Associate Professor Sayani Majumdar has published an invited perspective article in APL Machine Learning, as part of the journal’s Special Collection Neuromorphic Technologies for Novel Hardware AI. The article explores the potential, challenges, and development path of ferroelectric tunnel junction (FTJ) technology, particularly for applications in non-volatile memory and neuromorphic computing.
“I tried to use my long experience in this field to map the progress of the technology, starting from oxide perovskite ferroelectric devices to CMOS back-end compatible ferroelectric tunnel junctions and circuits,” says Majumdar.
FTJs offer key advantages such as scalability, high on/off switching ratio, fast read/write operations, low leakage, and high endurance. The article highlights how specific device architectures can be tailored for different neuromorphic applications—such as volatile or non-volatile memory or programmable synaptic behavior—by engineering the depolarizing field. These innovations could enable highly integrated neuron-synapse circuits with significantly reduced complexity.
The publication also addresses challenges such as the relatively low on-current in CMOS-compatible FTJs and reviews promising architectural solutions for large-scale integration.
Read the full article here: https://pubs.aip.org/aip/aml/article/3/2/020902/3344306
The work was supported by the Research Council of Finland (Academy of Finland), Business Finland, and the Chips Joint Undertaking through the ARCTIC-KDT project.