Hello dear friends!
Heterojunctions are produced by deposition (often epitaxy) of a semiconductor on an other to achieve excellent components. Silicon being the most commmon material, Si1-x
is used as a material with a different bandgap, despite the drawbacks of germanium: lattice constant very mismatched, main electron valley in <111> direction versus <100>... As opposed, the GaP crystal is known to resemble Si closely: same zincblende lattice (called diamond when the atoms are identical), lattice constants matched to 500ppm, main electron valley in <100> direction too. GaP's bandgap is also bigger and differs more from Si, nice.http://www.ioffe.ru/SVA/NSM/Semicond/Si/index.htmlhttp://www.ioffe.ru/SVA/NSM/Semicond/GaP/index.htmlhttp://www.ioffe.ru/SVA/NSM/Semicond/Ge/index.html
So it is seducing to produce epitaxies of GaxPxSi1-2x on Si
. The less distorted crystal should conduct better, and the valley in <100> direction should accept a bigger proportion x. Since these heteroepitaxies aren't common up to now, many difficulties must arise; I try to address one here, the exactly equal amount of Ga and P
. It won't happen naturally because Si crystals readily host the small P atoms but not Ga, and the resulting doping in Si must be so huge that no component can be built.
My suggestion is to use an already formed GaP single-crystal as an epitaxy source
. Semiconducting GaP is exactly balanced, perfect for a usable epitaxial layer. I imagine something like a short-pulsed laser could evaporate GaP to keep the proportion, and evaporate Si from an other source, for simultaneous deposition on the epitaxy target.
The GaP source may evaporate one element more easily at the beginning, despite the laser pulse atomizes a volume of solid with almost no selectivity. The initial surface of GaP may also be less clean. The answer is to have a shutter and begin the deposition some time after beginning the evaporation.
If the evaporated P and heavier Ga atoms fly to the epitaxy target for 100µs and 150µs (as an example) after the laser pulse, the epitaxy composition may fluctuate over the time hence the depth. The answer is a pulse repetition rate much faster than said 100µs. As 100µs flight time may spread as a 50µs Gaussian, 50ns repetition rate will smooth the distribution to far better than 1ppm.
The evaporation rate of Ga and P differ at the epitaxy target, which shouldn't hence be too hot.
Marc Schaefer, aka Enthalpy