講演情報
[8p-N304-9]Towards Hybrid Memory - Computing Architectures: Resistive Switching and Synaptic Plasticity in Magnetic Barium Hexaferrite Thin Films
〇Abhishek Atulbhai Gor1, Aziz Lokhandwala1, Ankur Solanki1, Chetna C. Chauhan2, N. M. Devashrayee2 (1.PDEU, 2.NIrma Uni.)
キーワード：
Computation-In-Memory(CIM)、Ferrite、Magnetic Materials
The von Neumann bottleneck necessitates novel in-memory computing architectures. While memristive devices emulate biological synapses, their coexistence with established magnetic storage materials remains largely unexplored. Here, we demonstrate unipolar and bipolar resistive switching, highly tunable via compliance current, alongside neuromorphic functionalities in barium hexaferrite (BaFe12O19, BaM) thin films.
In this work we successfully emulated key synaptic functions, including paired-pulse facilitation (PPF), potentiation–depression (PD), spike-timing-dependent plasticity (STDP), spike-number-dependent plasticity (SNDP), and spike-rate-dependent plasticity (SRDP).
We systematically examined the influence of varying magnetic fields on these IV/synaptic responses, revealing a strong interplay between magnetic and memristive properties. This dual functionality highlights the potential of BaM thin films as hybrid magnetic-neuromorphic platforms capable of simultaneously supporting data storage and in-memory computation.
Device stability and phase purity were confirmed under ambient conditions. With an energy consumption of ~30 nJ per 15 ms spike, these findings highlight BaM thin films as compelling material platform for energy-efficient crossbar architectures and provide a pathway toward integrating magnetic functionality with neuromorphic computing in next-generation hybrid in-memory computing systems.
In this work we successfully emulated key synaptic functions, including paired-pulse facilitation (PPF), potentiation–depression (PD), spike-timing-dependent plasticity (STDP), spike-number-dependent plasticity (SNDP), and spike-rate-dependent plasticity (SRDP).
We systematically examined the influence of varying magnetic fields on these IV/synaptic responses, revealing a strong interplay between magnetic and memristive properties. This dual functionality highlights the potential of BaM thin films as hybrid magnetic-neuromorphic platforms capable of simultaneously supporting data storage and in-memory computation.
Device stability and phase purity were confirmed under ambient conditions. With an energy consumption of ~30 nJ per 15 ms spike, these findings highlight BaM thin films as compelling material platform for energy-efficient crossbar architectures and provide a pathway toward integrating magnetic functionality with neuromorphic computing in next-generation hybrid in-memory computing systems.