switching biases during the flickering portion of the measurement is assigned as that pixel’s switching scan. This procedure allows the maps of the local switching bias shown in figure5.5 to be generated. The colour represents the tip bias triggering the local events. From these, polarisation reversal event sizes can be extracted as the area of individual regions sharing the same switching bias. As will be discussed in more detail in section5.4, these local switching bias maps also allow domain nucleation, motion, and merging events to be distinguished.
5.3 Switching at a glance
Figure 5.5:Local polarisation switching maps extracted from alternating switching scans and PFM measurements under increasing tip bias. The left (a,b) and right (c,d) columns show the switching bias in PZT-Nuc and PZT-Mot respectively. The top (a,c) and bottom (b,d) rows show switching maps acquired with a negative and positive tip bias respectively, corresponding to switching from written down-polarised domains to up and from as-grown up to down.
In this section, the differences in polarisation switching between PZT-Nuc and PZT-Mot are discussed in detail. Qualitative differences in the switching mechanisms can be clearly seen between the two samples, as well as between switching from the as-grown (up to down) and the written (down to up) polarisation states under positive and negative tip biasVtip, respectively. For PZT-Nuc, polarisation reversal proceeds predominantly by the nucleation and growth of new domains, with relatively little motion of the pre-existing walls of the initial domain configuration. This is particularly noticeable under
negative tip bias, shown in figure 5.5(a), where the two side walls of the down-oriented central stripe domain move inwards by only 20 nm between -1.0 V and -1.5 V. Beyond this threshold, multiple new up-oriented domains nucleate in the central region and grow rapidly outwards, gradually merging until almost complete polarisation reversal at -4 V. Under positive tip bias, we observe significantly higher activation thresholds for polarisation switching via domain wall motion, which sets in at 3 V. The motion contributes to somewhat larger 100 nm inward displacements of the existing domain walls.
Subsequently, very rapid growth of new domains via point nucleation occurs beyond 3.75 V, leading to almost complete polarisation reversal. For PZT-Mot, meanwhile, polarisation reversal under both positive and negative tip bias proceeds almost exclusively via the motion of pre-existing domain walls seen in figure5.5(c,d). For positive tip bias, these walls, initially relatively flat, begin moving via small displacements of around 25 nm at close to ±1 V. For negative tip bias, the motion appears quite regular and generalised to the entire domain walls, while for positive tip bias the displacements remain extremely limited until approximately 2.3 V is reached, at which point large jumps can be observed, leading to complete polarisation reversal. Under negative tip bias, the domain walls also appear to roughen more noticeably while they move.
The effect of domain writing history on the overall switching behaviour can also be seen from the evolution of the proportion of the switchable surface. This is calculated as the ratio of the switched area to the total area in which the polarisation is initially oriented opposite to the applied electric field, and shown in figure 5.6(a). In both samples, onset of domain wall motion occurs at a higher bias when switching from the as-grown polarisation state with positive bias, and the switched surface proportion first increased slowly close to the onset of switching, then much more rapidly at higher bias. At negative bias, corresponding to switching written domains where additional defects caused by the writing are expected to provide additional pinning to the domain walls, the switching is much more gradual. In both samples, the switching rates clearly show a difference between a relatively limited regime of domain wall motion at lower positive bias, and then more rapid switching, whether by further domain wall motion or nucleation and growth of new domains at higher bias values.
The influence of domain writing history is further observed in the dif-ferences between the first and second local hysteresis loops, taken with a stationary tip and averaged over 25 locations, acquired by switching spec-troscopy PFM [129] and shown in figure 5.6(b). Changes in the positive bias branch in the second loop seem to suggest that the high bias applied during the first loop affected the local switching thresholds in the films.
Because of these differences in the overall switching dynamics, the event size distributions are extracted separately for positive and for negative tip bias
5.3 Switching at a glance
Figure 5.6: (a) Normalized switching rates in PZT-Nuc and PZT-Mot samples, showing earlier onset of switching at negative bias, but more rapid full polarisation reversal at positive bias. Changes in switching rates corresponding to the onset of rapid domain nucleation and merging can also be observed. (b) First and second SSPFM hysteresis loops averaged over 25 locations. Changes of the positive coercive bias between the second and first loop are indicative of injection and redistribution of defects, known to occur in atomic force microscopy measurements on ferroelectric thin films [114,135,160].
in each sample.
Figure 5.7: Maps showing different types of switching events - domain nucleation (red), domain wall motion (blue) and domain merging (green) - in the PZT-Nuc (a,b) and PZT-Mot (c,d) samples at negative and positive tip bias, respectively. The
scalebars are 200 nm.