Choice of tip, signal stability and practical aspects of Piezoresponse-Force-Microscopy

TL;DR

This study evaluates the reproducibility and stability of Piezoresponse-Force-Microscopy signals, highlighting the benefits of diamond coated tips and proper sample polishing for accurate ferroelectric measurements.

Contribution

It demonstrates that diamond coated tips improve PFM signal stability and background noise, and emphasizes the importance of sample polishing for reliable PFM quantification.

Findings

Diamond coated tips reduce signal deviation to 38.3%.

Background signals account for 11.6% of domain contrast.

Proper polishing significantly enhances PFM measurement accuracy.

Abstract

Piezoresponse force-microscopy (PFM) has become the standard tool to investigate ferroelectrics on the micro- and nanoscale. However, reliability of PFM signals is often problematic and their quantification is challenging and thus not widely applied. Here, we present a study of the reproducibility of PFM signals and of the so-called PFM background signal which has been reported in literature. We find that PFM signals are generally reproducible to certain extents. The PFM signal difference between 180{\deg} domains on periodically-poled lithium niobate (PPLN) is taken as the reference signal in a large number of measurements, carried out in a low frequency regime (30-70 kHz). We show that in comparison to Pt coated tips, diamond coated tips exhibit improved signal stability, lower background signal and less imaging artifacts related to PFM which is reflected in the spread of measurements. This is attributed to the improved mechanical stability of the conductive layer. The average deviation of the mean PFM signal is 38.3%, for a diamond coated tip. Although this deviation is relatively high, it is far better than values from literature which showed a deviation of approx. 73.1%. Additionally, we find that the average deviation of the background signal from 0 is 11.6% of the PPLN domain contrast. Thus, the background signal needs to be taken into account when quantifying PFM signals and should be subtracted from PFM signals. Those results are important for quantification of PFM signals, since PPLN might be used for this purpose when PFM signals measured on PPLN are related to its macroscopic d33 coefficient. Finally, the crucial influence of sample polishing on PFM signals is shown and we recommend to use a multistep polishing route with a final step involving 200 nm sized colloidal silica particles.