Fabrication and characterization of arrays of GaAs all-optical logic gates

Abstract

Optical bistable elements fabricated from gallium arsenide (GaAs) have demonstrated potential as optical logic elements for use in optical parallel processing systems [1]. In previous experiments, the devices have consisted of small, uniform slabs of active material, either bulk GaAs or GaAs-AlGaAs multiple-quantum-well structures [2]. These were placed between mirrors to form nonlinear Fabry-Perot etalons. The active regions of these earlier devices were grown with top and bottom “windows” of AlGaAs to reduce the exciton surface-recombination rate. The overall switching time of these devices was limited by the carrier relaxation time to several nanoseconds. The present devices were produced from a 1.5-μm-thick GaAs layer, grown by molecular beam epitaxy (MBE) on a single-crystal GaAs substrate with an underlying 0.2-μm-thick Al₀.₄Ga₀.₆As etch stop layer, but no top window. The absence of the window resulted in an order-of-magnitude reduction in the recovery time [3]. Some of the wafer was etched to form arrays of small GaAs mesas or “pixels” on a single substrate. The etched surface structure served to further reduce the recovery time, as well as producing an array of separated devices on a single substrate [4]. Each individual pixel of the array measured 9 μm × 9 μm, and the pixels were spaced 20 μm apart from each other (Fig. 1). Individual samples of the array, as mounted in etalons, typically contained several thousands of pixels. However, the fabrication process used for these devices can be easily scaled to larger sizes. Large arrays of optically bistable elements may become important components of powerful optical parallel processing systems. Such processors could make efficient use of the inherent parallelism of optics.