With expanding applications and growing performance requirements in Power, RF and Optoelectronics markets, leading device manufacturers are looking for new ways to characterize yield-limiting defects that will help them achieve faster development and ramp times, higher product yields and lower device costs. Full-surface, high sensitivity defect inspection and accurate process control feedback has enabled the industry to improve substrate quality as well as to optimize the yields on epitaxy growth processes.
     As device manufacturers continue to push the boundaries of process designs, the requirements for defect inspection and overall yield management become increasingly more stringent and critical. The Candela unified surface and photoluminescence (PL) defect inspection platform enables high sensitivity inspection and defect classification at production throughputs of a wide range of critical defects (e.g. micro scratches, stacking faults, basal plane dislocations) and effectively separates front-surface defects and buried defects on transparent SiC substrates and epitaxial material. In addition, automated defect classification capabilities reduce the time required to identify, source and correct various yield-limiting defects such as carrots, triangles, sub-micron pits and others.
     The process of growing III-V epitaxy has unique challenges. The large mismatch in the lattice constant and the thermal expansion coefficient between epitaxy layer and substrate causes high lattice stress which leads to cracking on and through the epitaxy layer, making parts of the wafer unsuitable for device production. This cracking can be minimized by using a suitable buffer layer and optimizing the epitaxy reactor conditions. Improper epitaxy reactor conditions may also cause other device reliability killer defects such as micropits, craters, epi droplets and/or bumps.
     This study discusses how multiple complementary techniques such as scatterometry, reflectometry, ellipsometry and photoluminescence could be used together for simultaneous detection and classification of multiple critical defects on compound semiconductor wafers. We demonstrate how feedback from defect inspection equipment can be used to screen incoming substrate wafers and to monitor and optimize the performance of CVD reactors during the epitaxy process.