Exploring Complementary Approaches in Electron Microscopy for Neuromorphic Devices
Neuromorphic hardware, inspired by the architecture and functionality of the human brain, aims to emulate cognitive processes efficiently and accurately. Electron microscopy (EM) has emerged as a powerful tool for the characterization of neuromorphic devices, providing insights into the structural, morphological and chemical properties critical for understanding device performance and behavior. In first place, recent advancements and challenges in characterization by EM are critically discussed, highlighting the importance of interdisciplinary collaboration between materials scientists and electrical engineers. Although EM offers one of the highest lateral resolutions available, the standard techniques reach their limitations in neuromorphic devices, where the processes involved may occur spatially and temporally localized. In this thesis, distinct complementary EM approaches have been used to address this issue and to access higher level information to correlate structural and spectroscopic features with electrical performance. One approach is to combine different perspectives (e.g., plan-view and cross-section) that provide qualitatively distinct information. In a second approach, complementary spectroscopic methods were combined for a neuromorphic device made of several very thin oxide layers (HfO2, Al2O3 and NbOx) to build a purely electronic switching model based on charging and discharging of trap states in HfO2.
A third approach is the combination of the advanced in situ SEM methods active voltage contrast and resistive contrast imaging. These techniques provide the possibility to correlate the critical spiking dynamics of a self-organized nanoparticle network at the percolation threshold with the evolving current path. These methods revealed a changing potential distribution during spiking and a changing position of nano-gaps in the current path after spiking.
Preview
Rights
Use and reproduction:
No license. The provisions of the German Copyright Act (UrhG) apply.
Please note that individual components of the publication may be subject to other licensing or copyright conditions.