Cavity-Mode Dispersion Spectroscopy Overview

• Cavity-mode dispersion spectroscopy is a set of methods that measure shifts in cavity resonances and linewidth broadening to extract refractive and absorptive properties.
• It employs advanced techniques such as optical frequency combs and transient methods to resolve thousands of modes over broad spectral ranges with high precision.
• Applications include greenhouse gas metrology, trace gas detection, and integrated photonics, offering calibration‐free molecular sensing.

Cavity-mode dispersion spectroscopy denotes a class of spectroscopic methods in which the cavity resonances themselves are the primary observables. In the strict Fabry–Perot sense, molecular dispersion is inferred from shifts of cavity mode center frequencies, while absorption is inferred from linewidth broadening; in a broader photonic-resonator usage, the same logic extends to integrated dispersion, local resonance mismatches, antiresonances, and coupled-cavity minibands whose spectral structure encodes refractive, modal, or collective interactions (Johansson et al., 2018, Rutkowski et al., 2017, Cygan et al., 2024, Lian et al., 2015).

1. Definition and scope

In cavity-enhanced molecular spectroscopy, the central dispersive observable is the cavity mode shift produced by the frequency-dependent real part of the intracavity response. In broadband cavity-enhanced complex refractive index spectroscopy (CE-CRIS), the molecularly induced mode shift is written as

$\nu_\varphi(\nu)=\frac{c}{4\pi n}\varphi(\nu),$

while the absorption-induced linewidth contribution is

$\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$

The important point is that the shift and broadening are directly proportional to the resonant real and imaginary parts of the refractive index and are independent of cavity length and mirror reflectivity, which is why the method is described as calibration-free with respect to those cavity parameters (Johansson et al., 2018).

A closely related formulation appears in cavity-enhanced mode dispersion spectroscopy of CO near 1560 nm, where the absorption and dispersion spectra are treated as the real and imaginary parts of the same normalized complex line-shape function, and the measured observable for dispersion is the frequency shift $\Delta \nu$ of cavity longitudinal modes caused by the molecular refractive index (Huang et al., 2024). Heterodyne dispersive cavity ring-down spectroscopy (HCRDS) uses the same physical quantity, but extracts it from the optical frequency carried by the transient ring-down field rather than from scanned mode profiles (Cygan et al., 2024).

A broader usage of the term includes cases where the “dispersion” under study is not the refractive index of an intracavity gas but the frequency organization of cavity modes themselves. In photonic crystal coupled resonator optical waveguides (CROWs), the measured miniband $\omega(k)$ can become intrinsically asymmetric because the single-cavity driven mode profile depends on driving frequency; in integrated microrings, the measured object is the resonance-frequency landscape $D_{\mathrm{int}}$; in coupled-resonator nonlinear optics, the relevant dispersion observable is the local mismatch $2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$ (Lian et al., 2015, Diakonov et al., 2024, Gentry et al., 2014).

2. Resonance theory and dispersive observables

The underlying resonance condition can be written as

$\Phi(\nu_q)=\Phi_0(\nu_q)+\Phi_n(\nu_q)=2\pi q,$

where $\Phi_0$ is the empty-cavity phase and $\Phi_n$ is the additional phase due to the intracavity sample. Broadband measurement of the resulting mode frequencies $\nu_q$ gives direct access to cavity dispersion, because deviations from an equidistant reference mode scale are themselves the spectroscopic signal (Rutkowski et al., 2017).

For mode-frequency metrology, the relevant cavity quantity is not only the phase index $\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$0 but the group index

$\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$1

Bradshaw’s analysis makes explicit that the free spectral range is governed by the group index, with

$\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$2

or more exactly by a cavity-averaged $\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$3. This distinction is not cosmetic: the phase index sets the single-frequency resonance condition, while the group index governs mode spacing, linewidth, quality factor, and frequency sensitivity (Bradshaw, 2010).

In broadband direct cavity-mode spectroscopy, the experimentally convenient quantity is the mode shift

$\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$4

defined relative to a reference cavity mode scale. The cavity group delay dispersion is then recovered from the curvature of that shift curve,

$\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$5

This framework allows mirror-coating dispersion, nonresonant gas refractivity, and resonant molecular refractivity to be extracted from the same set of measured cavity-mode positions (Rutkowski et al., 2017).

