Abstract

The III-V heterostructure based on the p-n junction is probably one of the most established semiconductor optoelectronic technology that revolutionized the development of light-emitting diodes (LEDs), lasers and photodetectors. Nonetheless, given the recent miniaturization of III-V light-emitting and detector devices (e.g. micro- and nanoscale LEDs [1], nanolasers, nanophotodetectors), the use of resistive p-type doping plays a crucial role on the devices’ overall resistance and optical losses. Importantly, the p-doping adds substantial cost and complexity to the epi-material growth and device fabrication. Here, we show that nanometric layers of AlAs/GaAs/AlAs forming a double-barrier quantum well (DBQW) (Fig. 1a), arranged in n-type unipolar micropillar or nanopillar array LEDs (Fig. 1b), can provide electroluminescence (EL) (emission at 806 nm from the active DBQW and bulk, Fig. 1c), light-sensing (0.56 A/W at 830 nm, not shown here), and negative differential conductance (NDC) [2] (Fig. 1c, inset) in a single nanophotonic device. Under the same forward bias, we show that enough holes are created in the DBQW to allow for radiative recombination without the need of p-type semiconductor doped layers, as well as pronounced photocurrent generation due to the built-in electric field across the DBQW that separates the photogenerated charge carriers. Time-resolved EL (TREL) reveals decay lifetimes of 5 ns, whereas photoresponse fall times of 250 ns are measured in the light-sensing process, Fig. 1d. The estimated internal quantum efficiency (IQE) of 0.23% in unipolar devices can be improved by using a thinner quantum well and low-doped cladding AlGaAs layers [2], and improved passivation methods by SiN_x thin films [3], resulting in a 40-fold improvement of IQE. The seamless integration of these multi-functions – EL, light-sensing and NDC – in a single nanophotonic device paves the way for compact, on-chip light-emitting and receiving nanocircuits needed for imaging, sensing, signal processing, data communication and neuromorphic computing applications [4].

© 2023 IEEE