High-precision local C–V mapping for ferroelectrics using principal component analysis

Abstract

Local C–V mapping based on scanning nonlinear dielectric microscopy (SNDM) is a useful tool for characterizing ferroelectric domain dynamics at the nanoscale. In this study, we realized high-precision C–V mapping through an improved measurement system that introduces a digitizer and post-signal processing. Coupled with the high capacitance detection sensitivity of SNDM, nontrivial patterns corresponding to the polarization response were observed even in harmonic images with an order as high as thirty. Using these high-order harmonic components for the C–V curve resynthesis led to improved measurement accuracy. In addition, further improvement was achieved by introducing noise reduction based on principal component analysis. From the C–V curves resynthesized in this manner, parameters that represent the curve features could be extracted and subsequently displayed in two-dimensional maps. Detailed analyses of the datasets obtained with a ferroelectric HfO2 film revealed interesting distributions concerning local polarization switching dynamics.