Modern hardware for artificial vision can benefit from multi-terminal synaptic transistors (MTSTs). MTST’s advantages include greater synaptic weight controllability, nonvolatile synaptic function, as well as a variety of controllable parameters. Recent developments in wide-bandgap metal oxide semiconductors (MOSs)-based synaptic transistors have shown promising prospects to accomplish ultraviolet (UV) artificial vision.
Image Credit: asharkyu/Shutterstock.com
Ongoing scientific investigations seek to mimic the five senses humans use to comprehend their surroundings. Among them, the majority of the information that the brain processes is acquired through the visual sense, demonstrating the vast significance of bionic vision in the creation of artificial systems.
Artificial Vision
The Automated Imaging Association (AIA) defines artificial vision as any industrial or non-industrial application in which a combination of hardware and software directs the operation of equipment to carry out tasks based on the acquisition and processing of images.
The human brain constantly absorbs information from its surroundings through sight and is especially sensitive to the apparent size of objects to sudden changes in light. Artificial vision is the field of research that enables computers to make decisions as quickly as the human brain by processing visual information.
Although the differences between artificial and computer vision are minor, they should not be mistaken. The goal of computer vision is to comprehend digital images after they have been edited and analyzed. Artificial vision, on the other hand, just needs to examine the visual elements that cause a certain action to be taken. For instance, an autonomous vehicle's artificial vision is aimed at spotting obstructions in order to prevent collisions.
UV vision
While artificial vision based on IR and the visible part of the electromagnetic spectrum is widely studied, ultraviolet (UV) vision is critical for many biological species to navigate their environment and avoid predators. UV vision can also be crucial for artificial systems used in astronomical surveys and the discovery of terrestrial planets.
Wide-Bandgap Metal Oxide Semiconductors
Suitable wide-band-gap metal oxide semiconductors (MOSs) offer numerous benefits for synaptic transistors driven by UV lasers. For example, superior optoelectronic characteristics, air stability, process compatibility, and cost-effectiveness. One-dimensional MOSs also have the advantage of having a high surface-to-volume ratio. The places of contact between neurons known as synapses are where information is transferred from one neuron to the next.
New advancements in bio-inspired synaptic transistors are based on ZnSnO nanowires using a low-cost electrospinning technique. These transistors can be successfully tuned by a 375 nm UV laser to produce air-stable synaptic characteristics. UV light with different pulse counts and light intensities is used in order to get the necessary electrical signals, such as paired-pulse facilitation (PPF), excitatory post-synaptic current (EPSC), and learning-forgetting process based on relative long-term potentiation (LTP). In ZnSnO nanowires, the production of free charge carriers is controlled by the adsorption and desorption of oxygen molecules.
The SnO2/ZnO hetero-structures that were discovered through the use of X-ray diffraction (XRD) and transmission electron microscopy (TEM) measurements can successfully reduce the recombination of free carriers in the ZnSnO nanowires induced by the UV radiation. Additionally, the high binding energies of the Sn-O and Zn-O bonds play a significant role in the stability of the device.
UV artificial vision based on the synaptic properties acquired from ZnSnO nanowires has been demonstrated in various applications. For example:
- Programmable optoelectronic logic gates - Combining an optical input of UV light with an electrical input of gate voltage can be used to alter both "AND" and "OR" gates. "AND-OR" cycles have been repeated multiple times in order to confirm the stability of the programmable "AND" and "OR" logic gates, and a stable transition between these two logic gates has been demonstrated.
- Handwritten recognition - The accuracy of pattern recognition is based on the linearity and symmetry of the weight update. The Modified National Institute of Standards and Technology (MNIST) database of handwritten digits is utilized to execute supervised learning on the ZnSnO synaptic transistor for neuromorphic computing. Simulations have shown how optoelectronic synaptic devices have the potential to be used in neural network computing in the future.
- Sensing-memory-processing system - Traditional CMOS often consists of separate detection, memory, and processing units, which results in significant energy loss and compute lag. Optoelectronic artificial synapses, in contrast, can combine memory and information processing with the ability to sense light. The information sensing/memory-processing system based on a ZnSnO optical synaptic array can perform real-time picture identification, in situ memorizing, and distinguishing input data.
Outlook
New UV-driven synaptic transistors based on ZnSnO nanowires have been developed using low-cost electrospinning coupled with a nanowire transfer approach. To obtain the associated synaptic signals, such as EPSCs, PPFs, and LTPs, UV light was administered with a range of pulse characteristics and light intensities. EPSCs produced from ZnSnO nanowire transistors show greater device environmental stability. Additionally, the SnO2/ZnO heterostructures detected by TEM and XRD demonstrate a reduction of the recombination of UV-induced free carriers.
More from AZoOptics: Recent Research on Quantum Cascade Lasers
References and Further Reading
Ruifu Zhou, Wenxin Zhang, Haofei Cong, Yu Chang, Fengyun Wang, Xuhai Liu, Metal oxide semiconductor nanowires enabled air-stable ultraviolet-driven synaptic transistors for artificial vision, Materials Science in Semiconductor Processing, Volume 158, 2023, 107344,ISSN 1369-8001, https://doi.org/10.1016/j.mssp.2023.107344
X.H. Liu, F.Y. Wang, J. Su, et al., Bio-inspired 3D artificial neuromorphic circuits, Adv. Funct. Mater. 32 (2022), 2113050. https://onlinelibrary.wiley.com/doi/abs/10.1002/adfm.202113050
S. Zhang, K.X. Guo, L. Sun, et al., Selective release of different neurotransmitters emulated by a p-i-n junction synaptic transistor for environment-responsive action control, Adv. Mater. 33 (2021), 2007350. https://doi.org/10.1002/adma.202007350
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.