Tunable Photonic-Molecule Resonators

• Tunable photonic-molecule resonators are systems where coupled optical elements hybridize into supermodes, enabling precise control over resonance frequencies, linewidths, and phase.
• They exploit diverse implementations—from Fabry–Pérot cavities and whispering-gallery modes to integrated ring resonators—to achieve applications in sensing, nonlinear optics, and quantum photonics.
• Dynamic tuning methods including mechanical, thermal, photochromic, and electro-optic techniques allow real-time adjustments in spectral characteristics and quality factors.

to=functions.shell 天天中彩票怎么买json {"command":"bash -lc 'python - <<\"PY\"\nimport urllib.request\nurl=\"http://export.arxiv.org/api/query?search_query=all:tunable+photonic+molecule+optical+resonator&start=0&max_results=3\"\ntry:\n print(urllib.request.urlopen(url,timeout=10).read(500).decode())\nexcept Exception as e:\n print('ERR',e)\nPY'"}ുവനന്തപുരം to=functions.shell 神彩争霸 anasiyana code: 0 Wall time: 0.3 seconds Output: ERR HTTP Error 403: Forbidden

to=functions.shell novitads ाठमाडौंարսোন {"command":"bash -lc 'python - <<\"PY\"\nimport urllib.request\nurl=\"https://export.arxiv.org/api/query?search_query=all:tunable+photonic+molecule+optical+resonator&start=0&max_results=3\"\nreq=urllib.request.Request(url,headers={\"User-Agent\":\"Mozilla/5.0\"})\ntry:\n data=urllib.request.urlopen(req,timeout=10).read(1000).decode()\n print(data)\nexcept Exception as e:\n print('ERR',e)\nPY'"}【อ่านข้อความเต็ม to=functions.shell to=functions.shell 天天中彩票人工්json {"command":"bash -lc 'python - <<\"PY\"\nimport urllib.request\nurl=\"https://export.arxiv.org/api/query?search_query=id:([2103.16548](/papers/2103.16548))&start=0&max_results=1\"\nreq=urllib.request.Request(url,headers={\"User-Agent\":\"[email protected]\"})\ntry:\n with urllib.request.urlopen(req,timeout=20) as r:\n b=r.read(1000)\n print(repr(b[:200]))\nexcept Exception as e:\n print('ERR',type(e),e)\nPY'"} to=functions.shell 新天天彩票 json {"command":"bash -lc 'curl -I -A \"Mozilla/5.0\" https://export.arxiv.org/api/query?search_query=id:([2103.16548](/papers/2103.16548))&start=0&max_results=1 | head'"}【อ่านข้อความเต็ม to=functions.shell Tunable photonic-molecule optical resonators are optical analogues of chemical molecules in which two or more resonant photonic elements are electromagnetically coupled so that isolated cavity resonances hybridize into collective supermodes. In this class of systems, tuning can act on bare resonance frequencies, inter-resonator coupling, modal loss, intracavity phase, or even optical-path topology, depending on whether the platform is a Fabry–Pérot microcavity, a whispering-gallery-mode pair, a photonic-crystal cavity array, a coupled microring network, or a single multimode ring engineered to emulate a molecule (1207.1274, Palekar et al., 2021). The resulting devices are used for spectral engineering, controllable light–matter interaction scenarios, nonlinear optics, sensing, and quantum photonics (1207.1274, Zhang et al., 2020).

1. Coupled-resonator formalism and supermode formation

The canonical description of a photonic molecule starts from coupled-mode theory. For two weakly coupled cavities with normalized complex field amplitudes $a_1(t)$ and $a_2(t)$, the standard equations are

$$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$

with $\omega_i$ the isolated-cavity resonant frequencies, $\gamma_i$ the total loss rates, and $\kappa$ the evanescent coupling rate. The normal-mode frequencies are

$$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$

which reduce to $\Omega_\pm=\omega_0\pm\kappa$ in the symmetric case, giving a splitting $2\kappa$ (1207.1274).

