Periodically Poled Microring Resonator

• Periodically poled microring resonators are integrated optical cavities in thin-film lithium niobate that employ spatially periodic ferroelectric domain inversion for quasi-phase matching.
• They enable efficient nonlinear optical processes such as second-harmonic generation, optical parametric oscillation, and spontaneous parametric down-conversion through engineered resonances and tunable coupling.
• Device designs balance high peak efficiency with broad bandwidth through careful control of geometry, coupling, and temperature tuning to optimize cavity resonance and nonlinear interaction.

A periodically poled microring resonator is an integrated optical cavity in which a ring-shaped or racetrack-shaped waveguide in lithium niobate is endowed with a spatially periodic inversion of the ferroelectric domain, so that the sign of the effective second-order nonlinear coefficient alternates along the optical path and implements quasi-phase matching for cavity-enhanced $\chi^{(2)}$ processes. In the thin-film lithium niobate platform, this concept has been used for high-efficiency second-harmonic generation, low-threshold optical parametric oscillation, and broadband spontaneous parametric down-conversion, with implementations ranging from fully poled z-cut microrings to racetrack resonators containing only a short periodically poled section, and to ring-like multi-resonator systems in which the poled interaction region is placed in a shared Mach–Zehnder interferometer arm rather than around the full circumference (Lu et al., 2019).

1. Definition, material platform, and device classes

The defining feature of a periodically poled microring resonator is periodic modulation of the nonlinear coefficient rather than periodic modulation of geometry. In lithium niobate microrings, periodic poling flips the sign of the effective $\chi^{(2)}$ tensor element seen by the optical mode, enabling quasi-phase matching for SHG, OPO, and SPDC without requiring exact modal phase matching. The periodic structure is therefore encoded in the ferroelectric domain pattern. By contrast, a photonic-crystal microring resonator uses periodic corrugation of the ring boundary to engineer linear mode splitting; the latter is a geometrically periodic device, not a periodically poled one (Peng et al., 1 May 2025).

Several distinct periodically poled resonator classes are represented in the literature summarized here. One class is the dual-resonant z-cut thin-film lithium niobate microring for SHG, in which a telecom TE$_{00}$ mode near $1617\,\mathrm{nm}$ and a TM$_{00}$ mode near $808\,\mathrm{nm}$ are simultaneously resonant and linked by first-order quasi-phase matching (Lu et al., 2019). A second class is the z-cut air-clad periodically poled thin-film lithium niobate microring resonator engineered for type-0 TM–TM SHG via the largest tensor element $d_{33}$, with a measured normalized SHG efficiency of $5{,}000{,}000\%/\mathrm{W}$ and a single-photon coupling rate $g/2\pi = 1.2\,\mathrm{MHz}$ (Lu et al., 2020). A third class is the submicron-poled thin-film lithium niobate microring designed for backward quasi-phase matching of a $780\,\mathrm{nm}$ pump to degenerate $\chi^{(2)}$0 signal and idler, simultaneously supporting Type-0 $\chi^{(2)}$1 and Type-I $\chi^{(2)}$2 OPO in the same device (Yang et al., 23 May 2026).

The term also extends naturally to ring-type resonators whose nonlinear section is only a fraction of the cavity. In a 2026 realization, a thin-film lithium niobate racetrack resonator incorporates a short PPLN section in one arm of a tunable Mach–Zehnder interferometer coupler, with only the telecom pump resonant and the second harmonic generated in single pass (Hefti et al., 22 May 2026). A related generalization appears in two linearly uncoupled ring-like resonators nonlinearly coupled through a periodically poled Mach–Zehnder interferometer arm; although not a single microring, the device preserves the central ingredients of periodically poled resonant $\chi^{(2)}$3 photonics and is explicitly framed as “microring-class” from a design perspective (Stefano et al., 2024).

2. Device geometry and ferroelectric domain engineering

Thin-film lithium niobate on insulator is the common substrate across the periodically poled microring implementations considered here. In the 2019 dual-resonant SHG device, the platform is z-cut, undoped, congruent thin-film lithium niobate on insulator with a $\chi^{(2)}$4 LN film, a remaining unetched LN slab of $\chi^{(2)}$5, a microring radius of $\chi^{(2)}$6, and a ring waveguide width of $\chi^{(2)}$7 (Lu et al., 2019). In the 2020 type-0 TM–TM device, the ring radius is likewise $\chi^{(2)}$8 and the ring is air-clad, with a cross-section designed to support quasi-TM modes at both $\chi^{(2)}$9 and $_{00}$0 (Lu et al., 2020). In the 2026 backward-QPM OPO, the microring is fabricated on z-cut TFLN with film thickness $_{00}$1, etch depth $_{00}$2 leaving a $_{00}$3 slab, ring radius $_{00}$4, and ring width $_{00}$5 (Yang et al., 23 May 2026).

