High-stability, low-power and broadband-spectral-response flexible carbon nanotube photoelectronic synaptic transistors modified with D-π-A-π-D photosensitive molecules

The development of neuromorphic optoelectronic devices with broadband response, low power consumption and high stability has emerged as a prominent research focus due to their potential application in artificial intelligence. Hence, we report a flexible carbon nanotube-based photoelectronic synaptic transistor array modified by a D-π-A-π-D structured photosensitive molecule, 4,8-diyldi-5,2-thiophenediylbis[4-(N,N-bis(4-octyloxyphenyl)amino)phenyl]benzo[1,2-c:4,5-c']bis([1,2,5]thiadiazole) (PM329), in the device channels. The incorporation of PM329 into single-walled carbon nanotube (SWCNT) channels enables broadband photoresponse ranging from 365 nm to 1050 nm, alongside outstanding device performance, including high stability (operational after two months of air storage), resistance to water and oxygen interference, long retention times (78.50 % of the current retained after 10000 s during slow relaxation) and ultralow power consumption (as low as 25 aJ per event). These properties are attributed to the precise energy band alignment between PM329 and SWCNTs, facilitating efficient charge transport and enhanced photoresponse. Furthermore, the neuromorphic devices exhibit critical synaptic behaviors, including excitatory postsynaptic currents (EPSC), short-term potentiation (STP), long-term potentiation (LTP) and paired-pulse facilitation (PPF). Key functionalities such as image perception, associative learning and long-term memory were successfully demonstrated under pulsed light stimulation. Notably, when integrated into a machine vision system, the transistor arrays enable automobile rear-end collision detection and image perception capabilities. This work highlights a robust and scalable strategy for developing carbon-based optoelectronic synaptic devices with high stability, energy efficiency and practical applicability, offering significant potential for artificial intelligence and neuromorphic computing.