I. Core Research Summary

• High-precision direct-write semiconductor carbon nanotube conductive channel via aerosol printing

Ultrasonic atomization purifies (6,5) single-walled carbon nanotube ink, enabling CNC‑controlled, precise deposition of a 50 μm channel array. The resulting thin film features an interwoven carbon‑nanotube network that delivers high‑mobility electrical conductivity. The device is fabricated at a low temperature of 100 °C on a flexible PI substrate, ensuring no thermal damage.

• Dual-mode synaptic response, with gate-voltage–independent regulation of plasticity.

Turn-on under negative gate voltage Excitement Mode : p‑type doping is achieved by ion implantation into carbon nanotubes under compressive stress, with the current steadily increasing; switching between positive and zero gate bias. Inhibition mode , pressure induces ion release from the channel, leading to current decay; the tactile response can be independently tuned and decoupled from pressure stimulation, thereby replicating the intrinsic synaptic plasticity of biological systems.

• Long-Term Memory Characteristics of Tactile Signals

The pyramidal structure constrains the ion diffusion rate, and upon pressure release, the dopant state decays slowly. The pressure‑pulse signal can be retained for up to 8 × 10⁴ seconds (≈22 hours), far surpassing the sub‑second memory timescale of conventional ionic gel devices, thereby enabling continuous temporal storage of tactile events.

• Ultra-low-power stable sensing

The operating voltage is as low as −0.1 V, and the single‑synapse weight can be tuned with energy consumption as low as 0.5 nJ. After 100 consecutive stress cycles, its electrical performance shows no significant degradation, and it operates stably within the human‑body‑compatible temperature range of 20–36 °C.

• In-situ Fluorescence Visualization of the Doping Mechanism

Near-infrared photoluminescence (PL) enables real-time observation of localized quenching in doped regions of carbon nanotubes under pressure, providing direct evidence that ion‑gel deformation gives rise to a locally reversible doping–dedoping mechanism and offering a characterization tool for optimizing synaptic performance.

II. Complete Preparation Process and Core Advantages of Aerosol Spraying

(1) Complete Tactile Synaptic Aerosol Printing Process

• Spin-coat the PI precursor solution onto a silicon wafer, then thermally cure it on a flexible polyimide substrate; deposit Cr/Au source–drain and side-gate metal electrodes via electron-beam evaporation.
• Purified (6,5) SWCNT‑based ink with low viscosity, specially formulated for ultrasonic atomization;
• An aerosol nozzle performs site-specific direct writing of transistor conductive channels, with a printing speed of 0.5 mm/s, a stage temperature maintained at 100°C, and uniform deposition of carbon nanotube networks.
• Tetrahydrofuran is used to clean and remove residual additives, followed by delamination of the flexible PI film.
• A pyramid‑shaped polyurethane ionogel was fabricated via silicone mold casting and then used to encapsulate the device by conformally adhering to the carbon‑nanotube channel.

(II) Five Core Process Advantages (Compared with Lithography, Inkjet, and Vacuum Coating)

• High-precision, fixed-point direct writing of carbon nanotubes

Aerosol‑beam focusing enables the fabrication of micron‑scale, uniform carbon‑nanotube films, allowing for the selective deposition of carbon nanotubes exclusively in the channel region. In contrast, inkjet printing tends to spread over large areas, and photolithography requires multiple mask steps, thereby significantly simplifying the preparation of the transistor’s active layer.

• Low-temperature flexible compatible molding

The entire process employs a 100°C substrate‑treatment protocol that preserves the PI flexible substrate, enabling direct fabrication of neuromorphic devices on various thin, flexible substrates and accommodating curved‑surface processing for wearable electronics.

• The carbon nanotube network exhibits excellent uniformity.

Ultrasonic atomization breaks up carbon‑nanotube agglomerates, while the uniform spreading of droplets forms a continuous, interwoven network. As a result, the device’s on‑state current and synaptic response consistency are significantly superior to those achieved by other printing approaches.

• Digital, maskless rapid iteration

With CAD, you can instantly adjust the channel’s width and length and configure array layouts. Tactile‑sensory devices of varying dimensions can be prototyped in just a few hours—no custom lithography masks required—making it ideal for developing electronic skins across multiple form factors.

• Compatibility of wide-carbon-based inks

Compatible with various single-walled and multi-walled carbon nanotube and graphene dispersions, this single system can fabricate a range of carbon-based flexible electronic components, including sensors, synaptic devices, and actuators.

