Electrically Tunable Silica Microtoroid Resonators

• Electrically tunable silica microtoroids are optical resonators that employ whispering-gallery modes and integrated actuation (thermal, capacitive, or electro-optic) to precisely shift resonances.
• Their designs leverage advanced fabrication techniques such as CO₂-laser reflow and microelectrode integration to achieve ultra-high quality factors and narrow linewidths.
• Measured performance metrics demonstrate sharp tuning rates and preserved optical quality, enabling practical applications in optical filtering, switching, and quantum photonic systems.

Searching arXiv for the cited microtoroid papers and closely related work to ground the article in published sources. An electrically tunable silica microtoroid is a whispering-gallery-mode microresonator in which an applied electrical signal shifts the optical resonance of a silica toroidal cavity without negating the high-$Q$ behavior that makes microtoroids useful for narrow-linewidth filtering, cavity optomechanics, and nonlinear photonics. In the silica microtoroid literature represented here, electrical tuning has been realized by two distinct mechanisms: localized thermo-optic tuning with an integrated microheater, which changes the resonant wavelength through Joule-heated refractive-index variation (Tang et al., 2014), and on-chip capacitive actuation in a cavity opto-electromechanical system, which deforms a slotted microtoroid and thereby shifts the whispering-gallery resonance while preserving an intrinsic optical quality factor up to $4.5\times 10^7$ (Baker et al., 2016). A more recent computational design extends the concept into quantum photonics by combining a silica microtoroid with a lithium-niobate tuning element to align polarization-resolved resonances for frequency-bin selection of polarization-entangled photons (Zhang et al., 3 Jul 2026).

1. Device concept and physical definition

A silica microtoroid is a toroidal optical resonator formed from silica, typically by defining a microdisk and then reflowing its rim with a CO$_2$ laser to obtain an ultra-smooth toroidal boundary. Optical confinement occurs through whispering-gallery modes (WGMs), whose resonance condition depends on the effective refractive index and optical path length. In one formulation used for a computationally designed electrically tunable microtoroid, the modal resonance frequencies satisfy

$\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$

with $L \approx 2\pi R$ and $R \approx D/2$ (Zhang et al., 3 Jul 2026).

Two experimental implementations define the electrically tunable silica microtoroid in different ways. In the capacitive architecture, silica microtoroid based cavity opto-electromechanical systems incorporate two patterned gold electrodes on the non-reflowed central silica region adjacent to a radial slot, so that an applied voltage creates an attractive capacitive force and radial deformation of the resonator (Baker et al., 2016). In the thermo-optic architecture, a microheater fabricated on the same fused-silica substrate but located about $200~\mu\mathrm{m}$ from the microtoroid raises the local temperature and red-shifts the resonance through the thermo-optic effect (Tang et al., 2014).

The common feature is electrical control over a high-$Q$ silica WGM cavity. The main distinction is the actuation pathway: mechanical deformation for capacitive tuning, thermal index tuning for microheater integration, and electro-optic perturbation in an adjacent lithium-niobate element in the quantum-device design [(Baker et al., 2016); (Tang et al., 2014); (Zhang et al., 3 Jul 2026)].

2. Resonator geometry, materials, and mode structure

In the capacitive cavity opto-electromechanical implementation, the microtoroid has major radius $R \approx 60~\mu\mathrm{m}$ after CO$_2$ reflow, minor rim diameter $4.5\times 10^7$0, and an undercut silicon pedestal radius $4.5\times 10^7$1. The silica refractive index is reported as $4.5\times 10^7$2 at $4.5\times 10^7$3, the intrinsic optical quality factor reaches up to $4.5\times 10^7$4, and the free spectral range is approximately $4.5\times 10^7$5, corresponding to about $4.5\times 10^7$6 (Baker et al., 2016).

In the microheater-based device, the substrate is UV-grade fused silica of thickness $4.5\times 10^7$7. The pre-reflow microdisk diameter is $4.5\times 10^7$8 and the disk is supported on a cylindrical silica pillar of diameter $4.5\times 10^7$9. After CO$_2$0-laser reflow, the toroid has major radius $_2$1 and a minor cross-sectional radius on the order of $_2$2--$_2$3. The final measured quality factor is approximately $_2$4 under critical coupling with a fiber taper (Tang et al., 2014).

