Localized electromechanical interactions in ferroelectric P(VDF-TrFE) nanowires investigated by scanning probe microscopy

We investigate the electromechanical interactions in individual P(VDF-TrFE) nanowires in response to localized electrical poling via a conducting atomic force microscope tip. Spatially resolved measurements of piezoelectric coefficients and elastic moduli before and after poling reveal a striking dependence on the polarity of the poling field, notably absent in thin films of the same composition. These observations are attributed to the unclamped nature of the nanowires and the inherent asymmetry in their chemical and electrical interactions with the tip and underlying substrate. Our findings provide insights into the mechanism of poling/switching in polymer nanowires critical to ferroelectric device performance.

to be insufficient for poling these films, as shown in Supporting Information Sec. III., and thus only the 100 nm-thick films were used for the localised poling studies. The samples were mounted on a 1 cm magnetic disc suitable for conductive AFM measurements using a Bruker Multimode 8 (with Nanoscope V controller) microscope. Calibration for both PFM and QNM modes was performed using dedicated calibration samples (PFM-SMPL and PSFILM-12M by Bruker ) as described in Supporting Information Sec. III.  Figure 1d shows the uncalibrated lateral PFM signal amplitude (i) and phase (ii) recorded simultaneously with the corresponding vertical PFM signal for TF. Figure 1e shows the lateral PFM amplitude from the NW, obtained simultaneously with [b ii]. The extracted results for the effective piezoelectric coefficient, d eff , from the NW and TF samples, with different tip-velocities during poling, and different times after the poling are presented in Table I.
Several important observations can be made following these experiments and from the results presented in Table I:(i) the pristine "unpoled" regions of the TF and NW had different d eff values. While the films exhibited values close to zero as expected, the NWs exhibited a non-zero, down-poled oriented value. This can be attributed to the inherent self-poled nature of the NW, as has been discussed at length previously 16 . (ii) The efficiency of the poling procedure in NWs was found to be velocity-dependent and asymmetric with regards to up-and down-poling (with down poling usually stronger). While symmetric poling was achieved in the TF sample, NW poling remained asymmetric even with a slower scan. (iii) The poled regions in the TF gave rise to distinct lateral PFM signals that were mostly absent    Fig. 1). This region is located away from the NW, since the down-poling process resulted in a significant displacement of the NW relative to its original location (where the residue was found). Following the successful location of the poled regions (Fig. 3c), a slow, high-resolution scan (0.1 Hz, 1024 samples/line) was performed (Fig. 3d). is not yet a full understanding of these effects. Of these, we rule out the effect of electrode surface oxide, as this was shown to be considerably reduced when using gold electrodes (as in our case) 11,35 . However, Schottky effect might induce preferential (and self-) poling to the material due to the differences in work-function (see also SI 37,38 . A basic distinction between the NW and the TF configurations is the non-clamped nature of the NW on the substrate, both laterally and relative to the substrate 39,40 . The absence of lateral PFM from the poled regions of the NW, compared to the clear, polarization dependent signal from the TF sample ( Fig. 1 and Supp. S3), is an interesting manifestation of this.
This result is surprising, considering that P(VDF-TrFE) β-phase crystal structure results in the preference of 60 • rotations of the horizontal dipole configuration (see Fig. 4) 11,28 .
Therefore some lateral signal is expected during up-or down-poling, and its absence in the NWs is intriguing. In addition, the distinct nano-indentation of NWs and TFs also demonstrates this, as shown by finite-element simulations (see Supporting Information, Fig. S6).
If so, this lack of constraint is expected to allow relaxation of the material to more stable configurations, and indeed we observe changes in the extracted d eff values obtained from the NWs in the range of minutes, while these changes are notably reduced from the TF samples. It might also account for the larger values usually measured (and in this work as well) for NW compared to thin films. When considering a NW section being poled, the energy needed to pole an equivalent clamped section would likely be higher than the energy needed to overcome the restriction imposed by the clamped configuration.
The lack of constraints may also explain the increased mechanical freedom of the NW system, wherein the asymmetries between the two polarization states are more pronounced.
As mentioned above, the tip-NW-substrate is asymmetrical; in addition, the P(VDF-TrFE)/metal interface is inherently asymmetrical, as the molecular layer closest to the metal is different for each polarization, and for the pristine NW. Apart from work function related phenomena, one can consider the molecular nature of the interface. should not result in an inherent asymmetry, therefore it is reasonable that the former case is prevalent; however, we cannot currently distinguish between the two alternatives, and future work will be concerned with this issue. The local damage observed in some of the up-poled regions may then be due to increased mechanical stresses induced by the presence of conflicting forces in these regions, resulting perhaps in amorphization or local loss of ferroelectric ordering, and thus the correspondingly reduced d eff extracted. This is also consistent with the observed reduced elastic modulus in this region, as the elastic modulus of polymers is expected to increase with crystallinity 23, 48 . The sudden µm-scale movement of the NW during poling (Fig. 2) could be related to a non-linear response and the instantaneous expansion of the domain such that mechanical stresses resulting from the switch are released through this movement, which is otherwise not possible in TF samples.
Recently, Guo and Setter have developed a model 28 based on bulk (or volume) dipoledipole interactions to explain thickness dependent preferred orientations of P(VDF-TrFE) TFs. By summing up the interaction energies of the horizontal and the 60 • rotated orientations (Fig. 4b), for a given volume, they point to a cross-over of the energies when the thickness of the TF is comparable with the grain size; thus inducing preference of the horizontal orientation for "very" thin films 28 . In contrast to TFs, NWs inherently have similar We attribute these observations to the different electronic and chemical surface interactions of the P(VDF-TrFE) with the substrate and/or ambient, which prevail due to the uniform geometry of the NW, and the reduced mechanical clamping which otherwise plays a significant role in TF samples. We believe that these result shed light on the electromechanical interplay in regards to ferroelectric P(VDF-TrFE) NWs, as well as on fundamental processes in this material that are not commonly observed when studying TFs.