The second-order nonlinear optical susceptibility χ⁽²⁾ determines how efficiently a material converts light from one frequency to another — second harmonic generation, parametric down-conversion, sum and difference frequency generation. Larger χ⁽²⁾ means more efficient conversion, which means smaller devices, lower power requirements, and stronger quantum correlations in the generated photon pairs. In bulk crystals, χ⁽²⁾ is fixed by the crystal structure and the electronic band structure. You get what nature provides.
Ramesh, Brown, Ricks, and collaborators (arXiv 2602.23246, February 2026) demonstrate a material where χ⁽²⁾ is designed rather than inherited. Asymmetric coupled quantum wells in AlGaAs/GaAs — thin layers of semiconductor with different well widths and barrier heights — produce interband optical transitions with broken inversion symmetry. The asymmetry, engineered at the atomic layer level during epitaxial growth, generates a second-order nonlinear response that doesn't exist in either constituent material on its own.
GaAs has a bulk χ⁽²⁾ of about 360 pm/V at telecommunications wavelengths — already large by nonlinear optical standards. The coupled quantum well structure achieves χ⁽²⁾ of 2750 pm/V at 1550 nm, more than seven times the bulk value. The enhancement comes from engineering the quantum well transitions to maximize the asymmetric matrix elements between the interband states. The quantum wells concentrate the oscillator strength at specific energies and create asymmetric charge distributions that conventional interband transitions in bulk crystals cannot produce.
The design space is large. The well widths, barrier heights, and layer compositions are all adjustable parameters that control the transition energies, oscillator strengths, and asymmetry. Quantum mechanical calculations — solving the Schrödinger equation for the heterostructure and computing the nonlinear susceptibility from the resulting wavefunctions — predict the χ⁽²⁾ before growth. The experimental measurements match the calculations, confirming that the enhancement is understood quantitatively and can be optimized further.
The wavelength matters. 1550 nm is the telecommunications C-band — the low-loss window of optical fiber. Second harmonic generation from 1550 to 775 nm, and parametric down-conversion from 775 to 1550 nm, are the processes needed for generating entangled photon pairs at telecom wavelengths for quantum key distribution and quantum networking. Bulk crystals used for this purpose (periodically poled lithium niobate, for example) work well but are separate components that must be coupled to semiconductor photonic circuits. A semiconductor source of entangled photons, integrated on the same chip as the detectors and modulators, would simplify the architecture.
The material is the device. The quantum wells don't need to be placed inside a cavity or waveguide to generate the nonlinear response — the response is in the material itself. Integration into a photonic circuit adds resonant enhancement on top of the intrinsic material improvement.