Cai et al. formulated the same physics for an array of $N$ evanescently coupled photonic-crystal cavities through the Hamiltonian

$a_2(t)$0

with $a_2(t)$1 the photon-tunneling rate. For two cavities tuned into resonance, the normal modes split by $a_2(t)$2, while with unequal losses the eigenfrequencies become $a_2(t)$3, and the coupling can be extracted from

$a_2(t)$4

(Cai et al., 2013).

A notable extension replaces multiple physical resonators with multiple transverse modes of a single ring. In the two-mode case, the multimode single-ring photonic molecule is described by

$a_2(t)$5

or equivalently by the non-Hermitian matrix

$a_2(t)$6

Its eigenvalues $a_2(t)$7 determine both resonance-frequency splitting and linewidth splitting through

$a_2(t)$8

with the zero-detuning, lossless limit giving $a_2(t)$9 (Lu et al., 14 Jan 2026).

In Fabry–Pérot implementations, the molecule-like spectral structure is controlled through the longitudinal resonance condition. For the mechanically tunable polymer/air Bragg microcavity introduced by Palekar et al., the $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$0-th longitudinal mode satisfies

$$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$1

or

$$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$2

with $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$3 for a mixed air/polymer spacer (Palekar et al., 2021). This formulation makes explicit that tuning may act on cavity length, effective index, or mirror phase.

2. Implementations and physical architectures

Palekar et al. realized a lithographically defined Fabry–Pérot platform by two-photon lithography in Nanoscribe IP-DIP resist with refractive index $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$4 at $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$5 nm. The voxel height is $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$6 nm, enabling direct 3D “printing” of dielectric mirror stacks and spacers on arbitrary substrates. Their polymer/air Bragg mirrors use alternating polymer and air layers with $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$7, $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$8, quarter-wave dimensions near $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$9 nm, $\omega_i$0 nm and $\omega_i$1 nm, and $\omega_i$2 to $\omega_i$3 layer pairs. Both a hybrid cavity, comprising a bottom conventional DBR and a top air-Bragg reflector, and an all-air-Bragg cavity were studied, with the active medium placed either on the lower mirror or suspended in the spacer (Palekar et al., 2021).

In photonic-crystal implementations, Cai et al. used linear-defect L5 cavities in a 160 nm GaAs membrane with embedded InAs quantum dots and a $\omega_i$4 nm photochromic polymer overlayer. The two-cavity molecule consisted of cavities separated by five rows of holes, $\omega_i$5 center-to-center, while the three-cavity molecule arranged three identical cavities in line. Focused optical addressing of individual cavities was achieved with $\omega_i$6 spots (Cai et al., 2013).

Whispering-gallery photonic molecules provide a distinct geometry. Peng et al. studied direct evanescent coupling between free silica microtoroids or microspheres and on-chip polymer-coated silica microtoroids. The free microtoroids had radius $\omega_i$7, the free microspheres $\omega_i$8, and the coupling gap was tuned with a 3-axis nanopositioner of $\omega_i$9 nm resolution. A tapered fiber with $\gamma_i$0 waist launched light at $\gamma_i$1 nm into one resonator only, while the second resonator was mechanically positioned to tune the overlap (Peng et al., 2013).

Integrated ring-based molecules span both multi-resonator and single-resonator topologies. The fully symmetric three-resonator photonic molecule of the silicon-nitride platform uses three identical rings of radius $\gamma_i$2 at the vertices of an equilateral triangle, each coupled to its two neighbors and to its own bus waveguide; monolithic PZT actuators are deposited over each bus waveguide with a 2 $\gamma_i$3 lateral offset (Wang et al., 2021). The heterogeneous photonic molecule combines a silicon ring resonator of radius $\gamma_i$4 with a photonic-crystal nanobeam side-coupled across an edge-to-edge gap $\gamma_i$5 nm and overlap length $\gamma_i$6 (Smith et al., 2019).