The bus–ring coupling geometry is likewise part of the device definition because the cavity must be engineered at widely separated wavelength bands. The 2019 SHG device uses a single pulley bus waveguide with width about $_{00}$6 and gap $_{00}$7 to couple both telecom and near-visible modes (Lu et al., 2019). The 2026 OPO ring uses a wraparound or pulley coupler with bus width $_{00}$8, coupling gap $_{00}$9, and wraparound angle $1617\,\mathrm{nm}$0, specifically chosen for $1617\,\mathrm{nm}$1 TM phase-matched coupling while maintaining access to $1617\,\mathrm{nm}$2 TE and TM modes (Yang et al., 23 May 2026).

Periodic poling can be implemented either after the ring is defined or before it is etched. In the 2019 microring, radial nickel electrodes are patterned on top of the ring after device definition, the chip is heated to $1617\,\mathrm{nm}$3, and six high-voltage pulses are applied to induce ferroelectric domain inversion; after electrode removal, HF etching reveals alternating inverted and non-inverted domains with a measured period of $1617\,\mathrm{nm}$4 and duty cycle of about $1617\,\mathrm{nm}$5 (Lu et al., 2019). In the 2020 high-$1617\,\mathrm{nm}$6 device, poling electrodes are arranged radially around the ring with tooth width $1617\,\mathrm{nm}$7 and period $1617\,\mathrm{nm}$8, and several $1617\,\mathrm{nm}$9, $_{00}$0 pulses are applied at elevated temperature, after which piezoresponse force microscopy and HF-etched mock-up rings confirm alternating domains and approximately $_{00}$1 duty cycle (Lu et al., 2020). In the 2026 backward-QPM OPO, periodic poling is implemented before the ring is defined: nickel electrodes are patterned in a radial periodic pattern that tracks the future ring path, voltage is applied to invert ferroelectric domains in a submicron grating along the ring circumference, and the microring waveguide is subsequently etched within the poled region; inverted domains are again verified by HF etching (Yang et al., 23 May 2026).

Submicron poling introduces additional fabrication constraints because the electrode pattern must track the curved ring path with the correct period along arc length while maintaining uniform period and duty cycle over the full $_{00}$2 circumference. In the backward-QPM OPO, the required period is $_{00}$3, which the work describes as at the edge of what is feasible in TFLN, achieved by optimized electrode lithography and field strengths together with careful wet etching using RCA at $_{00}$4 for $_{00}$5 to reduce domain-wall defects (Yang et al., 23 May 2026). This suggests that domain fidelity, not only waveguide loss, becomes a primary performance determinant once the poling period enters the submicron regime.

3. Quasi-phase matching in microring coordinates

The generic quasi-phase-matching condition for a three-wave $_{00}$6 process is

$_{00}$7

where $_{00}$8, $_{00}$9, and $808\,\mathrm{nm}$0 are propagation wavevectors including propagation direction, $808\,\mathrm{nm}$1 is the QPM order, and $808\,\mathrm{nm}$2 is the grating wavevector (Yang et al., 23 May 2026). For SHG in a straight waveguide this is commonly written as

$808\,\mathrm{nm}$3

or, in first order,

$808\,\mathrm{nm}$4

(Lu et al., 2019, Hefti et al., 22 May 2026).

In a microring resonator, this continuous condition maps onto discrete azimuthal mode numbers. The 2019 SHG treatment writes the nonlinear coupling coefficient in a form containing the Kronecker delta condition

$808\,\mathrm{nm}$5

where $808\,\mathrm{nm}$6 and $808\,\mathrm{nm}$7 are the pump and second-harmonic azimuthal mode numbers and $808\,\mathrm{nm}$8 is the number of poling periods around the ring (Lu et al., 2019). In the fabricated device, simulations give $808\,\mathrm{nm}$9 and $d_{33}$0, and with $d_{33}$1 and $d_{33}$2 one obtains $d_{33}$3, so that $d_{33}$4, establishing exact first-order QPM (Lu et al., 2019).

