Digital-Alloy Bragg Mirrors in High-Q Microcavities for Polariton Lasing

TL;DR

The paper addresses the challenge of achieving ultrahigh-quality GaAs microcavities by replacing conventional AlGaAs DBRs with short-period GaAs/AlAs digital-alloy superlattices. It combines meticulous MBE growth, transfer-matrix modeling with a nonlocal excitonic response, and cryogenic optical characterization to show that SPSL DBRs dramatically reduce interface roughness, enable precise λ/4 periodicity, suppress dislocation propagation, and allow absorption tuning. The optimized MC2 structure achieves a polariton lasing threshold of approximately $P_{th} \approx 570~\mathrm{W/cm^{2}}$ with a measured $Q_{exp} \approx 5.4\times 10^{4}$, nearly double the simple ternary-alloy prediction and underscoring the need for SPSL-specific refractive-index modeling that includes quantum confinement and excitonic effects. Overall, the digital-alloy approach provides a scalable route to high-$Q$ polariton devices and yields deeper insights into SPSL optical properties for device design.

Abstract

We present an approach to the molecular-beam epitaxy of high-Q planar GaAs-based microcavities in which the AlGaAs high-index layers of the distributed Bragg reflectors (DBRs) are replaced by short-period GaAs/AlAs superlattices (digital alloys) with similar optical properties. This design enables a significant reduction of interface roughness, precise control of the quarter-wavelength optical thickness and the effective Al content, suppression of the propagation of structural defects, and efficient tuning of intrinsic absorption at the polariton emission wavelength via optimization of the superlattice parameters. Using this approach, we fabricate a microcavity with a low polariton-lasing threshold of approximately 570 W/cm$^2$ and a high experimental quality factor of about 5.4 x $10^4$. This value exceeds by almost a factor of two the theoretical estimate obtained within an equivalent ternary-alloy model. We demonstrate that accurate modeling of the stop-band characteristics and the Q factor requires incorporating the modified electronic density of states in the superlattice, including quantum-confinement and excitonic effects.

Figures (8)

• Figure 1: Loss channels in a planar GaAs/Al$_x$Ga$_{1-x}$As semiconductor microcavity that limit the achievable $Q$ factor. $Q_{\mathrm{theor}}$ denotes radiative leakage through the mirrors. $Q_{\mathrm{rough}}$ denotes scattering due to interface roughness. $Q_{\mathrm{per}}$ denotes losses associated with deviations from the $\lambda/4$-periodicity in the DBRs. $Q_{\mathrm{sd}}$ denotes scattering from structural defects. $Q_{\mathrm{abs}}$ denotes intrinsic absorption in the constituent materials.
• Figure 2: Dependence of the microcavity $Q$ factor on the number of pairs in the bottom DBR (the top DBR contains five fewer pairs) for different values of the absorption coefficient $\kappa$.
• Figure 3: (a) Experimental (red) and theoretical (blue) reflectance spectra of the MC1 structure at a small negative detuning. (b) PL spectrum at small negative detuning measured under nonresonant excitation ($E_{\mathrm{pump}} = 1.65\eV$) showing the lower and upper polariton branches together with impurity-related transitions of the SPSL.
• Figure 4: Optical characterization of half-microcavities with the SPSL-based DBRs and the GaAs layer thicknesses of 5.1nm (HMC1) and 4nm (HMC2). (a, b) Reflectance spectra showing the stop band of the half-microcavity and the heavy-hole ($\mathrm{QW_{Xhh}}$) and light-hole ($\mathrm{QW_{Xlh}}$) exciton transitions in the GaAs quantum well. (c, d) PL spectra under excitation at 2.33eV, revealing the superlattice exciton resonances ($\mathrm{SL_{Xhh}}$) and an impurity tail ($\mathrm{SL_{imp}}$).
• Figure 5: Atomic force microscopy (AFM) images of the surface of the HMC3 (a) and MC1 (b) samples. The HMC3 structure, which employs a ternary-alloy-based DBR, exhibits pronounced surface roughness with an RMS value of $\sim$ 1.1nm (c). In contrast, the MC1 structure with an SPSL-based DBR shows a much smoother surface with an RMS value of $\sim$ 0.1nm (d). Panels (e) and (f) show Nomarski micrographs of the HMC3 and MC1 samples, respectively.