The computational quantum-photonic design adopts a larger geometry tailored to frequency-bin operation near $_2$5. There, the toroid diameter is $_2$6, the minor radii are $_2$7, and with $_2$8 the free spectral range is approximately $_2$9, or about $\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$0 at $\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$1 (Zhang et al., 3 Jul 2026). COMSOL eigenmode simulations with an axisymmetric fine mesh of approximately $\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$2 elements and maximum size $\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$3 are used to compute field profiles and effective indices for two polarization families labeled “H-like” and “V-like” (Zhang et al., 3 Jul 2026).

These geometries show that “electrically tunable silica microtoroid” is not a single fixed device class but a family of silica toroidal WGMs whose dimensions, surrounding structures, and actuation elements are selected according to target wavelength, tuning mechanism, and system role.

3. Electrical tuning mechanisms

The capacitive tuning mechanism is explicitly electromechanical. A voltage $\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$4 applied between the patterned electrodes produces an attractive force

$\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$5

which deforms the microtoroid radially by $\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$6. The associated dispersive resonance shift is written as

$\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$7

where $\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$8 is the optomechanical coupling, $\omega_m = \frac{m c}{n_{\mathrm{eff}} L},$9 is the effective stiffness of the slotted structure, and $L \approx 2\pi R$0 is the optical tunability coefficient (Baker et al., 2016). For the geometry with $L \approx 2\pi R$1, electrode width $L \approx 2\pi R$2, and slot width $L \approx 2\pi R$3--$L \approx 2\pi R$4, finite-element modeling predicts $L \approx 2\pi R$5--$L \approx 2\pi R$6, while the measured value is $L \approx 2\pi R$7 (Baker et al., 2016).

The thermo-optic tuning mechanism is based on Joule heating in an integrated copper microheater. The dissipated power is $L \approx 2\pi R$8, the local temperature rise is $L \approx 2\pi R$9, and the WGM resonance condition is

$R \approx D/2$0

Assuming the geometric contribution is negligible because the thermal expansion coefficient is much smaller than the thermo-optic coefficient, the wavelength shift is

$R \approx D/2$1

(Tang et al., 2014). The thermo-optic coefficient is given as $R \approx D/2$2 at $R \approx D/2$3, whereas the thermal expansion coefficient is $R \approx D/2$4 and contributes less than $R \approx D/2$5 of the wavelength shift (Tang et al., 2014).

The quantum-photonic design uses a third mechanism: electro-optic perturbation in a nearby lithium-niobate patch. An applied electric field induces an index change in LN via the Pockels effect,

$R \approx D/2$6

with $R \approx D/2$7 and $R \approx D/2$8 for LN. The resonance shift for polarization $R \approx D/2$9 is obtained by first-order perturbation theory,

$200~\mu\mathrm{m}$0

yielding tuning rates $200~\mu\mathrm{m}$1 and $200~\mu\mathrm{m}$2 (Zhang et al., 3 Jul 2026).

Taken together, these studies show that electrical tunability in silica microtoroids is a mechanism-level designation rather than a single materials recipe. The silica resonator provides the high-$200~\mu\mathrm{m}$3 WGM platform, while the electrical control may be thermal, mechanical, or electro-optic depending on the integrated actuator.

4. Fabrication and integration strategies

The capacitive microtoroid is fabricated from a silicon wafer with $200~\mu\mathrm{m}$4 thermal SiO$200~\mu\mathrm{m}$5. The process flow begins with photolithography using AZ 1518 and buffered oxide etch to define slotted disks, followed by a second lithography step with AZ nLOF 2020 to define electrodes, e-beam evaporation of $200~\mu\mathrm{m}$6 W and $200~\mu\mathrm{m}$7 Au, and lift-off to produce slotted disks with gold electrodes. A third lithography protects the central region during a first XeF$200~\mu\mathrm{m}$8 undercut that releases the outer silica annulus through the slot. CO$200~\mu\mathrm{m}$9 laser reflow with approximately $Q$0 pulses forms the toroidal rim while the central region remains on the silicon heat sink, preserving the electrodes. A second XeF$Q$1 undercut then releases the slot and final structure (Baker et al., 2016).