III. Five Core Technological Innovations of This Study

Innovation 1: Precise Preparation of Aerosol‑Based SWCNT Neural‑Morphic Transistor Channels

Innovation 2: A Pyramid-Structured Ionogel-Based Slow-Diffusion Mechanism

Innovative three‑gate voltage–dependent regulation of excitatory/inhibitory bidirectional synaptic transmission

Innovation 4: Ultra-Low-Voltage, Low-Power Bionic Tactile System

Innovative In-Situ PL Visualization and Doping Characterization Platform 5

IV. Existing Process Challenges and Corresponding Solutions for Overcoming Them

Challenge 1: In conventional printing, carbon nanotubes exhibit severe aggregation, resulting in poor transistor switching performance.

Challenge 2: In conventional ionogels, ion diffusion is rapid, resulting in tactile memory lasting only a few seconds.

Challenge 3: The lithography process is complex, and the prototyping cycle for flexible synapses is lengthy.

Challenge 4: Most tactile devices exhibit only a single type of response, making it impossible to mimic neural plasticity.

Challenge 5: High-temperature processes damage flexible wearable substrates.

V. Future Diversified Industrial Application Scenarios

• Bionic Prosthetic Tactile Skin

• Smart Sensing Glove for Rehabilitation Training

• Humanoid Robot Sensing Neural System

• Human–computer interaction flexible touch device

• In Vitro Neural Bionics Research Platform

Figure 1. Tunable excitatory and inhibitory synaptic behavior in ion‑gel‑gated SWCNT transistors. a) Schematic of a biological tactile neuronal system: under the influence of a sensing‑gated voltage, excitatory and inhibitory neurotransmitters are selectively released; b) Schematic of an artificial tactile neuronal system built on an ion‑gel‑gated SWCNT transistor employing a three‑dimensional patterned polyurethane ion‑gel matrix; c) Chemical structures of the mobile anions [FAP]⁻ and cations [EMIM]⁺ contained within the ion‑gel matrix; d) Proposed schematic of the operating mechanism of the artificial tactile synapse; e) Overall concept of the tunable synaptic plasticity of the ion‑gel‑gated SWCNT transistor: the tactile synapse can elicit distinct responses—either an excitatory or an inhibitory mode—to the same stimulus, depending on the applied gating voltage.

Figure 2. Pulse-dependent sensing and memory performance under Vds = −0.1 V. a) Pressure response of the tactile synapse at different pulse amplitudes (pulse duration: 500 ms); b) Pressure response of the tactile synapse at different pulse durations (pressure pulse amplitude: 8.71 kPa); c) Comparison of peak Ids values (ΔIds/I0) and synaptic weights (SW) as a function of varying pressure pulse amplitudes; d) Comparison of ΔIds/I0 and SW across different pulse durations; e) Decay curve of the leakage current (Ipeak − I0)/I0 over the first 240 seconds when no pressure pulse is applied, as shown in panel a; f) Decay curve of the leakage current (Ipeak − I0)/I0 over the first 240 seconds when no pressure pulse is applied, as shown in panel b.

Figure 3. Excitatory and inhibitory synaptic responses under sequential pulse stimulation. a, b) Changes in excitatory and inhibitory responses of the postsynaptic current Ids as a function of pulse amplitude: 100 pulses, pulse width of 500 ms, and inter‑pulse interval of 10 s. c) Values of I9.5s/In0 corresponding to the first 10 pulses in panel a (excitatory mode, upper panel) and panel b (inhibitory mode, lower panel). d, e) Excitatory and inhibitory modulation of the postsynaptic current Ids as a function of gating voltage: pressure pulse amplitude of 5 kPa, pulse duration of 500 ms, and inter‑pulse interval of 10 s. f) Values of I9.5s/In0 corresponding to the first 10 pulses in panel d (excitatory mode, upper panel) and panel e (inhibitory mode, lower panel).

Figure 4. Analysis of the retention time and cycling stability of the synaptic tactile sensor. a) Current decay curve of the synaptic tactile sensor after 24 consecutive pressure pulses with a 40‑second inter‑pulse interval. The pressure pulse amplitude was 4.3 kPa, and the pulse duration was 300 ms. b) Cycling stability test results of the synaptic tactile sensor after 100 consecutive loading cycles. The pressure pulse amplitude was 1.56 kPa, with a pulse duration of 400 ms; the inter‑pulse interval was 280 s.

Figure 5. Doping characteristics of (6,5) single-walled carbon nanotubes as a function of time and pressure, characterized by in situ photoluminescence imaging. a) Optical image of the channel region in a (6,5) SWCNT network transistor, highlighting the pyramidal structure of the ionic gel; b, c) Trends in the drain current (Ids) during and after pressure application (green and orange circles indicate the time points at which frames were recorded; see Movie S1 in the Supporting Information); d) Photoluminescence images corresponding to the orange circles (frames 1–8 in panel c), illustrating changes before, during, and after pressure loading: blue regions indicate reduced photoluminescence intensity, reflecting enhanced p-type doping of the single-walled carbon nanotubes; dashed lines denote the positions of the source and drain electrodes.