Direct observation of nanoscale dynamics of ferroelectric degradation

Ferroelectric materials, known for their polarization reversal under applied electric fields, are widely used in electronic nanodevices. However, repeated electric field application leads to ferroelectric degradation, a polarization reversal failure that significantly impacts the reliability and longevity of these devices. This degradation manifests as changes in coercive field, remnant polarization, and resistivity under cyclic electric loading. A comprehensive understanding of the mechanisms and evolution of ferroelectric degradation is crucial, especially given the increasing use of nanoscale devices. Several mechanisms have been proposed to explain ferroelectric degradation, including domain wall pinning (charge carriers/defect dipoles accumulate at domain walls, locking domains), domain nucleation suppression (injected charges from electrodes prevent domain nucleation and growth), and microcrack formation (long-term cycling initiates microcracks from grain boundaries or domain walls). Most theories attribute degradation to excess charges from space or injection, accumulating at domain walls and hindering polarization. However, the precise link between charge accumulation and degradation has remained unclear. Direct, real-time observation of charge accumulation and domain structure evolution under cyclic electric fields is essential to elucidate this relationship. Advances in in-situ transmission electron microscopy (TEM), particularly using MEMS-based nano-chips, have made such nanoscale observations technically feasible.

The literature extensively discusses ferroelectric degradation, often linking it to domain wall pinning, where charge carriers or defect dipoles accumulate at domain walls, leading to domain locking. Some studies propose a competitive process between domain wall pinning and unpinning, with detrapping of charge carriers also occurring. Another proposed mechanism is domain nucleation suppression, where injected charges from the electrodes hinder the nucleation and growth of domains near the sample surface. The formation of microcracks initiated from grain boundaries or domain walls has also been suggested as a contributing factor. While charge accumulation at domain walls is frequently cited as a major contributor to ferroelectric degradation, the precise pathway connecting charge accumulation to the observed degradation has remained elusive. The lack of direct observational data at the nanoscale on charge accumulation and its real-time impact on domain structure evolution has hindered a complete understanding of this phenomenon.

This study employed in-situ biasing transmission electron microscopy (TEM) with MEMS-based nano-chips to investigate the domain switching behavior of Pb(Mg_1/3Nb_2/3)O₃-0.38PbTiO₃ (PMN-0.38PT) single crystals. A focused ion beam fabricated a PMN-0.38PT lamella, which was then positioned between two platinum electrodes with a 4 µm gap. A cyclic electric field was applied along the ±[100] crystallographic direction, and the resulting domain structure evolution was monitored using scanning transmission electron microscopy-differential phase contrast (STEM-DPC) imaging. This technique allowed for observation of the charge distribution at domain walls during cyclic electric loading. The bipolar triangle wave used for the electric biasing test had a peak voltage of ±4.7 V and a frequency of 1/30 Hz, resulting in a peak applied electric field of ±1.175 MV/m. Electron diffraction patterns were acquired to identify the polarization direction of domains, while STEM-DPC imaging was used to map the local electric field and monitor charge accumulation at domain walls. Electron Energy Loss Spectroscopy (EELS) was also used to measure plasma peak shifts at domain walls, providing quantitative information on charge density changes. The stability and reliability of STEM-DPC imaging were validated through supplementary data and analysis to account for potential artefacts from defocusing and domain switching during the experiment.

The in-situ TEM observations revealed the nucleation and growth of *c* domains (domains with out-of-plane polarization) from the original domain walls during cyclic electric loading. The area of the *c* domains increased exponentially with the number of cycles. STEM-DPC imaging showed increased local electric field amplitude at domain walls after cyclic loading, providing direct evidence of charge accumulation. EELS measurements confirmed this charge accumulation, quantifying the change in charge density at the domain walls. Notably, the *c* domains exhibited a 'frozen' state, remaining unresponsive to the applied electric field, even after the field was reversed. This lack of response is a key factor in the observed ferroelectric degradation. Analysis showed that the accumulated charges stabilize the *c* domains by acting as compensation charges, weakening the depolarization field and reducing the influence of the external electric field. The formation of the *c* domains was a consequence of cyclic electric loading, not a single event. Furthermore, ferroelectric hysteresis loops showed a decrease in the switchable area of domains with increasing cycles, while the growth of unswitchable domains increased – a clear indication of ferroelectric degradation. The 'frozen' state of *c* domains persisted for extended periods, remaining even 18 months after the experiment concluded. Stress-induced nanodomains were observed in some *a₂* domains, influencing their switching behavior, although the *c* domains in these regions remained immobile.

This study provides direct evidence linking charge accumulation at domain walls to the formation of frozen *c* domains, thereby explaining ferroelectric degradation. The findings bridge the gap in understanding how space charges or injected charges contribute to this degradation. While the nanoscale nature of the experiment might limit the direct applicability to bulk materials, the results are highly relevant to thin-film ferroelectrics and nanodevices where degradation remains a significant challenge. The observed initiation of frozen domains at domain walls suggests that other polarization transition zones, such as phase boundaries, interfaces, and surfaces, are also potential sites for frozen domain formation. This is supported by the observation of frozen domains near the sample edge, a region with a reconstructed surface and randomly distributed polarizations. Future studies could investigate the influence of different materials, electrode types, and applied electric field parameters on the degradation process.

This research demonstrated that ferroelectric degradation in PMN-0.38PT single crystals is driven by the formation of frozen *c* domains originating from domain walls due to charge accumulation. The accumulated charges stabilize these domains, rendering them unresponsive to the applied electric field. This provides crucial insights into the nanoscale dynamics of ferroelectric degradation in thin films and nanodevices, highlighting the importance of charge management strategies for improving device reliability. Future work could explore different ferroelectric materials and investigate the influence of various factors on frozen domain formation and growth.

The study focused on PMN-0.38PT single crystals with specific dimensions and experimental conditions. The generalizability of these findings to other ferroelectric materials and different device geometries requires further investigation. The use of in-situ TEM inherently limits the size and complexity of the samples that can be studied. The exact nature and origin of the accumulated charges warrant further exploration using advanced characterization techniques. While the frozen state of the *c* domains was observed to persist for an extended time, long-term stability studies would strengthen these observations.