Nonlinear and linear conductance modulation and synaptic plasticity in stable tin-zinc oxide based-memristor for neuro-inspired computing

• Published in: Materials science in semiconductor processing Vol. 186; p. 109111
• Main Authors: Rajwali Khan, Shahid Iqbal, Fazal Raziq, Maram, Pardha Saradhi, Chakrabortty, Sabyasachi, Sangaraju, Sambasivam
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.02.2025
• Subjects: Artificial synapses; Bipolar resistive switching; Data retention; Endurance; Memristors; Oxygen vacancies; Oxygen vacancies; Endurance; Bipolar resistive switching; Data retention; Artificial synapses; Memristors
• DOI: 10.1016/j.mssp.2024.109111
• ISSN: 1369-8001

Inducing post-transition metals in an oxide semiconductor system has a high potential for use in storage for neuromorphic computing. It is challenging to find a material that can be switched stably between multiple resistance states. This research explores the memristive properties of Sn (post-transition metal)-doped ZnO (SZO) thin films, emphasizing their application in memristor devices. The (magnetron sputtered) synthesized SZO thin films in the form of Ag/SZO/Au/Ti/SiO₂ device demonstrated a clear bipolar resistive switching (BRS) behavior with VSET and VRESET of 1.0 V and −0.75 V, respectively. The memristor could change between a high resistance state and a low resistance state with a high RON/OFF rate of 104, mimicking synaptic behaviors such as potentiation and depression. This switching is attributed to the formation and dissolution of Ag filaments within the SZO layer, influenced by the migration of Ag⁺ ions and the presence of oxygen vacancies. These vacancies facilitate the formation of conductive filaments under positive bias and their dissolution under negative bias. The endurance and retention tests showed stable switching characteristics, with the memristor maintaining distinct HRS and LRS over 100 cycles and retaining these states for over 5K seconds without significant degradation. Finally, the nonlinearity values for potentiation and depression were αp∼1.6 and αd ∼ -0.14, suggesting that the memristor may be more responsive to increasing synaptic weights in biological systems. The linearity response at a very small pulse width showed the device is more applicable for neuromorphic applications. The observed memristor combined with stable endurance and retention performance, suggests that this memristor structure could play a crucial role in the development of artificial synapses and memory technologies.