LAUSR

Rising demand for artificial intelligence (AI), especially generative AI, is increasingly limited by the so-called von-Neumann bottleneck. Emerging nonvolatile memory devices, e.g., ferroelectric devices based on hafnium oxide, have been suggested to overcome this bottleneck by enabling near- and in-memory computing through synaptic bit cells. However, these technologies still face challenges in the area of reliability and circuit architecture. In this article, we review recent advances in ferroelectric hafnium oxide-based synapses. In addition, we present challenges that need to be overcome before their industrial application and outline future directions for improving these devices further.