This process depends on several critical geometric constraints. The radial slot width is $Q$2--$Q$3 and is introduced to increase mechanical compliance. The two gold electrodes, each nominally $Q$4 wide, run tangentially on either side of the slot and terminate in contact pads P1/P2. The outer electrode is supported by a single narrow silica anchor of approximately $Q$5--$Q$6 width to the silicon substrate. The lateral gap from the toroid rim to the metal is at least $Q$7 to prevent evanescent WGM overlap and excess loss; finite-element modeling indicates more than $Q$8 field decay over that distance (Baker et al., 2016).

The microheater-based device is fabricated entirely by femtosecond laser three-dimensional micromachining on fused silica. Two classes of grooves are ablated for the heater: $Q$9-wide, $R \approx 60~\mu\mathrm{m}$0-deep electrode grooves and $R \approx 60~\mu\mathrm{m}$1-wide, $R \approx 60~\mu\mathrm{m}$2-deep resistive-element grooves. Electroless plating selectively fills these grooves with copper, producing a microheater of footprint $R \approx 60~\mu\mathrm{m}$3 and total DC resistance approximately $R \approx 60~\mu\mathrm{m}$4 (Tang et al., 2014). The microdisk is then formed by water-immersion femtosecond-laser ablation with layer-by-layer annular scanning, followed by CO$R \approx 60~\mu\mathrm{m}$5 laser reflow at $R \approx 60~\mu\mathrm{m}$6, duty cycle $R \approx 60~\mu\mathrm{m}$7, and irradiation time about $R \approx 60~\mu\mathrm{m}$8 (Tang et al., 2014).

The quantum-photonic design is a computational architecture rather than an experimentally fabricated device, but its integration parameters are specific. The lithium-niobate tuning element has thickness about $R \approx 60~\mu\mathrm{m}$9 and sits $_2$0 from the toroid rim; the total voltage gap including electrodes is approximately $_2$1 (Zhang et al., 3 Jul 2026). This suggests a hybrid electro-optic integration strategy in which the silica toroid remains the WGM host while a thin-film LN actuator supplies polarization-dependent tuning.

5. Measured and computed performance

The experimental capacitive device demonstrates a quadratic frequency shift $_2$2. At $_2$3, the measured resonance shift is approximately $_2$4, corresponding to about $_2$5 cavity linewidths, and the breakdown voltage is approximately $_2$6--$_2$7, limited by field emission or air discharge (Baker et al., 2016). A harmonic modulation bandwidth is observed at the mechanical radial-breathing mode frequency $_2$8 with $_2$9, enabling direct optical modulation at about $4.5\times 10^7$00 (Baker et al., 2016). At the same time, the optical cavity lifetime yields $4.5\times 10^7$01 for $4.5\times 10^7$02 at $4.5\times 10^7$03, corresponding to an upper optical modulation bandwidth of about $4.5\times 10^7$04 (Baker et al., 2016). Transient switching ring-down is approximately $4.5\times 10^7$05, or about $4.5\times 10^7$06, dominated by the slowest mechanical quality factor (Baker et al., 2016). Importantly, the optical $4.5\times 10^7$07 remains in the mid-$4.5\times 10^7$08 range with electrodes present, the same as control devices without electrodes, and insertion loss through fiber taper coupling is unchanged (Baker et al., 2016).

The microheater device exhibits a linear dependence of resonance drift on $4.5\times 10^7$09, with slope $4.5\times 10^7$10, equivalent to approximately $4.5\times 10^7$11 (Tang et al., 2014). Over the range $4.5\times 10^7$12 to $4.5\times 10^7$13, corresponding to power up to about $4.5\times 10^7$14, the red shift is about $4.5\times 10^7$15, or approximately $4.5\times 10^7$16 (Tang et al., 2014). At $4.5\times 10^7$17 the power is $4.5\times 10^7$18, giving a tuning efficiency of about $4.5\times 10^7$19 or about $4.5\times 10^7$20 (Tang et al., 2014). The thermal response reaches $4.5\times 10^7$21 of the full shift in less than or approximately $4.5\times 10^7$22, with a similar cooling time (Tang et al., 2014).