The type-0 TM–TM design aimed at the largest tensor element $d_{33}$5 uses the corresponding waveguide expression

$d_{33}$6

yielding an optimal period $d_{33}$7 for TM$d_{33}$8 at $d_{33}$9 and TM$5{,}000{,}000\%/\mathrm{W}$0 at $5{,}000{,}000\%/\mathrm{W}$1 (Lu et al., 2020). The 2019 TE$5{,}000{,}000\%/\mathrm{W}$2TM$5{,}000{,}000\%/\mathrm{W}$3 SHG device instead requires $5{,}000{,}000\%/\mathrm{W}$4 by the same design logic, and explicitly notes that a TM$5{,}000{,}000\%/\mathrm{W}$5TM$5{,}000{,}000\%/\mathrm{W}$6 design using $5{,}000{,}000\%/\mathrm{W}$7 would require $5{,}000{,}000\%/\mathrm{W}$8, a more difficult period to implement uniformly around the ring with the poling technology used there (Lu et al., 2019).

Backward quasi-phase matching in a microring modifies the momentum picture more radically. In the 2026 OPO, the device is designed for degenerate backward phase matching with pump at $5{,}000{,}000\%/\mathrm{W}$9 and signal and idler near $g/2\pi = 1.2\,\mathrm{MHz}$0 propagating in opposite directions. At degeneracy,

$g/2\pi = 1.2\,\mathrm{MHz}$1

so that

$g/2\pi = 1.2\,\mathrm{MHz}$2

and the QPM condition reduces to

$g/2\pi = 1.2\,\mathrm{MHz}$3

The paper states that under these conditions “the poling wavevector compensates only the pump momentum,” so that a single submicron period $g/2\pi = 1.2\,\mathrm{MHz}$4 simultaneously phase matches both Type-0 and Type-I interactions because both share the same TM pump mode at $g/2\pi = 1.2\,\mathrm{MHz}$5 (Yang et al., 23 May 2026). This is the basis of the paper’s description of the poling as polarization-insensitive at degeneracy.

The ring resonance condition remains, independently,

$g/2\pi = 1.2\,\mathrm{MHz}$6

or equivalently $g/2\pi = 1.2\,\mathrm{MHz}$7 with $g/2\pi = 1.2\,\mathrm{MHz}$8, so quasi-phase matching alone is not sufficient; the participating wavelengths and polarizations must also coincide with cavity resonances (Yang et al., 23 May 2026). In practice, temperature tuning is repeatedly used to align those resonances after fabrication (Lu et al., 2019, Yang et al., 23 May 2026).

4. Cavity enhancement, resonance alignment, and nonlinear coupling

Periodically poled microring resonators derive their utility from simultaneous quasi-phase matching and resonant field buildup. For cavity SHG in the undepleted regime, the 2019 dual-resonant device gives the on-chip efficiency

$g/2\pi = 1.2\,\mathrm{MHz}$9

with external coupling rates $780\,\mathrm{nm}$0 and $780\,\mathrm{nm}$1, total loss rates $780\,\mathrm{nm}$2 and $780\,\mathrm{nm}$3, and detunings $780\,\mathrm{nm}$4 and $780\,\mathrm{nm}$5 (Lu et al., 2019). Under exact dual resonance and critical coupling for both modes, the maximum normalized efficiency simplifies to

$780\,\mathrm{nm}$6

showing the central role of the nonlinear coupling rate $780\,\mathrm{nm}$7 and the intrinsic quality factors (Lu et al., 2019).

The nonlinear coupling strength is determined by tensor element, domain duty cycle, and modal overlap. In the 2019 TE$780\,\mathrm{nm}$8TM$780\,\mathrm{nm}$9 SHG microring, the relevant tensor element is $\chi^{(2)}$00, the duty cycle is $\chi^{(2)}$01, and the extracted effective susceptibility is $\chi^{(2)}$02; with simulated overlap factor $\chi^{(2)}$03, the corresponding theoretical coupling rate is $\chi^{(2)}$04 (Lu et al., 2019). In the 2020 type-0 TM–TM device, the measured coupling rate is $\chi^{(2)}$05 and the figure of merit $\chi^{(2)}$06 is reported as approximately $\chi^{(2)}$07 using the loaded linewidth, or about $\chi^{(2)}$08 using intrinsic loss only (Lu et al., 2020).