The newer single-ring paradigm creates the “molecule” within one cavity. In the multimode single-ring photonic molecule, transmissive mode converters are co-directional gratings inside a multimode ring; for TE$\gamma_i$7–TE$\gamma_i$8 coupling, each section uses period $\gamma_i$9, corrugation depth $\kappa$0, and $\kappa$1 periods, with phase matching $\kappa$2 (Lu et al., 14 Jan 2026). In thin-film lithium niobate, a racetrack resonator of total round-trip length $\kappa$3 mm supports bright TE$\kappa$4 and dark TM$\kappa$5 families; a long-lived photorefractive grating then hybridizes them into a reconfigurable single-ring photonic molecule (Zhang et al., 4 Jun 2026).

3. Tuning modalities

Different platforms realize tunability through different physical perturbations.

Mechanism	Representative platform	Reported result
Mechanical compression	Polymer/air Bragg Fabry–Pérot cavity (Palekar et al., 2021)	Blue shifts $\kappa$6 nm up to $\kappa$7 MPa
Photochromic index control	GaAs photonic-crystal molecule (Cai et al., 2013)	Local reversible shift up to $\kappa$8 nm
Thermal detuning + gap control	Coupled WGM hybrid resonators (Peng et al., 2013)	Splitting tuned from $\kappa$9 to $$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$0 GHz
Stress-optic PZT actuation	Symmetric three-ring Si$$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$1N$$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$2 molecule (Wang et al., 2021)	$$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$3 GHz at $$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$4 V, $$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$5 nW DC power
Acoustic dynamic Bragg mirror	Coupled microrings on lithium-niobate-on-sapphire (Zhu et al., 27 Nov 2025)	$$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$6, $$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$7 at $$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$8 mW
Photorefractive grating writing	Single TFLN racetrack molecule (Zhang et al., 4 Jun 2026)	$$\Omega_\pm=\frac{\omega_1+\omega_2}{2}\pm\sqrt{\kappa^2+(\Delta\omega/2)^2}, \qquad \Delta\omega=\omega_1-\omega_2,$$9 GHz over $\Omega_\pm=\omega_0\pm\kappa$0 THz bandwidth
Molecular photoswitching	Azobenzene-functionalized silica toroid (2002.04644)	$\Omega_\pm=\omega_0\pm\kappa$1 nm $\Omega_\pm=\omega_0\pm\kappa$2 over $\Omega_\pm=\omega_0\pm\kappa$3 h
Drive-phase control	Two-cavity optical molecule (Wang et al., 2016)	One cavity can be darkened by tuning only $\Omega_\pm=\omega_0\pm\kappa$4

Mechanically tunable Fabry–Pérot cavities exploit compression of air gaps while polymer layers remain essentially unchanged. For small strain $\Omega_\pm=\omega_0\pm\kappa$5, the resonance shift is

$\Omega_\pm=\omega_0\pm\kappa$6

so $\Omega_\pm=\omega_0\pm\kappa$7 strain at $\Omega_\pm=\omega_0\pm\kappa$8 nm gives $\Omega_\pm=\omega_0\pm\kappa$9 nm. FEA+TMM in the polymer/air Bragg system yields a tuning slope of $2\kappa$0 nm/MPa up to $2\kappa$1 MPa (Palekar et al., 2021).

Photochromic tuning perturbs the cavity dielectric function locally. Cai et al. used 1,3,3-Trimethylindolinonaphthospirooxazine diluted in PMMA, where near-UV exposure at $2\kappa$2 nm locally increases the refractive index and red-shifts the selected cavity. The perturbative frequency shift is

$2\kappa$3

and the shift is fully reversible with green $2\kappa$4 nm illumination at $2\kappa$5 (Cai et al., 2013).

Thermal tuning appears in several distinct roles. In WGM hybrid resonators, it coarsely aligns dissimilar cavity modes to degeneracy through the resonance law $2\kappa$6, while mechanical gap control then tunes the coupling coefficient $2\kappa$7 (Peng et al., 2013). In the heterogeneous ring–nanobeam molecule, a temperature sweep from $2\kappa$8 to $2\kappa$9 drives the detuning $N$0 through zero and continuously changes the mixing angle $N$1 defined by $N$2, thereby converting the supermodes from ring-like to beam-like (Smith et al., 2019).

