Decoupling thermo-mechanical signals in ionic hydrogels via deep operator networks.

Area of Science:

• Soft Matter Physics
• Materials Science
• Artificial Intelligence

Background:

• Responsive hydrogels generate coupled electrical signals from thermal and mechanical stimuli.
• Superposition of ionic thermoelectric and piezoionic effects poses a challenge in soft ionotronics.
• Disentangling these signals is crucial for developing advanced soft electronic devices.

Purpose of the Study:

• To develop a novel model for disentangling inseparable electrical signals in responsive hydrogels.
• To address the fundamental challenge in soft ionotronics caused by coupled stimuli.
• To enable real-time, field-level separation of thermal and mechanical influences.

Main Methods:

• Proposed the T-DeepONet model, integrating transformer temporal modeling with DeepONet spatial encoding.
• Trained the model on a synthetic dataset from finite element simulations.
• Developed a method to map coupled voltage fields to independent temperature and pressure distributions.

Main Results:

• T-DeepONet achieved 98.2% R2 accuracy in resolving distinct spatiotemporal signatures of thermal diffusion and mechanical transients.
• Demonstrated accurate disentanglement across synchronous and asynchronous loading scenarios.
• Achieved an inference latency of approximately 100 ms for real-time analysis.

Conclusions:

• Established a general framework for real-time multiphysics signal disentanglement in soft matter.
• The T-DeepONet model offers a pathway for high-fidelity tactile perception in soft robotics.
• Bridged advances in nonequilibrium ion transport with operator learning for future innovations.