Fabrication Of Carbon Nanotube Thin Film Transistors And Their Applications In Artificial Synapses And Low-power CMOS Circuits

Semiconducting single-walled carbon nanotubes（sc-SWCNTs）are considered as one of the excellent semiconductor materials which have the potential to bring revolutionary development in the post-Moore era.As an important electronic component,semiconducting single-walled carbon nanotube thin film transistors（sc-SWCNTs TFTs）have the advantages on excellent electrical performance,good reliability,high stability,as well as good compatibility with flexible electronics,silicon-based electronics and complementary metal oxide semiconductor（CMOS）technology.Moreover,these three-terminal sc-SWCNT devices can easily implement parallel control of the signal processing/transmission.Therefore,sc-SWCNT TFTs have played a pivotal role in the fields of biochemical sensors,integrated circuits,flexible printed electronics,and wearable electronics,and have great potential for building high-efficiency low-power artificial intelligence chips and neuromorphic sensory systems based on Non-Von Neumann architecture by the hardware method to breakthrough the existing dilemma due to the disaggregation of compute and storages.The visual system is indispensable for human interaction with the outside world,and its basic functional unit is biological synapse which can further transmit and process action potential information.Therefore,it is necessary to develop Non-Von Neumann type artificial vision systems to simulate biological visual perception.The constructions of artificial synaptic devices which can perceive,process,and memorize optical information,as well as CMOS processing circuits which can convert constant detection signals to pulsed electrical signals,are key steps in achieving the above goals.Although SWCNT TFT has the opportunity to play a significant role in artificial vision applications,it still faces some challenges inevitably.Due to the insensitivity of sc-SWCNT for illumination,the exploration is still underway on how to achieve high-performance multimodal SWCNT photoelectric artificial synapses compatible with silicon-based electronics,elucidate the mechanism of the synergistic effect,flexible SWCNT photosynaptic transistors with better biocompatibility.In addition,how to obtain enhancement-mode and low-power SWCNT CMOS devices and circuits compatible with carbon-based neuromorphic devices has become the hotspot and difficulty in this field to carbon-based electronics.In regard to these research difficulties,we developed SWCNT photoelectric synaptic transistors compatible with silicon and flexible substrates,explored the coupling effect of photoelectric signals,integrated more advanced synaptic functions in one device,clarified the working mechanism of SWCNT photoelectric artificial synapse and constructed the SWCNT CMOS inverter with ultra-low power consumption in this thesis.The main research contents and results are summarized as follows:（1）Optically and electrically modulated printed SWCNT synaptic transistors with simple structure were fabricated by using lightly-doped silicon with photogating effect as the single bottom gate electrode.SWCNT TFTs based on p-type and n-type lightly doped silicon showed the positive and negative photoresponse under the laser illumination,respectively.Drain currents of the synaptic devices increase gradually under the continuous light/electric stimulation so that several important synaptic behaviors（low-pass filtering and nonvolatile memory performance）were simulated.The mechanism of photoelectric synaptic plasticity has been proved,which mainly includes gate voltage changes of lightly-doped silicon under illumination,and the trapping state of the charge carriers existing at the surface/interfaces of/between the sc-SWCNTs and the high dielectric constant（k）oxide insulators.Finally,the SWCNT photoelectric synaptic transistor successfully imitated several high neural activities such as excitatory,inhibitory and logic operation by coupling the photoelectric stimulations.（2）Photosensitive materials were introduced to improve the photoelectric responses and implement multimodal input/output of Si-based SWCNT synaptic transistors.The deposition of CsBi₃I₁₀ on the top of SWCNT synaptic transistors based on lightly-doped Si substrates by spin coating can increase the photoelectric response of the original device by dozens of times,which is due to the excellent photosensitivity of CsBi₃I₁₀ and the effective dissociation of photogenerated carriers at the interface between SWCNT and CsBi₃I₁₀.Multimodal optoelectronic artificial synapse was implemented by using synergistic training of photoelectric signals and a series of complex neural activities was emulated incuding NOT-OR（NOR）logic,reproducible flash memory operated by light programming and electric erasing,and Classical conditioning of Pavlov’s dog.A single-layer perceptron based artificial neural network using long-term potentiation（LTP）/long-term depression（LTD）model characteristics was simulated to be trained for the recognition of MNIST handwritten digits with the maximum accuracy of about 85.46%.Besides,the SWCNT photosynaptic transistor based on heavily-doped silicon bottom gate was constructed by using Pd source/drain electrodes and rhodamine 6G（Rh6G）photosensitive layer,and optimizing the preparation process of active layer.The device can simulate nonvolatile memory and flash memory,and realize the better weight update（LTP/LTD）under optical and/or electrical stimulation.（3）Flexible ionic crosslinked poly（4-vinylphenol）（ILs-c-PVP）based SWCNT electric-double-layer（EDL）transistors were fabricated by fully-solution-process on polyimide（PI）substrates to imitate biological synaptic function and improve light response.SWCNT electrolyte-gate synaptic transistors displayed a low voltage range（±1 V）,high On/Off ratios（>10⁶）,low off currents(～10^-12 A),and good mechanical flexibility.The number of mobile ions and capacitance value of dielectrics are related to the weight concentrations of ILs in the electrolyte,which further influence the electrical performance and synaptic function imitation of the SWCNT synaptic transistors.The short-term synaptic behaviors,such as excitatory postsynaptic current（EPSC）and paired-pulse facilitation（PPF）,can be mimicked through the electrical stimulation in this device.Furthermore,the flexible ILs-c-PVP based SWCNT photoelectric synaptic transistors can be obtained by adding a lead-free material of BiI₃ into the electrolyte,and the transition of photosynaptic plasticity from short-term memory（STM）to long-term memory（LTM）were simulated under the optical signals.Surprisingly,the turn-on voltage of the SWCNT TFTs using BiI₃-ILs-c-PVP dielectrics decreases and approaches 0V,which lays a good foundation for low-power synaptic devices and electronic circuits.（4）The methods and mechanisms of threshold voltage adjustment and polarity conversion for SWCNT TFTs were explored in order to build SWCNT CMOS inverters with ultra-low power consumption.SWCNT TFTs can be turned on at 0 V by using BiI₃-doped ILs-c-PVP dielectrics and adjusting the sc-SWCNT printing process.It is because that the existence of ultrathin Ag I with ultralow work function can reduce the work function of Ag gate electrode,and the modulation of the sc-SWCNT ink concentrations and printing times can decrease the density of interface trapped charges.Subsequently,n-type SWCNT TFTs were selectively achieved via printing the polarity conversion ink consisting of 2-amino-2-methyl-1-propanol（AMP）with electron donating ability.Finally,the printed SWCNT CMOS inverters consisting of E-mode p/n-type TFT devices were successfully constructed with low operation voltage and full rail-to-rail output characteristics.Surprisingly,the record unit static power consumption of 6.75 f Wμm^-1 at V_DD of 0.2 V can be obtained（compared with reported flexible SWCNT CMOS inverters）which exhibit great potential in carbon nanotube-based power-efficient applications.In summary,the research results obtained in this work are of great significance for promoting the further application and development of SWCNT TFT devices in the field of neuromorphic-based artificial visual electronics.