Electrically assisted tuning spans both stress-optic and acoustic regimes. The monolithic PZT actuators on ultra-low-loss silicon nitride provide $N$3 with $N$4 GHz/V and leakage current $N$5 nA at $N$6 V, corresponding to $N$7 nW per actuator (Wang et al., 2021). In contrast, the acoustically controlled microring molecule uses a traveling acoustic wave to generate a dynamic Bragg mirror with

$N$8

thereby coupling CW and CCW modes with rate

$N$9

This tuning does not merely shift a resonance; it inserts a frequency-selective mirror into the circulating path (Zhu et al., 27 Nov 2025).

All-optical and long-lived tuning can also be mediated by molecular or photorefractive media. Azobenzene monolayers on silica toroids shift the effective index through trans–cis isomerization under $a_2(t)$00 nm illumination, following $a_2(t)$01 (2002.04644). In thin-film lithium niobate, interference between bright and dark modes writes a photorefractive grating with envelope $a_2(t)$02, producing a coupling

$a_2(t)$03

that can be written, erased, and rewritten optically (Zhang et al., 4 Jun 2026).

4. Spectral characteristics, quality factors, and programmability

Mirror design strongly constrains the accessible spectrum in Fabry–Pérot photonic molecules. For the polymer/air Bragg mirrors, Palekar et al. found $a_2(t)$04 for $a_2(t)$05, $a_2(t)$06 for $a_2(t)$07, and $a_2(t)$08 for $a_2(t)$09. For $a_2(t)$10, the stop-band width narrows from $a_2(t)$11 nm for the $a_2(t)$12 polymer layer to $a_2(t)$13 nm for the $a_2(t)$14 case (Palekar et al., 2021). This directly couples mirror geometry to mode selectivity and achievable $a_2(t)$15.

The same study reported $a_2(t)$16 at $a_2(t)$17 for the hybrid cavity, decreasing to $a_2(t)$18 at $a_2(t)$19 MPa; the all-air-Bragg cavity decreased from $a_2(t)$20 to $a_2(t)$21. More mirror pairs increase $a_2(t)$22, with $a_2(t)$23 for $a_2(t)$24 air-Bragg pairs coupled to a high-index DBR (Palekar et al., 2021). These values clarify that mechanical tunability and high $a_2(t)$25 are coupled design variables rather than independent targets.

Photochromically tuned photonic-crystal molecules provide finer spectral granularity. Cai et al. reported a spectrometer-limited tuning resolution $a_2(t)$26 nm and incremental tuning step $a_2(t)$27 nm. In the two-cavity GaAs molecule, the bare detuning was initially $a_2(t)$28 nm; at resonance the observed normal-mode splitting was $a_2(t)$29 nm, corresponding to $a_2(t)$30 GHz. More than $a_2(t)$31 full red/blue cycles were demonstrated with no degradation of $a_2(t)$32 (Cai et al., 2013).

Integrated ring molecules combine high $a_2(t)$33 with matrix-controlled spectral complexity. The three-resonator PZT-controlled system achieved loaded $a_2(t)$34 and intrinsic $a_2(t)$35 with PZT, compared with $a_2(t)$36 and $a_2(t)$37 for a reference without PZT; the $a_2(t)$38 reduction in $a_2(t)$39 quantified the optical penalty of actuator integration (Wang et al., 2021). The multimode single-ring molecule reached loaded $a_2(t)$40 and $a_2(t)$41 at full splitting, with intrinsic $a_2(t)$42 exceeding $a_2(t)$43 for each branch, while the normalized splitting $a_2(t)$44 swept from $a_2(t)$45 to $a_2(t)$46 as $a_2(t)$47 varied from $a_2(t)$48 to $a_2(t)$49 (Lu et al., 14 Jan 2026).

