Memristive domain walls in ferroelectric BiFeO3 thin films for brain-inspired electronics
Today’s technology demands ever-increasing computational power for applications such as the Internet of Things and self-driving cars. Tasks such as image recognition consume large amounts of energy on current computers, while further transistor scaling is approaching physical limits. Inspired by the brain’s exceptional efficiency in processing information, neuromorphic computing seeks to develop powerful yet energy-efficient hardware. Advanced materials may play a key role in solving these problems.
One promising class of materials concerns multiferroic oxides. When grown as a thin film, they self-assemble into nano-scale domains that differ in the orientation of a specific order parameter, such as the electric polarization in the case of ferroelectrics. These regions are separated by domain walls, forming complex and interconnected networks. Interestingly, domain walls display higher conductivity than the host material and can be tuned to some extent by applying electrical signals, similar to a resistor with memory (a so-called memristor). As a result, the domain walls form conductive pathways within the network - resembling the connections between neurons in the brain.
In his thesis, Jan Rieck examines domain walls in the prominent multiferroic material bismuth ferrite (BiFeO3) and their potential for brain-inspired electronics. To this end, he characterizes in detail the lateral electrical transport along domain walls through the network, using both established advanced measurement techniques and novel nanotechnology methods applied to this material for the first time. The findings highlight the fascinating transport properties of self-assembled domain wall networks and address both the possibilities and challenges of realizing neuromorphic hardware based on these systems.