Discover the SciOpen Platform and Achieve Your Research Goals with Ease.
Search articles, authors, keywords, DOI, etc.
The growing demands of artificial intelligence hardware have exposed a fundamental latency bottleneck in conventional floating-gate memory arrays. Although two-dimensional (2D) floating-gate memories have achievd nanosecond-level operations, their large-scale integration remains hindered by the scalability issues of the commonly used hBN dielectric layer. Here, we demonstrate scalable ultrafast non-volatile floating-gate memory arrays with hybrid Y2O3/HfO2 dielectrics as the tunneling layer, achieving ~ 60 ns ultrafast programming/erasing speeds, high extinction ratios up to ~ 109, long-term data retention approaching 105 s, and robust cycling endurance exceeding 1800 cycles with multi-bit storage capabilities. A 5 × 5 memory array shows outstanding uniformity in memory window sizes, on/off ratios, and programmed/erased threshold voltages. Furthermore, the floating-gate memory devices also demonstrate efficient optical erasure with 1,441 A/W responsivity and 9 × 1016 Jones detectivity. The promising uniformity of these memory arrays highlights the potential for larger-scale integration and its scalability for high-density storage and photodetection applications.

This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0, https://creativecommons.org/licenses/by/4.0/).
Comments on this article