Transversely Pumped AlGaAs Microcavity

• Transversely Pumped AlGaAs Microcavity is a photonic platform that integrates vertical free-space pumping with lateral waveguide confinement to facilitate strong light-matter interactions.
• Engineered DBRs, chirped stacks, and patterned gratings optimize pump transparency and enhance nonlinear conversion efficiencies across various device implementations.
• Applications span low-threshold polariton condensation, ultrafast Kerr switching, and parametric wavelength conversion linking trapped-ion transitions to telecom networks.

A transversely pumped AlGaAs waveguide microcavity is an AlGaAs-based integrated photonic structure in which the optically active or generated fields propagate in-plane in a waveguide, Bragg-reflection waveguide, or laterally patterned microcavity, while the pump is incident from free space across the vertical epitaxial stack. In the literature, this architecture appears in several closely related forms: a ridge waveguide combined with a vertical DBR microcavity resonant for the pump and guiding signal and idler along the chip (Simeone et al., 13 Mar 2026); a GaAs/AlGaAs waveguide with a shallow 1D grating that supports a symmetry-protected bound state in the continuum for low-density polariton condensation under normal-incidence femtosecond pumping (Riminucci et al., 2022); and waveguide-relevant extensions of planar GaAs/AlGaAs microcavity designs in which transverse pump injection is limited or enabled by stop-band engineering of the vertical multilayer (Poellmann et al., 2016). Across these realizations, the defining technical problem is not merely lateral guiding, but simultaneous control of vertical pump access, in-plane modal confinement, bandgap-selective absorption, and strong light-matter or nonlinear optical interaction in the same AlGaAs heterostructure.

1. Architectural concept and material platforms

The most explicit realization is a ridge waveguide etched into a multilayer AlGaAs heterostructure grown along the vertical axis, with signal and idler guided along the chip plane and a planar Fabry–Pérot microcavity for the pump formed by top and bottom DBRs. In the device designed to link a trapped-ion transition to the telecom C-band, the pump is at $\lambda_p = 640\,\mathrm{nm}$ and is resonantly enhanced between a 48-period bottom DBR and a 36-period top DBR, while the waveguide core contains four periods of alternating Al$_{0.50}$Ga$_{0.50}$As and Al$_{0.80}$Ga$_{0.20}$As layers that provide the quasi-phase-matched nonlinear region (Simeone et al., 13 Mar 2026). The same paper emphasizes that all layers have Al fraction $\ge 50\%$ to ensure transparency at $640\,\mathrm{nm}$.

A second platform replaces the vertical DBR cavity with a planar GaAs/AlGaAs slab waveguide patterned by a shallow 1D grating. The epitaxial stack consists of a $500\,\mathrm{nm}$ Al$_{0.8}$Ga$_{0.2}$As lower cladding, a $_{0.50}$0 core containing 12 GaAs quantum wells and 13 Al$_{0.50}$1Ga$_{0.50}$2As barriers, and a $_{0.50}$3 GaAs cap. The grating is $_{0.50}$4 long, $_{0.50}$5 wide, has period $_{0.50}$6, and is etched to depths of $_{0.50}$7, $_{0.50}$8, or $_{0.50}$9 (Riminucci et al., 2022). Here the microcavity is not a vertical DBR cavity but a laterally modulated waveguide cavity in momentum space.

Related GaAs/AlGaAs planar microcavities supply design rules that transfer naturally to waveguide geometry. One important example uses an AlAs $_{0.50}$0 cavity containing three sets of four $_{0.50}$1 GaAs quantum wells, with Al$_{0.50}$2Ga$_{0.50}$3As/AlAs DBRs and operation near $_{0.50}$4 at $_{0.50}$5 (Poellmann et al., 2016). Another important comparator is the Al$_{0.50}$6Ga$_{0.50}$7As/Al$_{0.50}$8Ga$_{0.50}$9As DBR platform, where replacing GaAs in the mirrors opens an optical pumping window from $_{0.80}$0 to $_{0.80}$1 while keeping the GaAs cavity absorbing (Shih et al., 2023). Together these results define the core AlGaAs design logic: low-gap material is placed only where carrier generation or nonlinear interaction is desired, while higher-gap AlGaAs layers are used in claddings and DBRs to control both confinement and pump transparency.

