Solution-Processed Zinc Oxide Nanoparticles for Thin Film Optoelectronic Neuromorphic Devices

Abstract

Listen

Translate article

Metal oxide nanostructures are an important class of thin film electronic materials well-suited for applications, which can exhibit large changes in their electrical and optical properties upon exposure to different external stimuli, gas and/or chemical environments at their surfaces [1, 2]. Zinc oxide (ZnO) in particular, has been garnering increased interest due to its unique combination of properties that are advantageous for electronics and optoelectronics such as wide band gap, large exciton binding energy, low-cost, ease of processing and availability [3]. This has led to the development of ZnO nanoparticles and thin films wherein the large surface areas and tunable properties of ZnO nanostructures make them particularly suitable for applications requiring enhanced performance with low-power and decreased cost [4].ZnO nanostructures have been used to demonstrate light-emitting diodes, photodetectors, gas sensors, transparent conductors, and memristors, among other applications [5]. Many of the desirable properties of ZnO arise because of surface states that can be tuned by various fabrication and processing steps and are further enhanced at nanoscale dimensions [4]. Recently, these attributes have shown that ZnO nanomaterials are promising for applications in optoelectronic neuromorphic computing [6]: ZnO nanostructured thin films have shown desirable properties analogous to long-term/short-term memory and synaptic plasticity in biological systems [7], which could allow for efficient analog computing devices that emulate brain functions and overcome performance limitations of conventional computing based on the von Neumann architecture [8]. Such neuromorphic materials have great potential for artificial neural networks, ultra-low power biologically inspired computing and vision [6].In this work, we present results on solution-processed ZnO nanoparticle thin films that display optoelectronic neuromorphic or synaptic behavior based on persistent photoconductivity [9]. The thin films were fabricated using nanoparticle inks (nanoinks) obtained via planetary ball milling (PBM) of bulk ZnO powders [10]. PBM is a solution-based nanofabrication approach that can quickly produce nanoink suspensions via colloidal grinding in a given solvent for low-cost thin film coatings without requiring complex processing.The thin film optoelectronic neuromorphic devices were fabricated by depositing ZnO nanoink onto flat insulating glass substrates followed by contact formation (see Fig. 1a for overall device structure). PBM was performed using ZnO powder in ethylene glycol or deionized water solvent with zirconia grinding beads. The grinding speed was varied between 200 and 1000 rpm. Analysis of the ZnO films after deposition showed they consist of nanostructured particles with sizes reaching below 100 nm, as displayed in the atomic force microscopy (AFM) images in Fig. 1b, depending on grinding conditions.The optoelectronic properties of the ZnO thin films were evaluated via two-terminal photoconductance measurements using a probe station connected to a precision source-measure unit, under ambient atmosphere and temperature. In order to determine long-term memory (LTM) and short-term memory (STM) properties for artificial optoelectronic synapses, the ZnO films were exposed to different width broadband light pulses: Fig. 1c shows photocurrent vs. time plots for a typical ZnO thin film device subjected to long pulses with large intervals, which exhibits the rise and decay of photoconductance associated desorption/adsorption of surface oxygen species in ZnO (i.e., oxygen surface vacancies that act as donors) [9]. On the other hand, Fig. 1d shows the photoconductive behavior for short pulses separated by short intervals – here the current continues to increase between pulses, which is indicative of an optoelectronic neuromorphic memory effect, or LTM. In other words, the system “remembers” the previous pulse state, in contrast to losing that information in the case of complete photocurrent decay between pulses (Fig. 1c), or STM. Thus, the ZnO films display optoelectronic memory properties that mimic brain functions/learning in order to recognize different optical signals, promising for synaptic devices and systems in neuromorphic memory and computing.[1] N. Nasiri, D. Jin, A. Tricoli, Adv. Opt. Mat., 7, 1800580 (2019).[2] C. Wang et al., Sensors, 10, 2088 (2010).[3] C. Klingshirn, Phys. Status Solidi B, 244, 3027 (2007).[4] R. Khokhra, B. Bharti, H.-N. Lee, R. Kumar, Sci. Rep., 7, 15032 (2017).[5] M. Laurenti, S. Porro, C. F. Pirri, C. Ricciardi, and A. Chiolerio, Crit. Rev. Solid State Mater. Sci., 42, 153 (2017).[6] Y. Wang et al., Adv. Intell. Syst., 3, 2000099 (2021).[7] R. D. Chandra, K. G. Gopchandran, ACS Appl. Electron. Mater., 3, 3846 (2021).[8] M. A. Zidan, J. P. Strachan, W. D. Lu, Nat. Electron., 1, 22 (2018).[9] S.-L. Gao et al., ACS Appl. Electron. Mater., 6, 1542 (2024).[10] R. Sapkota, P. Duan, T. Kumar, A. Venkataraman, C. Papadopoulos, Appl. Sci., 11, 9676 (2021). Figure 1