Growth of III-V semiconductor lasers on silicon

Abstract

The growth of reliable III-V semiconductor lasers on Si would be a significant step toward the fabrication of an opto-electronic integrated circuit, but reliable III-V semiconductor lasers grown on Si have yet to be reported. Many problems not encountered when growing III-V lasers on III-V substrates must be overcome before III-V lasers grown on Si become reliable. In the experiments described here many different methods are used to try to improve the reliability of III-V lasers grown on Si.

Data are presented showing that impurity induced layer disordering (IILD) can be greatly accelerated when an abundance of dislocations are present. The high dislocation density of III-V materials grown on Si makes this a problem that cannot be overlooked. This problem can be minimized but it puts limits on the kind of thermal processing that can be done to a III-V laser grown on Si. Perhaps the most difficult problem encountered when growing III-V lasers on Si is dislocations. If the dislocation density can be reduced, then lasers grown on Si should become more reliable. Several methods of reducing the dislocation density of III-V lasers grown on Si have been tried. Data are presented showing the Zn diffusions, strained layer superlattices, and lattice matched buffer layers are all somewhat effective at lowering the dislocation density. The high dislocation density of III-V lasers grown on Si makes InGaAs strained layer active lasers on Si an unlikely possibility because the dislocations tend to accumulate at the active region of these structures where dislocations are the most damaging.

Growth of III-V lasers on patterned GaAs-on-Si wafers has also been investigated. Growing over a selective area should reduce some strain and possibly lower the dislocation density. Data are presented showing that the dislocation density can be reduced by selective area epitaxy. But other problems arise when growing over a selective area and make device fabrication more difficult.