The computational quantum device targets polarization-preserving selection of nine frequency bins. With loaded quality factor $4.5\times 10^7$23 at $4.5\times 10^7$24, the loaded linewidth is about $4.5\times 10^7$25. The untuned horizontal/vertical mismatch is on the order of $4.5\times 10^7$26, corresponding to $4.5\times 10^7$27 linewidths in normalized units, but a single bias of $4.5\times 10^7$28 aligns the nine modes so that the residual span is only $4.5\times 10^7$29 linewidths, with maximum deviation $4.5\times 10^7$30 linewidths (Zhang et al., 3 Jul 2026). Over the nine channels, computed coupling efficiencies are $4.5\times 10^7$31 and $4.5\times 10^7$32, with leakage terms remaining $4.5\times 10^7$33 (Zhang et al., 3 Jul 2026).

A concise comparison of representative parameters is useful because the term encompasses multiple tuning modalities.

Implementation	Tuning mechanism	Representative performance
Slotted silica COEMS (Baker et al., 2016)	Capacitive electromechanical actuation	$4.5\times 10^7$34 measured; $4.5\times 10^7$35 shift at $4.5\times 10^7$36; optical $4.5\times 10^7$37 up to $4.5\times 10^7$38
Microheater-integrated toroid (Tang et al., 2014)	Thermo-optic tuning by Joule heating	$4.5\times 10^7$39 slope; $4.5\times 10^7$40 shift at $4.5\times 10^7$41; $4.5\times 10^7$42
Silica microtoroid with LN patch (Zhang et al., 3 Jul 2026)	Electro-optic tuning via Pockels effect in LN	$4.5\times 10^7$43 for nine-channel H/V alignment within $4.5\times 10^7$44 linewidths

These values are not directly interchangeable because they refer to different wavelengths, mode families, linewidths, and figures of merit. A plausible implication is that the most relevant comparison depends on whether the application is large static retuning, fast dynamic modulation, or polarization-preserving channel selection.

6. Applications, limitations, and design trade-offs

The capacitive device is presented for efficient radio-to-optical frequency conversion, optical routing, and switching applications (Baker et al., 2016). The same work identifies fast optical switching and add/drop functions with switching times of about $4.5\times 10^7$45, as well as tunable filters, lasers, and frequency combs on the silica microtoroid platform (Baker et al., 2016). Its principal advantage is that electrodes can be integrated while maintaining ultrahigh optical quality factor and unchanged fiber-taper insertion loss (Baker et al., 2016).

The microheater device is suited to local electrical control of resonant wavelength in a compact monolithic structure. The heater footprint is $4.5\times 10^7$46, it is positioned about $4.5\times 10^7$47 from the toroid, and it avoids placing metal on the resonator rim, which helps preserve the high-$4.5\times 10^7$48 state (Tang et al., 2014). Compared with an external heater, the reported response is significantly faster, though it remains fundamentally thermal and therefore much slower than electromechanical modulation (Tang et al., 2014).

The quantum-photonic design assigns a more specialized role to electrical tuning. The signal photon at $4.5\times 10^7$49 passes through the microtoroid while the entangled $4.5\times 10^7$50 idler bypasses it and functions as a reference for the selected frequency channel (Zhang et al., 3 Jul 2026). Because raw birefringence would otherwise reveal polarization, the electrically controlled LN element is used to equalize the response seen by horizontally and vertically polarized photons (Zhang et al., 3 Jul 2026). After filtering and tracing out frequency, the predicted polarization-state metrics are concurrence $4.5\times 10^7$51, Bell-state fidelity $4.5\times 10^7$52, and maximum CHSH parameter $4.5\times 10^7$53 (Zhang et al., 3 Jul 2026). If relative timing and phase are also controlled, the same structure can generate a nine-channel polarization-frequency hyperentangled state with effective dimension $4.5\times 10^7$54 and Shannon entropy $4.5\times 10^7$55 (Zhang et al., 3 Jul 2026).