2. Transverse pumping and vertical pump-access engineering

Under transverse pumping, the first obstacle is usually the vertical multilayer itself. In broad-band femtosecond excitation from above, conventional quarter-wave DBRs tend to reflect the pump, so only narrow leaky-mode transmission dips contribute to absorption. This was demonstrated in a GaAs/AlGaAs planar microcavity pumped with near-infrared femtosecond pulses of spectral FWHM $_{0.80}$2, at $_{0.80}$3 incidence and a $_{0.80}$4 spot. By shifting the DBR periods away from strict $_{0.80}$5—to approximately $_{0.80}$6 for the top DBR and $_{0.80}$7 for the bottom DBR—and adding a 20-pair chirped DBR behind the bottom mirror, the structure produced transmission windows near $_{0.80}$8, $_{0.80}$9, and $_{0.20}$0 while keeping the cavity mode fixed (Poellmann et al., 2016).

The result was a direct improvement in injection efficiency, defined as proportional to $_{0.20}$1. At $_{0.20}$2, the optimized structure improved injection efficiency by a factor of $_{0.20}$3 relative to a conventional microcavity, and at $_{0.20}$4 the factor reached $_{0.20}$5 (Poellmann et al., 2016). The same work showed, via transfer-matrix calculations, that the additional chirped DBR increases non-resonant absorption in the quantum wells by approximately $_{0.20}$6 through a second pass of transmitted pump light.

A closely related strategy appears in optically pumped AlGaAs micropillars. There, Al$_{0.20}$7Ga$_{0.20}$8As/Al$_{0.20}$9Ga$\ge 50\%$0As DBRs create a pump-transparent spectral window from $\ge 50\%$1 to $\ge 50\%$2, while the GaAs cavity remains absorbing. Pump wavelengths outside this window either encounter strong absorption in the top DBR or inefficient excitation of pump-level excitons. Crossing the DBR absorption edge by switching the pump from $\ge 50\%$3 to $\ge 50\%$4 reduces the lasing threshold from $\ge 50\%$5 to $\ge 50\%$6, and the modified DBRs raise the simulated power conversion efficiency from about $\ge 50\%$7 to about $\ge 50\%$8 (Shih et al., 2023).

These results establish a general principle for transversely pumped AlGaAs waveguide microcavities: high reflectivity at the signal or polariton frequency is not sufficient. Pump access must be engineered explicitly, either by stop-band detuning, by a pump-transparent DBR/cladding composition, or by a backside reflector that increases the effective absorption path. A plausible implication is that waveguide microcavities driven by femtosecond or broad-band pumps are often limited by vertical pump rejection rather than by the intrinsic properties of the guided mode.

3. Strong coupling, lateral confinement, and polariton implementations

In polaritonic realizations, the waveguide microcavity supports a photonic mode strongly coupled to quantum-well excitons. In planar GaAs/AlGaAs cavities this is described by the standard coupled-oscillator dispersion

$\ge 50\%$9

with measured Rabi splittings of approximately $640\,\mathrm{nm}$0 to $640\,\mathrm{nm}$1 in related GaAs/AlGaAs microcavities (Poellmann et al., 2016). In waveguide form, the same logic applies to guided modes whose vertical field maxima overlap the quantum wells.

The GaAs/AlGaAs BIC waveguide is the clearest example. Its 1D grating couples forward and backward TE$640\,\mathrm{nm}$2 guided modes, splits the photonic dispersion into bright and dark branches, and creates a symmetry-protected BIC on the dark branch at $640\,\mathrm{nm}$3. Excitons at $640\,\mathrm{nm}$4 couple to these modes with Rabi splitting $640\,\mathrm{nm}$5, producing bright and dark lower polariton branches (Riminucci et al., 2022). For a $640\,\mathrm{nm}$6-deep grating, the BIC excitonic fraction is tunable through the grating pitch: $640\,\mathrm{nm}$7 at $640\,\mathrm{nm}$8, $640\,\mathrm{nm}$9 at $500\,\mathrm{nm}$0, and $500\,\mathrm{nm}$1 at $500\,\mathrm{nm}$2 (Riminucci et al., 2022).