For gas-phase line spectroscopy, HCRDS makes the mode-shift interpretation explicit: $\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$6 while the cavity broadening is

$\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$7

The Kramers–Kronig linkage between dispersion and absorption therefore appears directly as a relation between cavity-mode positions and cavity-mode widths (Cygan et al., 2024).

3. Broadband implementations

Broadband cavity-mode dispersion spectroscopy became practical when optical frequency combs were combined with cavity-resolved readout. In one implementation, a stabilized Er:fiber comb and a mechanical Fourier transform spectrometer operated in a sub-nominal-resolution mode were used to characterize 16,000 cavity modes spanning 16 THz in terms of center frequency, linewidth, and amplitude. The measured center-frequency uncertainty ranged from 0.4 kHz to 3 kHz, the retrieved group delay dispersion of the cavity mirror coatings and pure N$\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$8 had 0.1 fs$\Gamma_\alpha(\nu)=\frac{c}{2\pi n}\alpha(\nu).$9 precision and 1 fs$\Delta \nu$0 accuracy, and the resonant refractivity of the $\Delta \nu$1 band of CO$\Delta \nu$2 was measured with $\Delta \nu$3 precision (Rutkowski et al., 2017).

Broadband CE-CRIS extended the same logic to direct measurement of the complex refractive index of entire molecular bands. Using an optical frequency comb and a mechanical Fourier transform spectrometer with sub-nominal resolution, it measured the absorption and dispersion spectra of three combination bands of CO$\Delta \nu$4 in the range between 1525 nm and 1620 nm. The measured cavity transmission spectrum contained about 15,000 cavity modes spanning 15 THz from 1505 to 1650 nm, and in the strongest band near 1610 nm the average precision of the Lorentzian fit parameters was about 30 Hz for center frequency and 10 Hz for linewidth (Johansson et al., 2018).

A more recent dual-comb realization used mode-locked erbium-fiber frequency combs and a length-tunable cavity to perform broadband dual-comb cavity mode dispersion spectroscopy on the entire $\Delta \nu$5 band of acetylene in the 1.5 $\Delta \nu$6m region. The weighted average acetylene pressure was 10.083(27) Pa, corresponding to a relative standard deviation of 0.27 %, and the spectral fluctuation corresponded to an absorption coefficient of $\Delta \nu$7 (Okubo et al., 23 Jun 2026).

Implementation	Primary readout	Representative scale
Comb + mechanical FTS	Center frequency, linewidth, amplitude of individual cavity modes	16,000 modes over 16 THz; 0.4–3 kHz center uncertainty (Rutkowski et al., 2017)
Broadband CE-CRIS	Lorentzian fits to cavity transmission modes	about 15,000 modes over 15 THz; about 30 Hz center precision in the strongest band (Johansson et al., 2018)
Dual-comb CMDS	Adjacent-mode spacing $\Delta \nu$8	0.27 % relative standard deviation for retrieved acetylene pressure; $\Delta \nu$9 fluctuation (Okubo et al., 23 Jun 2026)

These broadband implementations share a common structure. First, the cavity modes are individually resolved or reconstructed. Second, each mode is fit to obtain at least a center frequency and a linewidth. Third, smooth cavity-background terms such as mirror dispersion or nonresonant refractivity are separated from resonant molecular contributions. The dispersive channel is therefore frequency-metrological from the outset, whereas the amplitude channel remains more vulnerable to baseline distortions and normalization errors (Johansson et al., 2018, Okubo et al., 23 Jun 2026).

4. Transient and ring-down realizations

A major development was the transfer of cavity-mode dispersion readout from slow frequency scans to cavity transients. Cavity buildup dispersion spectroscopy (CBDS) measures the detuning between a non-resonant probe and a cavity resonance from the damped beat in the buildup transient. The detected intensity can be written in the simplified form

$\omega(k)$0

so the beat frequency directly localizes the cavity mode center. CBDS was reported to be currently 50 times less susceptible to detection nonlinearity compared to intensity-based methods, with a 20 Hz short-term sensitivity to cavity resonance shifts obtained in 400 $\omega(k)$1s (Cygan et al., 2020).