The limitations differ correspondingly. In the capacitive device, high DC bias is required for large tuning, with practical operation limited by breakdown near $4.5\times 10^7$56, and the mechanical ring-down quality factor limits ultimate switching speed to about $4.5\times 10^7$57 in the transient regime (Baker et al., 2016). In the microheater device, tuning speed is limited by thermal conduction time, power consumption is approximately $4.5\times 10^7$58--$4.5\times 10^7$59 for sub-nanometer shifts, and thermal crosstalk becomes a concern if many devices are tightly packed (Tang et al., 2014). In the quantum design, active tracking and feedback are required because at $4.5\times 10^7$60 the loaded linewidth is about $4.5\times 10^7$61 while the silica thermo-optic shift is several $4.5\times 10^7$62, so the bias near $4.5\times 10^7$63 must be held within approximately $4.5\times 10^7$64 (Zhang et al., 3 Jul 2026).

A common misconception is that electrical tuning in silica microtoroids necessarily implies direct electro-optic tuning of silica itself. The literature here does not support that simplification. Instead, the demonstrated and proposed devices rely on electrically induced mechanical deformation (Baker et al., 2016), electrically induced heating (Tang et al., 2014), or electro-optic modulation in an adjacent lithium-niobate element coupled to a silica toroid (Zhang et al., 3 Jul 2026).

7. Prospective developments

The capacitive study includes an explicit finite-element improvement path: interdigitated concentric electrodes with $4.5\times 10^7$65 fingers, each $4.5\times 10^7$66 wide and separated by $4.5\times 10^7$67 gaps, implemented directly on a solid non-slotted toroid (Baker et al., 2016). The simulated capacitance is approximately $4.5\times 10^7$68, described as a $4.5\times 10^7$69 increase, and the predicted tunability is $4.5\times 10^7$70, described as about a $4.5\times 10^7$71 improvement (Baker et al., 2016). In that configuration, only about $4.5\times 10^7$72 bias would be required to shift by one linewidth, and the paper states that this would enable low-voltage, GHz-bandwidth actuation with reduced DC requirements (Baker et al., 2016).

The microheater study proposes a different optimization trajectory: reducing the heater–toroid distance while avoiding metal contamination, tailoring heater geometry to adjust resistance and thermal coefficient, thinning the substrate or further undercutting the pillar for better thermal isolation, and exploring alternative materials with larger $4.5\times 10^7$73 (Tang et al., 2014). It also suggests buried electrical leads, multiple heating zones, and concentric heater rings for more compact arrays and finer spatial control (Tang et al., 2014). These proposals remain thermal in character and therefore prioritize tuning efficiency and integration density rather than high-speed response.

The computational quantum design points toward hybrid high-dimensional photonics. It combines a $4.5\times 10^7$74 silica toroid, a thin-film LN actuator, tapered-fiber access, and a tunable Fabry–Pérot analyzer with $4.5\times 10^7$75 and linewidth about $4.5\times 10^7$76 for idler-bin recognition (Zhang et al., 3 Jul 2026). The predicted coincidence-rate requirements and detector assumptions are specified quantitatively, and the calculated coincidence-to-accidental ratio remains much greater than $4.5\times 10^7$77, leaving $4.5\times 10^7$78 well above $4.5\times 10^7$79 in a realistic experiment (Zhang et al., 3 Jul 2026). This suggests that electrically tunable silica microtoroids may serve not only as classical photonic components but also as frequency-selective interfaces between polarization-entangled photon sources and frequency-encoded quantum systems.

Across these trajectories, the electrically tunable silica microtoroid emerges as a platform defined by a stable combination of ultrahigh-$4.5\times 10^7$80 silica WGM confinement and an external electrical control layer whose mechanism can be selected according to system constraints. The experimental record shows that substantial tuning can be added without obvious penalty to optical loss in both localized-heating and on-chip capacitive implementations [(Tang et al., 2014); (Baker et al., 2016)], while the more recent quantum-oriented design indicates how electrical tunability can be used not merely to shift a resonance but to enforce polarization-indistinguishable spectral selection across multiple channels (Zhang et al., 3 Jul 2026).