Citation: Chen, Q.; Han, T.; Zeng, J.;
He, Z.; Liu, Y.; Sun, J.; Tang, M.;
Zhang, Z.; Gao, P.; Liu, G.
Perovskite-Based Memristor with
50-Fold Switchable Photosensitivity
for In-Sensor Computing Neural
Network. Nanomaterials 2022, 12,
2217. https://doi.org/10.3390/
nano12132217
Academic Editors: Ki-Hyun Kim and
Deepak Kukkar
Received: 8 May 2022
Accepted: 27 June 2022
Published: 28 June 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
Perovskite-Based Memristor with 50-Fold Switchable
Photosensitivity for In-Sensor Computing Neural Network
Qilai Chen
1
, Tingting Han
2,3
, Jianmin Zeng
2
, Zhilong He
2
, Yulin Liu
4
, Jinglin Sun
2
, Minghua Tang
4
,
Zhang Zhang
3
, Pingqi Gao
1
and Gang Liu
2,
*
1
School of Materials, Sun Yat-Sen University, Guangzhou 510275, China; chenqlai@mail.sysu.edu.cn (Q.C.);
gaopq3@mail.sysu.edu.cn (P.G.)
2
Department of Micro and Nano Electronics, School of Electronic Information and Electrical Engineering,
Shanghai Jiao Tong University, Shanghai 200240, China; hantingt1997@163.com (T.H.);
jamy3531@sjtu.edu.cn (J.Z.); zhilong_he@sjtu.edu.cn (Z.H.); ldsunjinglin@163.com (J.S.)
3
School of Microelectronics, Hefei University of Technology, Hefei 230601, China; zhangzhang@hfut.edu.cn
4
School of Materials Science and Engineering, Xiangtan University, Xiangtan 411105, China;
ylliuxtu@163.com (Y.L.); tangminghua@xtu.edu.cn (M.T.)
* Correspondence: gang.liu@sjtu.edu.cn
Abstract:
In-sensor computing can simultaneously output image information and recognition results
through in-situ visual signal processing, which can greatly improve the efficiency of machine vision.
However, in-sensor computing is challenging due to the requirement to controllably adjust the
sensor’s photosensitivity. Herein, it is demonstrated a ternary cationic halide Cs
0.05
FA
0.81
MA
0.14
Pb(I
0.85
Br
0.15
)
3
(CsFAMA) perovskite, whose External quantum efficiency (EQE) value is above 80%
in the entire visible region (400–750 nm), and peak responsibility value at 750 nm reaches 0.45 A/W.
In addition, the device can achieve a 50-fold enhancement of the photoresponsibility under the same
illumination by adjusting the internal ion migration and readout voltage. A proof-of-concept visually
enhanced neural network system is demonstrated through the switchable photosensitivity of the
perovskite sensor array, which can simultaneously optimize imaging and recognition results and
improve object recognition accuracy by 17% in low-light environments.
Keywords:
memristor; perovskite; machine vision; photosensitivity; neural network; in-sensor computing
1. Introduction
As an essential branch of artificial intelligence (AI) technology, intelligent machine
vision has been extensively applied in scientific, industrial, and consumer markets and has
produced giant economic efficiency [
1
–
3
]. Traditional von Neumann architecture computers’
processing and memory blocks are physically separated [
4
,
5
]. When processing high-
throughput computing tasks, separate memory and computing modules will frequently
access and read data. Due to the different operating frequencies of the computing modules
and memory modules, this process reduces the speed of processing tasks and increases
energy consumption [
6
,
7
]. These drawbacks are particularly prominent in dealing with
machine vision tasks because the conversion of image or video signals is accompanied by a
large amount of data transmission and calculation [8].
Bioinspired neuromorphic in-memory or in-sensor digital/analog computing engi-
neering, with high energy efficiency and low energy consumption, endow them as a
potential candidate computing solution [
9
]. Modern image sensors on the basis of solid-
state semiconductor technology could dependably capture the optical information thanks
to the predefined photoresponsivity and internal chemical profile of silica-based materi-
als [
10
,
11
]. Unfortunately, the fixed photoresponsivity makes it impossible to implement
in-situ digital/analog computations in sensors. Therefore, traditional complementary
metal-oxide-semiconductor (CMOS) image sensors require complex peripheral circuits
Nanomaterials 2022, 12, 2217. https://doi.org/10.3390/nano12132217 https://www.mdpi.com/journal/nanomaterials