Resonance alignment is especially critical when more than one polarization channel is available. In the backward-QPM OPO, the detuning is defined as

$\chi^{(2)}$09

and the relative size of $\chi^{(2)}$10 and $\chi^{(2)}$11 determines whether the Type-0 or Type-I OPO dominates (Yang et al., 23 May 2026). The TE and TM infrared resonances shift differently with temperature, allowing selective activation of either channel in the same poled ring. In the earlier dual-resonant SHG microring, thermal tuning slopes are measured as $\chi^{(2)}$12 for the telecom TE$\chi^{(2)}$13 pump and $\chi^{(2)}$14 for the near-visible TM$\chi^{(2)}$15 second harmonic, yielding a Lorentzian SHG response with measured full width at half maximum $\chi^{(2)}$16 and predicted $\chi^{(2)}$17 (Lu et al., 2019).

Singly resonant racetrack realizations relax the dual-resonance constraint by resonantly enhancing only the pump. In the 2026 tunable-coupling racetrack, the pump power enhancement is defined as

$\chi^{(2)}$18

and because the second harmonic is generated in single pass,

$\chi^{(2)}$19

At critical coupling, the pump enhancement in the PPLN arm reaches $\chi^{(2)}$20, implying a predicted SH enhancement factor $\chi^{(2)}$21 relative to a non-resonant structure of the same nonlinear length (Hefti et al., 22 May 2026). This architecture stabilizes operation near critical coupling through a tunable Mach–Zehnder interferometer coupler rather than through fixed lithographic coupling.

5. Demonstrated performance and operating regimes

The earliest benchmark in the data set is the 2019 z-cut periodically poled thin-film lithium niobate microring for dual-resonant SHG. Pumped around $\chi^{(2)}$22, it achieved an on-chip normalized SHG efficiency of $\chi^{(2)}$23 and an absolute conversion efficiency of $\chi^{(2)}$24 with a low pump power of $\chi^{(2)}$25 in the waveguide (Lu et al., 2019). The loaded quality factors were $\chi^{(2)}$26 for the telecom TE$\chi^{(2)}$27 pump mode and $\chi^{(2)}$28 for the near-visible TM$\chi^{(2)}$29 second-harmonic mode (Lu et al., 2019).

A substantial increase followed in the 2020 type-0 TM–TM periodically poled thin-film lithium niobate microring resonator, which reported a normalized SHG efficiency of $\chi^{(2)}$30, almost $\chi^{(2)}$31-fold enhancement over the authors’ previous state of the art, together with intrinsic and loaded quality factors $\chi^{(2)}$32 and $\chi^{(2)}$33 for the fundamental mode and $\chi^{(2)}$34 and $\chi^{(2)}$35 for the second harmonic (Lu et al., 2020). The inferred single-photon coupling rate $\chi^{(2)}$36 and the projected scaled-device values $\chi^{(2)}$37 and $\chi^{(2)}$38 were explicitly tied to the prospect of unconventional photon blockade in coupled PPLN microring “photonic molecules” (Lu et al., 2020).

The backward-QPM microring extends performance from SHG into OPO. In a single submicron-poled ring, the measured symmetric SHG conversion efficiencies are $\chi^{(2)}$39 for Type-I SSHG and $\chi^{(2)}$40 for Type-0 SSHG, with a ratio of about $\chi^{(2)}$41, described as being in reasonable agreement with the tensor-based estimate

$\chi^{(2)}$42

(Yang et al., 23 May 2026). Under OPO operation, the on-chip thresholds are $\chi^{(2)}$43 for Type-0 and $\chi^{(2)}$44 for Type-I, with maximum conversion efficiencies of $\chi^{(2)}$45 and $\chi^{(2)}$46, respectively (Yang et al., 23 May 2026). The extracted effective nonlinear coupling rates are $\chi^{(2)}$47 and $\chi^{(2)}$48, and the corresponding predicted thresholds of $\chi^{(2)}$49 and $\chi^{(2)}$50 are reported to be in excellent agreement with experiment (Yang et al., 23 May 2026).

The singly resonant racetrack represents a different operating point in the efficiency–bandwidth space. Using a $\chi^{(2)}$51 PPLN section and a tunable Mach–Zehnder interferometer coupler, it achieves a measured on-chip SHG efficiency

$\chi^{(2)}$52

together with an experimentally normalized enhancement factor of $\chi^{(2)}$53 over a non-resonant structure and a bandwidth of roughly $\chi^{(2)}$54 (Hefti et al., 22 May 2026). The authors explicitly position this as an alternative to fully doubly resonant microrings when broad bandwidth and fabrication tolerance are prioritized over maximal peak efficiency.