Under non-resonant transverse pumping at $500\,\mathrm{nm}$3 with $500\,\mathrm{nm}$4 pulses at $500\,\mathrm{nm}$5 and a $500\,\mathrm{nm}$6 FWHM spot, this structure reaches polariton condensation at a threshold corresponding to a few $500\,\mathrm{nm}$7 per pulse after ALD passivation (Riminucci et al., 2022). The lowest threshold occurs for the $500\,\mathrm{nm}$8-deep, $500\,\mathrm{nm}$9-period design, where the BIC is more excitonic and is better isolated from bright lossy modes. The same study notes an estimated BIC polariton lifetime of a few hundred picoseconds.

Laterally structured (Al,Ga)As microcavities provide additional guidance on confinement in waveguide-like geometries. Shallow patterning of the cavity spacer produces lower-polariton traps with $_{0.8}$0 and graded interfaces whose lateral widths are $_{0.8}$1 and $_{0.8}$2. The effective potential is therefore smooth rather than abrupt, and confinement down to an effective potential width of $_{0.8}$3 is obtained (Kuznetsov et al., 2018). This suggests that transversely pumped AlGaAs waveguide microcavities defined by shallow patterning should be analyzed with graded lateral potentials rather than ideal square wells.

4. Nonlinear waveguide microcavities: SPDC, OPA, and wavelength conversion

The transversely pumped architecture also supports non-classical and parametric operation. In the inverse-designed AlGaAs device linking $_{0.8}$4 ions to telecom networks, the pump arrives from free space at an angle $_{0.8}$5 in the $_{0.8}$6–$_{0.8}$7 plane, with longitudinal component $_{0.8}$8. Phase matching for the counterpropagating type-II process is written as

$_{0.8}$9

and the nonlinear coupling is quantified by

$_{0.2}$0

The optimized structure raises $_{0.2}$1 from approximately $_{0.2}$2 to approximately $_{0.2}$3 and improves conversion efficiency from $_{0.2}$4 to $_{0.2}$5 pairs per pump photon, a $_{0.2}$6-fold increase (Simeone et al., 13 Mar 2026).

This device generates polarization-entangled photon pairs with one photon at $_{0.2}$7, resonant with the $_{0.2}$8 transition of $_{0.2}$9, and the other near $_{0.50}$00 in the telecom C-band. By simultaneously pumping at $_{0.50}$01 and $_{0.50}$02 and spectrally filtering the output, the remaining indistinguishable HV and VH pathways yield the Bell-state form

$_{0.50}$03

with a predicted heralded ion–telecom-photon entanglement rate of order of 2 events per minute under the assumptions stated in the paper (Simeone et al., 13 Mar 2026).

Related traveling-wave AlGaAs Bragg-reflection waveguides show what the nonlinear core can provide even without a resonant microcavity. A $_{0.50}$04-long, $_{0.50}$05-wide BRW operated as a $_{0.50}$06 OPA exhibits more than $_{0.50}$07 of parametric gain for both TE and TM modes in the $_{0.50}$08 region and reaches sensitivity of $_{0.50}$09 photon/pulse (Yan et al., 2022). A separate AlGaAs nanowaveguide racetrack microresonator with $_{0.50}$10 demonstrates continuous-wave four-wave mixing wavelength conversion with $_{0.50}$11 pump power, measured conversion efficiency of $_{0.50}$12, a $_{0.50}$13 enhancement over a straight nanowaveguide, and predicted optimized conversion efficiency of $_{0.50}$14 (Kultavewuti et al., 2016). Although these particular experiments are not transversely pumped, they delimit the attainable nonlinear interaction strengths in AlGaAs waveguide microcavities.

5. Ultrafast Kerr switching under transverse pump injection

A distinct operating regime uses the instantaneous electronic Kerr effect to shift a cavity resonance on sub-picosecond time scales. In planar GaAs/AlAs microcavities resonant near $_{0.50}$15, pump-probe reflectivity experiments with a $_{0.50}$16 pump and pulse duration $_{0.50}$17 produce a reversible decrease of the resonance frequency by more than half a linewidth. The switching speed is limited by the cavity storage time $_{0.50}$18 rather than by intrinsic material parameters, and the red shift corresponds to an increase of the GaAs refractive index by $_{0.50}$19 (Ctistis et al., 2011).

