From Nanoionic Processes to Neuromorphic Sensing

Memristive devices based on the valence change mechanism are highly interesting candidates for non-volatile data storage and for hardware representations of synapses and neurons in neuromorphic circuits. Although it has been early proposed that the switching mechanism is based on the field-driven movement of oxygen vacancies within nanosized filaments [1], details of the underlying physical-chemical processes are difficult to access experimentally. However, these details are highly relevant for the kinetics of the devices and the time stability of the resistances states. We employed photoemission electron spectoscopy (PEEM) to study the movement of oxygen vacancies during device operation and after resistance relaxation [2, 3]. By combining PEEM data with simulations of the ionic and electronic transport, we gained a deep understanding of the interplay between oxygen vacancy movement, the modulation of space charge zones and the switching performance.<br/>We use the knowledge to design VCM devices with tailored resistance relaxation [4] to gain spatio-temporal information in a neuromorphic sensing circuit, the time-difference encoder, developed at the University of Groningen [5].<br/>[1] Waser et al., Adv. Mater. 2009<br/>[2] Bäumer et al., Nature Commun. 2015<br/>[3] Bäumer et al., Nature Commun. 2016<br/>[4] J. Hellwig et al., Adv. Funct. Mater. 2024<br/>[5] Schoepe et al, Nature Commun. 2024