HCRDS uses the decaying cavity field rather than the buildup field. Its heterodyne ring-down signal is

$\omega(k)$2

and Fourier analysis of the decaying beat yields the exact frequencies of optical cavity modes and a dispersive spectrum of the gas sample. Using CO and HD line intensities as examples, the method demonstrated sub-permille accuracy and long-term repeatability of dispersion measurements at the $\omega(k)$3 level (Cygan et al., 2024).

Dual-comb cavity ring-down spectroscopy, as analyzed by Lisak et al., extends this logic to multiplexed readout. The method derives absorption and dispersion spectra from widths and positions of cavity modes, respectively, by multiheterodyne detection of many ring-down signals in parallel. For the coherently driven implementation, the reported mode halfwidth sensitivity was $\omega(k)$4 and 70 Hz in 1 s, while the mode position sensitivity was $\omega(k)$5 and 100 Hz in 1 s (Lisak et al., 2021).

Transient dispersive sensing in whispering-gallery resonators follows a closely related principle. In cavity ring-up spectroscopy (CRUS), a far-detuned pulse edge excites the cavity transiently, and the output

$\omega(k)$6

contains both a detuning-sensitive oscillation frequency and a detuning-sensitive peak height. The paper emphasizes that distinctive dispersive and dissipative transient sensing can be realised by simply measuring the peak height of the CRUS signal (Yang et al., 2016).

Taken together, these transient methods show that cavity-mode dispersion spectroscopy does not require a quasi-static resonance scan. The cavity eigenfrequency can instead be encoded in a time-domain beat note whose center frequency is itself the dispersive observable (Cygan et al., 2020, Cygan et al., 2024).

5. Coupled-cavity, integrated, and collective-mode dispersion

In coupled photonic crystal mode-gap cavities, the “dispersion” under study is the collective miniband rather than a single Fabry–Perot resonance. The standard tight-binding expectation,

$\omega(k)$7

predicts a symmetric band. FDTD calculations instead showed a pronouncedly asymmetric group-velocity curve, with the maximum at about $\omega(k)$8 in 2D and $\omega(k)$9 in 3D. The paper attributed this not to disorder but to the fact that the cavity mode profile itself is dispersive, so the coupling rate must be treated as $D_{\mathrm{int}}$0 rather than a constant (Lian et al., 2015).

In silicon microresonators used for resonant four-wave mixing, cavity-mode dispersion is naturally expressed as the local mismatch

$D_{\mathrm{int}}$1

A dual-cavity structure used localized mode coupling to compensate a measured intrinsic mismatch of 27 GHz in the primary ring. Thermal tuning of the auxiliary ring produced peak seeded wavelength conversion efficiency of $D_{\mathrm{int}}$2 dB, an 8 dB enhancement over the non-compensated state, across a free spectral range of 3.334 THz ($D_{\mathrm{int}}$3 nm) (Gentry et al., 2014).

On-chip Kerr-comb cavity spectroscopy generalizes the same idea to integrated dispersion mapping. A Kerr microcomb coupled to a high-$D_{\mathrm{int}}$4 SiN microring cavity was used to reconstruct

$D_{\mathrm{int}}$5

over 31 cavity lines spanning 31 THz, corresponding to 246 nm. A quadratic fit yielded

$D_{\mathrm{int}}$6

and the loaded quality factors ranged from about $D_{\mathrm{int}}$7 to $D_{\mathrm{int}}$8 across the measured spectrum (Diakonov et al., 2024).

Collective cavity-mode dispersion can also appear as an antiresonant phase feature rather than a shifted resonance peak. In arrays of dipole-coupled emitters matched to a cavity mode, subradiant collective states produce sharp antiresonances and steep output-field phase dispersion. The paper’s central quantity is the effective cooperativity

$D_{\mathrm{int}}$9

which can grow much faster than the usual linear $2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$0 scaling of independent emitters; the corresponding phase switch scales like $2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$1 (Plankensteiner et al., 2017).

