Dry Etching Damage and Alloy Composition Analysis of GaN-Based Semiconductors Using Electron Energy-Loss Spectroscopy

Dry etching damage and compositional fluctuation of GaN-based semiconductors and alloys are major issues to overcome to improve device fabrication processes and performance. Electron energy-loss spectroscopy (EELS) in a transmission electron microscope is a useful method for studying these phenomena because it can analyze the elemental composition and chemical bond states of a sample with high spatial resolution. However, because these various components of information are superposed in an EELS spectrum, the analysis has to be performed carefully. This requires multivariate statistical techniques, such as principal component analysis (PCA), to process the large amounts of data extracted from EELS. Herein we apply PCA to N-K core and In-M4,5 core edge EELS spectra to study the degree of ion-beam etching damage of GaN and the composition of GaInN ternary alloy quantum wells. Two data pretreatment procedures prior to the PCA analysis were examined to assess their comparative practical usefulness. The differential-based pretreatment method for PCA was useful when the shape of the EELS core edge structure was changed, whereas the power-law subtraction-based pretreatment method was useful when the intensity of the EELS core edge structure was variable.