Masking or Enhancing Physical and Chemical Contributors to the PFM Response

Resonant Piezoresponse Force Microscopy (R-PFM) techniques provide a wealth of information with respect to the electro-chemo-mechanical response of material (surfaces) through acquisition of amplitude (A), phase (θ), resonance frequency (ω), and quality factor (Q), associated with the amplitude and phase of the cantilever deflection, the viscoelastic properties at the tip-surface contact, and the quality factor of the contact resonance, respectively. However, the large amount of information thus acquired has been traditionally difficult to interpret and correlate to the materials` surface characteristics, except through the user`s visual interpretation. Further complicating such interpretations is the continued discovery of the various chemical and physical contributors to the PFM acquired signal, ranging from the piezoelectric response to electrostatic effects, charge injection, Joule heating, etc.
Most recently, machine learning approaches have been used successfully to identify clusters and areas of different contributors, in attempts to understand the piezoelectric response of different materials, as correlating to the underlying chemical (e.g., local changes in the chemical composition) and physical (e.g., presence of or proximity to domain boundaries and point defects, domain polarity, etc.) variations. However, in absence of a fingerprint of the various contributor, the separation of the piezoelectric correlation rather than simple "PFM response" remains unclear. Here we show that dimensional stacking and scaling of the acquired parameters can be leveraged to separate different contributors to the R-PFM parameters, enabling fingerprinting of various contributors (such as electrostatic effects or topographic cross-talks) and subsequently their masking within the R-PFM signal.