Acoustically and photorefractively programmed molecules add a dynamical dimension. In the acoustic microring platform, the measured splitting follows $a_2(t)$50, with strong coupling signaled by $a_2(t)$51; for $a_2(t)$52 mW, $a_2(t)$53 (Zhu et al., 27 Nov 2025). In the TFLN single-ring molecule, hybrid doublets emerged across more than $a_2(t)$54 longitudinal modes, with a full-width half-maximum coupling bandwidth $a_2(t)$55 THz and decay time $a_2(t)$56 min after the write beam was turned off (Zhang et al., 4 Jun 2026). This suggests a distinct operating regime in which configuration latency is slow but retention is long.

5. Light–matter interaction, molecular integration, and field localization

The Fabry–Pérot polymer/air Bragg platform was explicitly developed for controllable light–matter interaction scenarios. Palekar et al. considered molecules, nanoparticles, quantum dots, and 2D materials placed either on the lower mirror or suspended in the spacer, with positioning accuracy $a_2(t)$57 nm to locate the emitter at an antinode of $a_2(t)$58. The field-enhancement factor reaches up to $a_2(t)$59 relative to free space, the mode volume is typically on the order of $a_2(t)$60, and the Purcell factor

$a_2(t)$61

is estimated as $a_2(t)$62 for $a_2(t)$63 nm, $a_2(t)$64, $a_2(t)$65, and $a_2(t)$66. For a WS$a_2(t)$67 monolayer in a hybrid cavity, the reported vacuum Rabi splitting is $a_2(t)$68 meV at total cavity length $a_2(t)$69 and mode $a_2(t)$70 (Palekar et al., 2021).

The heterogeneous ring–nanobeam molecule offers a different route to field engineering. Because the supermode fields obey

$a_2(t)$71

thermal control of $a_2(t)$72 continuously changes not only the eigenfrequencies but also the spatial composition and effective mode volumes $a_2(t)$73. At $a_2(t)$74, the supermodes are equally mixed, while the nanobeam retains the smaller mode volume and stronger field concentration (Smith et al., 2019). This provides a direct mechanism for tuning modal participation of a high-confinement cavity without changing the physical gap.

Molecular functionalization can itself become the tuning medium. In the azobenzene-monolayer toroidal microresonator, the molecular layer thickness is $a_2(t)$75 nm and the refractive indices extracted by spectroscopic ellipsometry at $a_2(t)$76 nm are $a_2(t)$77 and $a_2(t)$78. After functionalization, the loaded $a_2(t)$79 at $a_2(t)$80 nm is $a_2(t)$81, and under $a_2(t)$82 nm pumping of $a_2(t)$83 mW the resonance shift is $a_2(t)$84 nm $a_2(t)$85; over $a_2(t)$86 h of continuous $a_2(t)$87 nm illumination it reaches $a_2(t)$88 nm $a_2(t)$89 (2002.04644). In this case the molecule is not merely an emitter or analyte but the active reconfiguration layer.

A common misconception is that “molecular” in this context necessarily refers to attached chemical molecules. In the dominant usage of the field, photonic molecules are coupled-resonator systems; however, the literature also contains platforms where molecular emitters or photoswitchable molecular monolayers are integrated into the resonator and participate directly in tuning or light–matter coupling (Palekar et al., 2021, 2002.04644).

6. Non-Hermitian, nonlinear, and topological regimes

Tunable photonic molecules are widely used for spectral engineering of nonlinear interactions. In the two-ring silicon-nitride molecule designed for degenerate squeezing, only the unwanted resonances are intentionally hybridized, while the two pumps and signal mode remain essentially unperturbed. The avoided crossing reaches a total splitting of $a_2(t)$90 GHz, and splitting the parasitic resonances by $a_2(t)$91 GHz suppresses their field enhancement by $a_2(t)$92. The device produced directly measured squeezing of $a_2(t)$93 dB and inferred on-chip squeezing of $a_2(t)$94 dB, with squeezing bandwidth $a_2(t)$95 GHz (Zhang et al., 2020).