The relevant constitutive relation is

$_{0.50}$20

and the resonance shift scales with the Kerr-induced index change as long as free-carrier generation is suppressed. That suppression is achieved by choosing the pump such that the pump photon energy is below half the bandgap and the sum of pump and probe photon energies remains below the bandgap. When the pump frequency is increased to $_{0.50}$21, non-degenerate two-photon absorption generates carriers and the response changes to a slower blue shift (Ctistis et al., 2011).

Subsequent optimization studies show that the observable switched magnitude is not monotonic in $_{0.50}$22. Instead, it is maximized when the pump pulse duration matches the cavity storage time. In GaAs/AlAs and AlGaAs/AlAs planar microcavities operating in the O-band, the shift is reduced when the backbone of the central $_{0.50}$23-layer has a greater electronic bandgap, while pumping with photon energies near the bandgap resonantly enhances the switched magnitude. Under optimized conditions, the model indicates that using an AlGaAs cavity can increase the shift three times to one linewidth (Yüce et al., 2015). This directly constrains transversely pumped AlGaAs waveguide microcavities intended for ultrafast modulation: higher $_{0.50}$24 does not automatically improve switching, and pump wavelength must be chosen with simultaneous regard to Kerr enhancement and multiphoton absorption.

6. Design trade-offs, misconceptions, and limitations

Several recurrent misconceptions are contradicted by the literature. One is that stronger mirrors automatically improve pumping. In fact, conventional quarter-wave DBRs often reject broad-band transverse pumps, so threshold and injection efficiency can be governed primarily by multilayer pump optics rather than by the guided or polaritonic mode itself (Poellmann et al., 2016). A second is that a wider pump-transparency margin always implies better device performance. Modified AlGaAs DBRs do open a useful $_{0.50}$25–$_{0.50}$26 pumping window, but increasing Al content also reduces refractive-index contrast and may require more mirror pairs for the same reflectivity and $_{0.50}$27 (Shih et al., 2023). A third is that deeper etching always improves BIC operation. In the GaAs/AlGaAs BIC waveguide, increasing depth from $_{0.50}$28 to $_{0.50}$29 lowers the condensation threshold, but $_{0.50}$30 etches introduce strong exciton quenching and require ALD passivation to recover condensation in even one structure (Riminucci et al., 2022). A fourth is that higher $_{0.50}$31 always improves ultrafast Kerr switching; the O-band microcavity studies show that the switched magnitude is maximized when pump pulse duration and storage time are matched, not when $_{0.50}$32 is simply maximized (Yüce et al., 2015).

Limitations are equally architecture-specific. In pump-engineered GaAs/AlGaAs microcavities, the enhancement window is angle-sensitive and optimized for a specific material composition and cavity frequency, while transfer-matrix calculations of pump absorption neglect excitonic resonances in the pump response (Poellmann et al., 2016). In the ion–telecom interface, the present signal spectrum overlaps the $_{0.50}$33 Sr$_{0.50}$34 linewidth only at a mismatch factor of about 50, motivating additional cavity engineering of the infrared modes (Simeone et al., 13 Mar 2026). In polaritonic BIC waveguides, nonradiative loss from etched sidewalls remains a central constraint, as shown by cathodoluminescence and by the effectiveness of $_{0.50}$35 Al$_{0.50}$36O$_{0.50}$37 ALD passivation (Riminucci et al., 2022).

Taken together, these studies define the transversely pumped AlGaAs waveguide microcavity not as a single device type, but as a design space in which vertical pump injection, in-plane confinement, and nonlinear or strong-coupling functionality are deliberately separated and then recombined. The mature forms of that design space already include low-threshold polariton condensation in a BIC waveguide, ultrafast Kerr switching in GaAs/AlAs and AlGaAs/AlAs microcavities, high-gain AlGaAs BRW parametric amplification, cavity-enhanced wavelength conversion, and inverse-designed counterpropagating SPDC that directly connects a trapped-ion transition at $_{0.50}$38 to the telecom C-band (Riminucci et al., 2022, Ctistis et al., 2011, Yan et al., 2022, Kultavewuti et al., 2016, Simeone et al., 13 Mar 2026).