# A Lithium Fluoride Interfacial Layer for Low-Voltage and Reliable Perovskite Memristors

**Authors:** Naresh Kumar Pendyala, Ignacio Sanjuán, Qun-Gao Chen, Wen-Ya Lee, Chu-Chen Chueh, Antonio Guerrero

PMC · DOI: 10.1021/acsaelm.5c02347 · ACS Applied Electronic Materials · 2025-12-19

## TL;DR

This paper introduces a lithium fluoride layer that improves the performance of perovskite memristors for low-voltage and reliable neuromorphic computing.

## Contribution

The use of a LiF interfacial layer is proposed to enhance memristor performance through improved ionic conductivity and low-voltage operation.

## Key findings

- LiF interlayer enables low-voltage operation (∼70–150 mV) with stable and reproducible performance.
- Devices show high cycling stability over >104 cycles with minimal variability.
- Memristors functioned as synaptic weights in a DNN for handwritten digit recognition.

## Abstract

Halide-perovskite materials have emerged as promising
candidates
for constructing reliable memristors, a key element for advancing
neuromorphic computing systems. While several perovskite formulations
have been tested, the nature of the external interfaces has not been
exploited to its full potential. In this study, LiF is employed as
an interfacial layer between a bromide-perovskite and the top contact.
The interlayer acts as a source of Li+ ions that facilitate
the formation of conducting filaments, combining the high ionic conductivity
of a halide perovskite and the small size of the Li+ ion.
The incorporation of a LiF layer significantly enhances device performance
at low operation voltages (∼70–150 mV) with a gradual
increase in conductance, rendering the devices suitable for analog
computation. Overall, devices yield stable and highly reproducible
results with high sensitivity to the external voltage. Notably, these
devices demonstrate high cycling stability during >104 cycles
with small variability in writing–erasing measurements. These
findings underline the potential of LiF-enhanced memristors for reliable
and energy-efficient neuromorphic computing applications. As a proof
of concept, these low-voltage memristors successfully functioned as
synaptic weights in an emulated deep neural network (DNN) for handwritten
digit recognition. Importantly, the use of LiF as an interlayer should
be universally valid for other families of materials used in memristor
applications.

## Linked entities

- **Chemicals:** LiF (PubChem CID 224478), Li+ (PubChem CID 28486)

## Full-text entities

- **Chemicals:** bromide (MESH:D001965), Halide (-), Li+ (MESH:D008094), LiF (MESH:C027651), Perovskite (MESH:C059910)

## Full text

_Full body text omitted from this summary view._ Fetch the complete paper as Markdown: https://tomesphere.com/paper/PMC12805799/full.md

## Figures

4 figures with captions in the complete paper: https://tomesphere.com/paper/PMC12805799/full.md

## References

22 references — full list in the complete paper: https://tomesphere.com/paper/PMC12805799/full.md

---
Source: https://tomesphere.com/paper/PMC12805799