The triple-state Si$a_2(t)$96N$a_2(t)$97 photonic molecule for degenerate optical parametric oscillation uses three identical rings in a linear array to create antisymmetric, central, and symmetric supermodes with frequencies

$a_2(t)$98

A supermode local-dispersion parameter

$a_2(t)$99

is tuned thermally through zero to optimize the four-wave-mixing phase-matching condition $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$00. The supermode splitting is in the tens of gigahertz range, the bare-ring FSR is $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$01 GHz, and the loaded $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$02 of the central supermode is $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$03 (Tomazio et al., 2024).

Single-ring photonic molecules also access non-Hermitian physics. In the transmissive-mode-converter platform, designing $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$04 causes the system to pass through exceptional points as the conversion efficiency $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$05 is tuned. Bright and dark supermodes appear near the diabolic point, while linewidth interchange and mode coalescence occur at the exceptional point (Lu et al., 14 Jan 2026). This establishes that photonic-molecule behavior does not require separate cavities; what matters is hybridization of resonant degrees of freedom.

The acoustic microring molecule extends tunability into topology. When the acoustically induced CW–CCW coupling $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$06 approaches the static inter-ring coupling $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$07, the simple four-mode photonic-molecule picture breaks down. In the limit $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$08, the dynamic Bragg mirror effectively “cuts” the outer ring, a photon must complete two physical round trips before its orientation returns to itself, and the effective cavity length doubles while the FSR is halved. Zhu et al. identify this as a transition toward Möbius-strip topology and state that full transfer-matrix theory, rather than perturbative coupled-mode theory, is then needed to reproduce the measured spectra (Zhu et al., 27 Nov 2025).

7. Applications, limitations, and evolving definitions

Applications reported across the literature include low-threshold single-mode microlasers, directional emission, coupled-resonator-induced transparency, slow-light waveguides, refractive-index and rotation sensing, Purcell-enhanced single-photon sources, strong-coupling polariton devices with 2D-semiconductor monolayers, nonlinear optics, low-threshold nanolasers, optical filters, analog optical computing, Brillouin lasers, and quantum photonic simulations (1207.1274, Palekar et al., 2021, Wang et al., 2021). Cai et al. further state that arrays of $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$09 photochromically addressable cavities should be feasible on a $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$10 chip, with the practical limit set by $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$11 spot-to-spot spacing and photochromic-film cross-talk (Cai et al., 2013).

Several technical limits recur. Strong coupling is not established merely by observing two lines: in WGM molecules, the $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$12 dB linewidth-splitting criterion is $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$13 (Peng et al., 2013), while in the acoustic platform it is quantified by cooperativity $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$14 with $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$15 marking strong coupling (Zhu et al., 27 Nov 2025). Tuning speed also varies widely across mechanisms: photochromic and photorefractive programming are reversible and low-power but slower, whereas acoustic and electro-optic methods are faster but generally require RF or bias circuitry (Cai et al., 2013, Zhang et al., 4 Jun 2026).

Another misconception is that tuning only means shifting bare resonances. The literature shows at least five distinct control axes: tuning $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$16 through thermal, electro-optic, or photochromic index change; tuning $$\frac{da_1}{dt} = (j\omega_1-\gamma_1)a_1-j\kappa a_2+s_{\rm in}, \qquad \frac{da_2}{dt} = (j\omega_2-\gamma_2)a_2-j\kappa a_1,$$17 mechanically through the inter-cavity gap; tuning linewidth asymmetry to access exceptional points; tuning intracavity occupation through drive phase; and tuning the optical path itself through a dynamic Bragg mirror or a written photorefractive grating (Wang et al., 2016, Zhu et al., 27 Nov 2025, Zhang et al., 4 Jun 2026). A plausible implication is that future classifications of photonic molecules will be organized less by geometry alone and more by which element of the effective Hamiltonian can be programmed.

The definition of a photonic molecule is itself evolving. Early work centered on distinct resonators coupled through evanescent overlap (1207.1274), whereas more recent single-ring multimode and photorefractive implementations show that the same supermode physics can be realized inside one lithographically defined cavity (Lu et al., 14 Jan 2026, Zhang et al., 4 Jun 2026). This broadening of scope preserves the central idea—engineered hybridization of resonant photonic states—while expanding the design space for tunable optical resonators.