Panithan Sriboriboon1,Huimin Qiao1,2,Owoong Kwon1,2,Rama Vasudevan3,Stephen Jesse3,Yunseok Kim1,2
Sungkyunkwan University1,Sungkyunkwan University (SKKU)2,Oak Ridge National Laboratory3
Panithan Sriboriboon1,Huimin Qiao1,2,Owoong Kwon1,2,Rama Vasudevan3,Stephen Jesse3,Yunseok Kim1,2
Sungkyunkwan University1,Sungkyunkwan University (SKKU)2,Oak Ridge National Laboratory3
Hafnium oxide-based ferroelectrics have been extensively studied because of their existence of ferroelectricity, even in ultra-thin film form. However, studying the weak response of the ultra-thin film requires high-sensitivity piezoresponse measurements. In general, resonance-enhanced piezoresponse force microscopy (PFM) has been used to characterize ferroelectricity by fitting between the acquired piezoresponse spectrum and a simple harmonic oscillation model. However, an iterative approach, such as traditional least square (LS) fitting, is highly sensitive to noise and can result in the misinterpretation of weak responses. In this study, we introduce a deep denoising autoencoder (DDA) and principal component analysis (PCA) to hybridize with a deep neural network (DNN) for the extraction of piezoresponse information. The DDA/PCA-DNN improves the PFM sensitivity down to 0.3 pm, allowing measurement of weak piezoresponse with low excitation voltage in 10-nm-thick Hf<sub>0.5</sub>Zr<sub>0.5</sub>O<sub>2</sub> thin films. Our hybrid approaches could provide more chances to explore the low piezoresponse of the ultra-thin ferroelectrics and could be applied to other resonance-based microscopic techniques.