While III–V light sources grown on silicon through heteroepitaxy suffer from inherent lattice mismatch and surface defects (as well as requiring high process temperatures which may damage underlying CMOS circuitry), 2D MoTe2 layers can be directly adhered to a silicon substrate via van der Waals interactions.
Publishing their results in the Nature Nanotechnology letters under the title "A MoTe2-based light-emitting diode and photodetector for silicon photonic integrated circuits", the MIT researchers describe a silicon waveguide-integrated light source and photodetector based on a p–n junction of bilayer MoTe2, emitting at 1160nm (infrared).
The MoTe2 p–n junction consists in an exfoliated bilayer of MoTe2 separated by a hexagonal boron nitride (h-BN) dielectric layer from a dual graphite gate. Instead of relying on dopants (typically seen as crystal impurities), the p- and n-type doping is induced electrostatically by applying different voltages to the graphite gates. The on-demand electrostatic split-gate configuration allows for diverse functionalities to be programmed, including transistors, light-emitting diodes (LEDs) and photodetectors, coupled with an underlying silicon photonic-crystal (PhC) waveguide (a holey silicon membrane).
For their demonstration, the researchers fabricated a grating coupler at the far end of the waveguide. When the p–n junction is operated as a LED, the emitted light couples to the waveguide where it travels in-plane to the grating coupler where it can be output. In the photodetector mode, incident light coupled into the waveguide by the grating coupler can be detected by the p–n junction (converting it to an electrical signal).