A further extension appears in multimode cavity polariton spectroscopy. For a single CO molecule interacting with two degenerate, orthogonally polarized Fabry–Perot modes, the one-excitation manifold is three-dimensional rather than two-dimensional, and spectroscopy shows three bright singly-excited states under suitable driving. The paper does not study a continuous cavity band $2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$2, but it demonstrates that discrete mode degeneracy and polarization structure can themselves become the spectroscopic variables (Fischer et al., 2022).

6. Accuracy, applications, and limitations

The most mature molecular implementations show that cavity-mode dispersion spectroscopy can be quantitatively competitive with absorption-based methods. In simultaneous measurements of cavity-enhanced absorption and dispersion spectroscopy of $2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$3C$2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$4O near 1560 nm, line intensities extracted from the dispersion-induced shifts of cavity modes agreed with those from CRDS within the experimental uncertainty of about 1 per thousand. For the representative R(27) budget, the total quoted uncertainties were 0.7‰ for CRDS and 0.8‰ for CMDS (Huang et al., 2024).

The same study also states an important limitation: the CMDS signal-to-noise ratio was about an order of magnitude worse than CRDS in that setup, and the CMDS fits therefore fixed the broadening parameters $2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$5 and $2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$6 to values first determined from CRDS. This is a useful corrective to the common misconception that frequency-based observables are automatically the most sensitive observables. The metrological attraction of CMDS in that work was not raw sensitivity but the fact that it measures only the frequencies, and frequency is the quantity that could be measured most precisely (Huang et al., 2024).

Another limitation concerns engineered anomalous dispersion inside cavities. Bradshaw’s analysis shows that reducing the group index enhances resonance sensitivity and can broaden a cavity resonance without reducing resonant buildup, but it also increases the vacuum field energy of a cavity mode. In the white-light limit the field normalization scales as $2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$7, the vacuum field energy scales as $2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$8, and the quantum noise of an ideal white light cavity diverges as the cavity finesse improves (Bradshaw, 2010). This suggests that cavity-mode dispersion spectroscopy in strongly anomalous media is constrained not only by classical line-shape modeling but also by modal noise and field normalization.

A different practical limitation is interpretive. In coupled photonic crystal CROWs, asymmetric minibands may be mistakenly attributed to fabrication imperfections or finite-size effects; the photonic-crystal mode-gap cavity analysis shows that asymmetry can be intrinsic to the cavity physics. The same paper is equally explicit that not all CROWs are asymmetric: the effect is strongest for shallow defect cavities near a waveguide band edge and is negligible for systems such as racetrack resonators and photonic crystal cavities far from the band edge (Lian et al., 2015).

Applications follow directly from the frequency-based character of the observable. The literature cited here places cavity-mode dispersion spectroscopy in greenhouse gas metrology, remote sensing calibration, planetary atmosphere analysis, trace gas detection, SI-traceable molecular density measurements, optical pressure standards, delay lines, optical memory, slow-light structures, and nonlinear integrated photonics (Huang et al., 2024, Rutkowski et al., 2017, Gentry et al., 2014).

A plausible implication is that single-comb-mode sources optimized for high-finesse cavities will expand the technique further. A freely controllable single-optical-frequency comb with an estimated single-comb-mode optical power of more than 10 mW was used for comb-mode-resolved CRDS with sensitivity up to $2\omega_{p,o}-\omega_{s,o}-\omega_{i,o}$9, two orders of magnitude higher than that of previously reported comb-based CRDS. The paper did not perform cavity-dispersion retrieval, but those source characteristics are directly aligned with future cavity-mode-shift measurements in very high-finesse resonators (Nishizawa et al., 9 Dec 2025).

Cavity-mode dispersion spectroscopy is therefore best understood not as a single instrument class but as a frequency-metrological viewpoint on resonators. Whether the measured object is the shift of a Fabry–Perot longitudinal mode, the width-position pair of a ring-down eigenmode, the integrated dispersion of an on-chip microring, or the asymmetric miniband of a coupled cavity chain, the common principle is that the cavity mode itself is the spectroscopic carrier of the dispersive information (Johansson et al., 2018, Cygan et al., 2024, Lian et al., 2015).