同济大学《‌CRPS》：基于碳纳米管/石墨烯/TPU的电子皮肤，用于仿生设备和智能可穿戴设备

成果简介

随着仿生设备和智能可穿戴设备的快速发展，电子皮肤（E-skin）在太空垃圾回收、医疗诊断和康复领域的应用潜力日益凸显。本文，同济大学袁光杰 副教授、李智军等在《Cell Reports Physical Science》期刊发表名为“Highly stretchable and microporous hybrid film-based electronic skins for biomimetic devices and smart wearables”的论文，研究报道了基于高伸展性微孔碳纳米管/石墨烯/热塑性聚氨酯复合薄膜的伸展/温度敏感型（S/T-S）和压力敏感型（P-S）电子皮肤。在 ε 变形率为 0%–340% 和 340%–480% 的范围内，S/T-S 电子皮肤的应变系数（GF）分别为 12.19 和 635.03。此外，预拉伸处理显著提升了其在拉伸-释放循环中的重复性。压力-拉伸型电子皮肤在5–20,005 Pa压力范围内具有1.79 kPa⁻¹的应变系数。凭借卓越性能，该电子皮肤被集成至仿生蛙舌结构，实现环境温度、伸缩及触觉信号检测。同时提出基于机器学习的足部姿势分析系统，其识别准确率高达96.2%。

图文导读

图1. Preparation process and morphology of the CNT/G/TPU hybrid film and structural schematics of the resulting E-skins

(A) Schematic illustration of the preparation process of the CNT/G/TPU hybrid film.

(B) Optical images of the original TPU and hybrid films (scale bars, 1 cm and 50 μm).

(C) Optical images showing the flexibility of the hybrid film under twisting and bending.

(D) Optical images of the hybrid film at ε values of 0% and 480% (scale bar, 80 mm).

(E) Structural schematic of the S/T-S E-skin.

(F) Structural schematic of the P-S E-skin.

图2. Stretching performance and mechanism of S/T-S E-skins with different G contents

(A) Relationship between ε and ΔR/R₀ of the E-skins with varying G contents.

(B) Dependency of the maximum GF and ε on the G content.

(C) FE-SEM image of the CNT/TPU film (scale bars, 1 μm and 100 nm).

(D) FE-SEM image of the G/TPU film (scale bars, 1 μm and 100 nm).

(E) FE-SEM image of the CNT/G/TPU film with a G content of 60 wt % (scale bars, 1 μm and 100 nm).

图3. Stretching sensing performance of the S/T-S E-skin

(A) Schematic diagram of the sensing mechanism during the stretching-releasing process (scale bars, 10 μm and 100 nm).

(B) I-V curves with various ε values.

(C) ΔR/R₀ variation as a function of time stepwise.

(D) Dependency of ΔR/R₀ on ε.

(E) Response and recovery times as ε varied from 0% to 10%.

(F) Hysteresis curves with ε values of 20%, 50%, and 100%.

(G) Long-term durability test with a maximum ε of 200% and 3000 stretching-releasing cycles after 5 prestretching cycles under 210% and typical cycles of the test.

图4. Temperature sensing performance of the S/T-S E-skin

(A) Schematic diagram of the sensing mechanism during the cooling-heating process.

(B) I-V curves at various temperatures.

(C) Relationship between ΔI/I₀ and temperature.

(D) Response and recovery times with the temperature varying from 25°C to 85°C.

(E) Hysteresis curves for temperatures ranging from 25°C to 85°C.

(F) ΔI/I₀ with the temperature varying from 25°C to 45°C, 65°C, or 85°C during 5 repeated cooling-heating cycles.

(G) Long-term durability test at temperatures varying from 25°C to 45°C during 50 cooling-heating cycles.

图5. Pressure sensing performance of the P-S E-skin

(A) Schematic diagram of the sensing mechanism during the loading-unloading process.

(B) I-V curves at various pressures.

(C) Relationship between ΔI/I₀ and pressure.

(D) Response and recovery times under pressures ranging from 5 to 50,005 Pa.

(E) ΔI/I₀ with pressures varying from 5 to 105, 1,505, 10,005, 15,005, 20,005, and 50,005 Pa during 5 loading-unloading cycles.

(F) Long-term durability test at pressures varying from 5 to 50,005 Pa during 3,000 loading-unloading cycles and typical cycles of the test.

图6. Application of the E-skins in biomimetic devices

(A) Schematic diagram of a biomimetic frog tongue and its potential application in the field of space debris retrieval.

(B) Variations in the ΔI/I₀ of the P-S E-skin and the ΔR/R₀ of the S/T-S E-skin after prestretching during the predatory process of the biomimetic tongue.

(C) Variations in the ΔI/I₀ of the S/T-S E-skin with different environmental temperatures.

图7. Application of the E-skins in smart wearables

(A) Schematic diagram of an injured athlete with an intelligent kneepad and smart insoles during the telediagnosis process.

(B) Schematic diagram of the detection platform.

(C) Schematic diagram of the intelligent kneepad and its electrical curves with different degrees of knee flexion as well as different temperatures of thermotherapy.

(D) Schematic diagram of the smart insole and its ΔI/I₀ as well as surface pressure distributions with different foot postures.

(E) Data processing of machine learning based on the KNN algorithm.

(F) Confusion matrix of machine learning outcomes.

小结

在此研究中，基于高伸展性微孔碳纳米管/聚烯烃/热塑性聚氨酯复合薄膜，成功制备了S/T-S和P-S电子皮肤，并实现了包括ε值、温度和压力在内的多种信号感知。S/T-S 电子皮肤在 0%–340% 和 340%–480% 的拉伸范围内具有 12.19 和 635.03 的 GF 值，在 25°C–85°C 的温度范围内具有 3.66 × 10−3°C −1 的值。P-S电子皮肤在5–20,005 Pa和20,005–50,005 Pa压力范围内分别具有1.79和0.2 kPa−1的GF值。此外，预拉伸工艺显著提升了电子皮肤的重复性，使其在拉伸-释放循环中保持固定ε值时最大ΔR/R0值基本恒定，解决了电子皮肤在大ε值下常见的不稳定问题，提高了应用可行性。此外，将S/T-S型与P-S型电子皮肤集成于仿生蛙舌装置，成功监测其伸展状态、环境温度及触觉信号。同时分别将两类电子皮肤集成于智能护膝与智能鞋垫。通过监测膝关节运动与热疗温度，同时提出结合机器学习的足部姿势分析系统，实现96.2%的高识别准确率。

文献：https://doi.org/10.1016/j.xcrp.2026.103109