Highly Stretchable, Directionally Oriented Carbon Nanotube/PDMS Conductive Films with Enhanced Sensitivity as...

Recent interest in the fields of human motion monitoring, electronic skin, and human-machine interface technology demands strain sensors with high stretchability/compressibility (ε > 50%), high sensitivity (or gauge factor (GF > 100)), and long-lasting electromechanical compliance. However, current metal- and semiconductor-based strain sensors have very low (ε < 5%) stretchability or low sensitivity (GF < 2), typically sacrificing the stretchability for high sensitivity. Composite elastomer sensors are a solution where the challenge is to improve the sensitivity to GF > 100. We propose a simple, low-cost fabrication of mechanically compliant, physically robust metallic carbon nanotube (CNT)-polydimethylsiloxane (PDMS) strain sensors. The process allows the alignment of CNTs within the PDMS elastomer, permitting directional sensing. Aligning CNTs horizontally (HA-CNTs) on the substrate before embedding in the PDMS reduces the number of CNT junctions and introduces scale-like features on the CNT film perpendicular to the tensile strain direction, resulting in improved sensitivity compared to vertically-aligned CNT-(VA-CNT)-PDMS strain sensors under tension. The CNT alignment and the scale-like features modulate the electron conduction pathway, affecting the electrical sensitivity. Resulting GF values are 594 at 15% and 65 at 50% strains for HA-CNT-PDMS and 326 at 25% and 52 at 50% strains for VA-CNT-PDMS sensors. Under compression, VA-CNT-PDMS sensors show more sensitivity to small-scale deformation than HA-CNT-PDMS sensors due to the CNT orientation and the continuous morphology of the film, demonstrating that the sensing ability can be improved by aligning the CNTs in certain directions. Furthermore, mechanical robustness and electromechanical durability are tested for over 6000 cycles up to 50% tensile and compressive strains, with good frequency responses with negligible hysteresis. Finally, both types of sensors are shown to detect small-scale human motions, successfully distinguishing various human motions with reaction and recovery times of as low as 130 ms and 0.5 s, respectively.