Application of Bayesian Machine Learning To Create A Low-Cost Silicon Failure Mechanism Pareto

The increasing challenges with relying on Physical Failure Analysis and inline inspection for ramping the yield are the reason that Volume Scan Diagnostics Analysis (VSDA) has become a mainstream methodology that supplements traditional yield learning. Because scan diagnostics are inherently noisy, the results often require expert knowledge to manually select the location that has the highest likelihood of being correct. In this paper, Failure Mechanism Analysis (FMA) applies the technique of Bayesian Machine Learning in a yield analysis system that can empirically estimate sources of yield loss using physical diagnostic information.