitter couples to a dark collective band-edge mode with large LDOS and appropriate symmetry (Krasnok et al., 2016). Conversely, percolation-enhanced decay in composite media is broadband in frequency yet dominated by evanescent, absorption-related channels rather than by a single narrow cavity resonance (Szilard et al., 2016). This suggests that “frequency-resolved Purcell effect” can refer either to sharp spectral peaks or to broad but still structured spectral dependence, depending on the environment.

Another distinction is between broadband and resonant viewpoints. Infinite-medium hyperbolic metamaterials motivated a broadband LDOS picture, but finite resonators show that the strongest enhancement can instead be set by discrete TM Fabry–Perot modes (Slobozhanyuk et al., 2015). ENZ switching is even narrower: the large change in Purcell factor appears only around the topological-transition wavelength and is weak away from it (Jahani et al., 2018). In circuit architectures, the same tension appears as a contrast between single-linewidth Purcell formulas and frequency-dependent admittance or port-overlap descriptions (Sete et al., 2015, Patel et al., 14 Mar 2025).

The principal caveat in most formulations is weak coupling. The antenna model explicitly assumes a fixed emitter dipole moment, electrically small emitters, and linear passive environments; it warns that very large $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$32 can invalidate the fixed-dipole approximation and require a strong-coupling treatment with mode splitting (Krasnok et al., 2015). Finite-size regularization is equally important: the Van Hove singularity belongs to the ideal infinite chain, while finite chains exhibit only large but finite resonances, broadened further by loss and leakage (Krasnok et al., 2016). ESR analyses assume $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$33 and low cooperativity $$F \equiv \frac{\gamma}{\gamma_0} =1+\frac{6\pi\varepsilon_0}{|\mathbf d_1|^2}\frac{1}{q^3} \Im\!\left[\mathbf d_1^\ast\!\cdot\! \mathbf E_{\rm s}(\mathbf r_d)\right].$$34, while the acoustic SiV experiment shows that the observed Purcell linewidth can be tens or hundreds of times broader than the intrinsic mechanical linewidth because of gas loading or ALD-related damping (Ranjan et al., 2019, Joe et al., 13 Mar 2025).

Taken together, these results define the frequency-resolved Purcell effect as a general spectral property of emitter–environment coupling. It may be governed by Green-tensor poles, band-edge singularities, input-impedance spectra, cavity-filtered transition manifolds, crossed chiral reflection coefficients, or discrete phonon modes. What remains invariant is the central principle: the spontaneous decay or radiation rate is controlled by the environment’s mode structure at the emitter’s transition frequency, weighted by symmetry, position, polarization, and coupling to the relevant channel (Krasnok et al., 2015, Vivas-Viaña et al., 2023).