Rapid subsurface detection of nanoscale defects in live microprocessors by functional infrared emission spectral microscopy.

We demonstrate the rapid and nondestructive detection of subsurface nanometer-size defects in 90 nm technology live microprocessors with a new technique called functional infrared emission spectral microscopy. Broken, leaky, and good transistors with similar photoemission images are identified from each other by their different emission spectra that are calculated as linear combinations of weighted basis spectra. The basis spectra are derived from a spectral library by principal component analysis. Leaky transistors do not exhibit apparent morphological damage and are undetectable by optical or scanning probe microscopy alone. The emission signals from two or more transistors combined incoherently, and defect detection is primarily limited by the signal-to-noise ratio of the detected spectrum and not by the separation distance of neighboring transistors.