Memristors Mimic Brain Synapses, Enabling Efficient Neuromorphic Computing.

The pursuit of energy-efficient computing increasingly focuses on neuromorphic systems, hardware designed to emulate the human brain’s remarkable ability to process information. A critical component of this endeavour lies in developing artificial synapses, devices capable of mimicking the dynamic adjustment of connections between neurons. Researchers from the Instituto de Nanociencia y Nanotecnología (INN) in Argentina and the Zernike Institute for Advanced Materials at the University of Groningen, alongside colleagues at Centro Atómico Bariloche and Instituto Balseiro, investigate the mechanisms underpinning synaptic behaviour in memristive devices. Their work, detailed in the article ‘Synaptic plasticity in Co/Nb:STO memristive devices: The role of oxygen vacancies’, demonstrates how the movement of oxygen vacancies within a cobalt/niobium oxide structure on a strontium titanate substrate modulates the device’s electrical resistance, enabling both short-term and long-term plasticity reminiscent of biological learning. The team, comprising Walter Quiñonez, Anouk Goossens, Diego Rubi, Tamalika Banerjee, and María José Sánchez, present experimental evidence and a computational model correlating oxygen vacancy distribution with changes in the Schottky barrier, a crucial element in controlling the device’s behaviour.

Long-term potentiation and depression, the cellular mechanisms underpinning learning and memory, inspire the development of artificial synapses, with memristors emerging as promising emulators of biological synaptic behaviour. Memristors, or memory resistors, exhibit a unique ability to change their electrical resistance based on the history of applied voltage or current, a property crucial for mimicking synaptic plasticity. Strontium titanate (SrTiO₃) currently receives significant attention as a key material in investigations into memristive devices intended for neuromorphic computing, a field aiming to build computer systems inspired by the structure and function of the brain.

Researchers actively manipulate oxygen vacancies within the SrTiO₃ crystal lattice to influence its electrical properties and facilitate resistive switching, the core principle behind memristive functionality. Oxygen vacancies are defects where an oxygen atom is missing from its regular position, creating a change in the material’s conductivity. The behaviour of interfaces between SrTiO₃ and other materials, such as platinum or hafnium oxide, proves critical in determining device performance. These interfaces form Schottky barriers, regions where differing electronic properties create a potential barrier for current flow, and their characteristics directly influence the memristor’s operation. Studies demonstrate how these vacancies modulate conductivity, creating pathways for charge transport and altering the Schottky barrier height and width.

Experimental findings reveal that resistance alterations follow a power-law during reading operations, indicative of short-term plasticity, mirroring the transient changes observed in biological synapses. Successive electrical pulses, however, induce stepwise resistance increases, representing long-term memory retention, analogous to the strengthening of synaptic connections through repeated stimulation. These observed behaviours align with computational models correlating oxygen vacancy distribution with Schottky barrier modulation, providing a comprehensive understanding of the device’s functionality.

Recent research demonstrates that cobalt/niobium-doped strontium titanate (Co/Nb:SrTiO₃) Schottky memristors exhibit plasticity behaviours directly linked to oxygen vacancy electromigration, the movement of oxygen vacancies under an electric field. This establishes a clear physical mechanism underpinning both short-term and long-term synaptic behaviours, confirming that the movement of these vacancies alters the Schottky barrier characteristics.

The findings establish a direct link between material properties and device behaviour, confirming that the modulation of the Schottky barrier, driven by oxygen vacancy movement, is central to the observed plasticity. This provides a predictive framework for device optimisation, allowing researchers to tailor memristor properties by controlling the distribution and movement of oxygen vacancies. Specifically, the research elucidates how oxygen vacancy electromigration alters the barrier height and width, dynamically adjusting the device’s resistance and enabling the emulation of synaptic weight changes, the strength of connections between artificial neurons.

The successful reproduction of experimental results by the developed model strengthens the validity of the proposed mechanism and provides a platform for further investigation. The model accurately captures the correlation between oxygen vacancy distribution and Schottky barrier modulation, allowing for the prediction of device behaviour under varying conditions, and enabling the design of memristors with tailored properties.

Consequently, strontium titanate-based memristors demonstrate considerable promise for applications in neuromorphic computing, offering a pathway to create more energy-efficient and powerful devices capable of mimicking the brain’s parallel processing capabilities.👉 More information