InGaAs is the workhorse III-V alloy for anything that has to emit, detect or amplify light at telecom wavelengths, where silicon is blind, and the live question is not whether it works but whether it can be integrated onto silicon and silicon nitride photonic chips at scale.
Summary
Indium gallium arsenide (InGaAs) is a ternary compound semiconductor, an alloy of indium arsenide and gallium arsenide whose bandgap can be tuned by composition. Its practical significance is that the bandgap can be pushed well below silicon’s, so InGaAs absorbs and emits at the 1550 nm telecom band and beyond, where silicon photodiodes are transparent. Source material notes this bandgap limitation directly: silicon single-photon avalanche diodes are mature and run at room temperature but cannot reach 1550 nm, while InGaAs-based SPADs can. It is normally grown epitaxially on indium phosphide substrates and used in heterostructures with InP, InAlAs or GaAs, which supply the barriers, passivation and lattice template.
The material appears in the literature in three distinct roles. First, as a detector absorber: gated InGaAs/InP SPADs are described as the mature non-cryogenic alternative to superconducting detectors for quantum key distribution, and GHz-gated arrays have been coupled to low-loss silica waveguides to build hybrid QKD receivers delivering secure key rates above 2 Mbps at short distance and 15 kbps over 100 km of fibre. Second, as an emitter: self-assembled InGaAs quantum dots are among the leading on-demand single-photon and entangled-photon sources, including energy-time entanglement from an InGaAs/GaAs dot grown monolithically on silicon. Third, as an active electronic layer bonded onto other platforms, for example epitaxial InGaAs transferred onto lithium niobate on insulator to build a surface-acoustic-wave acoustoelectric amplifier.
The parameters that decide InGaAs’s future are not exotic. For detectors: quantum efficiency, dark count rate and afterpulsing, all of which the room-temperature van der Waals detector work cites as InGaAs weaknesses alongside environmental hazards in fabrication. For integration: whether InGaAs can be grown or bonded onto silicon and silicon nitride without losing performance, with a heterogeneously integrated InGaAs-on-InP uni-travelling-carrier balanced photodetector on thin-core silicon nitride reported at 0.92 GHz bandwidth. For reliability: defect and surface chemistry, where trap states in an InP/InGaAs stack on silicon become increasingly active below 140 K and suppress current down to 5 K, and where ambient oxidation of a bare InGaAs surface extinguishes acoustoelectric gain within weeks unless capped with InP or ALD alumina.
Evidence base
- GHz-gated InGaAs/InP SPAD arrays coupled to low-loss silica waveguides produced compact hybrid QKD receivers with secure key rates over 2 Mbps at short distance and 15 kbps over 100 km of fibre.
- A self-differencing gating scheme on an InGaAs photodiode at 100 MHz recovered full quantum efficiency in under one pulse-repetition period, enabling dead-time-free continuous operation.
- A cryogenic InGaAs-InP photodiode biasing an SNSPD at about 2.3 K gave 80.7% system detection efficiency, 32.6 cps background dark counts and 57.5 ps minimum jitter.
- Energy-time entanglement was demonstrated from a single InGaAs/GaAs quantum dot grown monolithically on silicon, with Franson visibility of 64.0 plus or minus 7.0% for an 80 ps window.
- A heterogeneously integrated InGaAs-on-InP modified uni-travelling-carrier balanced photodetector on a 15 metre thin-core silicon nitride Mach-Zehnder interferometer achieved 0.92 GHz bandwidth.
- Epitaxial InGaAs bonded to X-cut lithium niobate on insulator gave 32 dB/mm stable non-reciprocity at 1.11 GHz for passivated devices, with bare-surface oxidation extinguishing the response within weeks.
- InGaAs SPADs are documented as sensitive at 1550 nm but suffering lower efficiency, high dark count rate, afterpulsing and fabrication environmental hazard, motivating a room-temperature black phosphorus alternative.
Open questions
- Can InGaAs/InP SPAD arrays scale beyond the demonstrated hybrid silica-waveguide receivers to monolithic pixel counts while holding negligible crosstalk and GHz gating?
- Do the trap states that suppress current in InP/InGaAs-on-silicon stacks below 140 K yield to growth or annealing changes, or do they cap cryogenic optical interconnect performance?
- Does the room-temperature van der Waals 1550 nm single-photon detector reproduce at array scale with dark-count and efficiency figures that beat InGaAs SPADs in a real link?
- Can monolithic InGaAs quantum dots on silicon push Franson visibility past the Bell violation threshold at long integration windows, not only at 80 ps?
Assessment drafted 2026-08-31 from up to 15 KB sources using the technology-scorecard framework; scores are a draft read pending review.