For SPDC, a two-resonator periodically poled ring-like system rather than a single microring achieves pair generation rates of up to $\chi^{(2)}$55 pump power for a single resonance and integrated pair generation rates of up to $\chi^{(2)}$56 pump power over $\chi^{(2)}$57, with reconfiguration enabling efficient generation over some $\chi^{(2)}$58 across the S, C, L, and U bands (Stefano et al., 2024). Although the nonlinear interaction is confined to a shared poled Mach–Zehnder interferometer arm, the work treats this as a functional generalization of the periodically poled microring concept.

6. Reconfigurability, limitations, and adjacent architectures

Reconfigurability in periodically poled microring systems is achieved primarily through temperature, phase shifters, or tunable coupling. In the backward-QPM OPO, temperature tunes the relative detuning between the $\chi^{(2)}$59 TM pump resonance and the $\chi^{(2)}$60 TE or TM infrared resonances; experimentally, Type-I operation is selected near $\chi^{(2)}$61 and Type-0 near $\chi^{(2)}$62 (Yang et al., 23 May 2026). In the two-resonator SPDC architecture, three phase shifters $\chi^{(2)}$63, $\chi^{(2)}$64, and $\chi^{(2)}$65 tune the pump comb, signal/idler comb, and effective coupling through the Mach–Zehnder interferometer, thereby avoiding the need for dispersion engineering to secure doubly resonant operation (Stefano et al., 2024). In the singly resonant racetrack, the effective bus–ring power coupling coefficient $\chi^{(2)}$66 is tunable from $\chi^{(2)}$67 to $\chi^{(2)}$68, allowing the device to be swept from undercoupled to overcoupled and stabilized at critical coupling after fabrication (Hefti et al., 22 May 2026).

Several limitations recur across the literature. Domain wall roughness, slight period variation, incomplete poling depth, and duty-cycle errors reduce the effective nonlinear coupling and help explain gaps between theoretical and measured efficiencies in the 2019 and 2020 devices (Lu et al., 2019, Lu et al., 2020). Nanofabrication nonuniformity around the azimuth also perturbs effective indices and resonance conditions, again lowering performance relative to idealized models (Lu et al., 2019, Lu et al., 2020). Photorefractive effects in lithium niobate are repeatedly identified as a limiting factor at higher pump powers: in the 2019 SHG ring, the measured absolute efficiency saturates at much lower power than predicted, and in the 2020 type-0 device the efficiency departs from quadratic scaling above roughly $\chi^{(2)}$69 pump because pump-power-dependent resonance shifts break the doubly resonant condition (Lu et al., 2019, Lu et al., 2020).

A common misconception is that periodicity in a microring necessarily denotes periodic poling. The photonic-crystal microring resonator on a hybrid SiN-on-LNOI platform illustrates the distinction. There, periodic corrugation of the ring boundary creates clockwise–counter-clockwise supermode splitting rather than quasi-phase matching, yielding a splitting bandwidth of $\chi^{(2)}$70, intrinsic $\chi^{(2)}$71, EO tuning coefficient $\chi^{(2)}$72, and splitting control of $\chi^{(2)}$73 with corrugation amplitude (Peng et al., 1 May 2025). The device is therefore relevant as a comparison point for periodic resonator engineering on lithium niobate, but it is not a periodically poled microring.

The application space of periodically poled microring resonators is correspondingly broad. The data explicitly connect them to second-harmonic generation, spontaneous parametric down-conversion, optical parametric oscillation, squeezed-light generation, frequency conversion, coherent Ising machines, and single-photon-level nonlinear optics (Lu et al., 2019, Lu et al., 2020, Yang et al., 23 May 2026). In the backward-QPM OPO, the same device below threshold is described as a plausible counter-propagating SPDC source with polarization diversity, since the same grating supports both TM–TM and TE–TE pair generation channels (Yang et al., 23 May 2026). More generally, the literature suggests two architectural directions: fully poled high-$\chi^{(2)}$74 rings for maximal efficiency and strong single-photon coupling, and partially poled or multi-resonator structures for broader bandwidth, easier tuning, and reduced reliance on simultaneous dual- or triply resonant alignment (Hefti et al., 22 May 2026, Stefano et al., 2024).