Self-adhesive polymer substrates turn independently optimized transistor modules into snap-together electronic skin that senses, learns, and computes without performance trade-offs.
(Nanowerk Spotlight) The human peripheral nervous system balances specialization against coordination. Receptor cells each do one thing exceptionally well: some detect pressure, others heat, others chemical signals. What transforms these isolated specialists into a functioning whole is a universal wiring scheme, synapses and nerve fibers, that connects them regardless of what they sense.
Electronic skin for robots faces an analogous challenge. The organic electrochemical transistors (OECTs), best suited for detecting sweat ions require different semiconductor materials and electrolytes than those designed to mimic synaptic learning or execute logic operations.
Fabrication processes can conflict directly: a solvent needed to deposit one semiconductor may dissolve the flexible substrate required by another. Monolithic integration, building everything on a single chip, forces every component to tolerate conditions optimized for its neighbors.
Rather than compromising each device to coexist during fabrication, the team built each at its best and then assembled them into a unified electronic skin through simple physical stacking.
(a) Schematic diagram of the perception process of robots (left) and humans (right). (b) “Neural building block” OECTs e-skin modules for sensing, synaptic emulation, and logic operations in HMI. (Image: Reproduced with permission from Wiley-VCH Verlag) (click on image to enlarge)
The SEBS substrates serve as both structural foundation and universal interface. Their native self-adhesiveness eliminates the need for external bonding agents, while their flexibility allows the assembled stack to conform to curved surfaces. The team validated these junctions through peel tests, cyclic deformation, and prolonged humidity exposure, confirming that the interfaces maintain both mechanical cohesion and electrical continuity under realistic operating conditions.
The first sensing module targets potassium and sodium ions in sweat, biomarkers tied to hydration status, muscle fatigue, and neural activity. It pairs a PEDOT:PSS channel with ion-selective polymer membranes in a coplanar side-gate architecture. This flat design avoids the liquid electrolyte encapsulation that complicates conventional top-gate structures and provides a smooth surface for subsequent module stacking.
The critical question was whether this simplified geometry could match the sensitivity of more complex designs. Across a three-order-of-magnitude concentration range, the module achieved sensitivities of 246 µA per decade for potassium and 60 µA per decade for sodium. Switching response times fell in the millisecond range, and the on/off current ratio held steady through 2000 continuous pulse cycles.
A second module addresses temperature, essential for both interaction safety and health monitoring. Here the team used a solid-state electrolyte made by embedding a biocompatible ionic liquid in a gelatin matrix. To push sensitivity beyond what thin-film designs typically deliver, they increased the PEDOT:PSS channel thickness through drop-casting. A thicker channel accommodates more ion injection under the same thermal driving force, amplifying the current response to small temperature shifts.
The underlying mechanism couples two effects. As temperature rises, the gelatin electrolyte transitions from gel to sol, sharply increasing ionic conductivity. Simultaneously, thermal expansion of the polymer substrate reshapes the conductive network within the channel. Together these processes yielded a temperature response of 32 µA per degree Celsius.
Beyond sensing, the platform includes a neuromorphic computing module that emulates biological synapses. This module pairs a P3HT semiconductor channel with an ionic gel electrolyte. Depositing P3HT posed a specific fabrication problem: its chloroform solvent dissolves SEBS. The team circumvented this by spray-coating the semiconductor onto a 120 °C hotplate, evaporating the solvent before it could attack the substrate.
The device reproduced key features of synaptic plasticity. Paired-pulse facilitation, a signature of short-term memory, decayed with relaxation times that closely tracked biological values. Adjusting pulse amplitude or duration shifted behavior from volatile short-term states to stable long-term potentiation. A learning-forgetting-relearning test captured experience-based efficiency gains: initial learning required 25 seconds, a second attempt took 11.5 seconds, and a third dropped to 5.4 seconds.
The fourth module brings logic operations to the sensory edge, enabling local preprocessing without routing signals to a central processor. The team built a complementary inverter by pairing p-type and n-type OECTs on a shared solid-state electrolyte, achieving a voltage gain of 11.74 V/V at a supply voltage of 0.8 V.
They also discovered stable anti-ambipolar behavior in one of their semiconductor materials when operated in a gelatin-based ionic gel. This unusual bell-shaped transfer characteristic allowed a single pair of transistors to switch between AND and NOR logic, or between OR and NAND logic, by changing the input voltage range alone.
Traditional flexible electronic skin logic requires separate circuits for each Boolean function, rapidly multiplying device count. The anti-ambipolar approach extracts multiple operations from the same physical circuit.
The assembled system operated under 30% stretching strain in both aqueous and gel electrolyte environments. Each sensing module maintained stable performance after 10 days in simulated sweat conditions. Because the architecture is modular, a robotic hand performing delicate chemical monitoring could carry a different configuration than an arm designed for thermal safety, all drawn from the same library of components.
Signal flow between modules currently depends on external wiring. A fully autonomous cascade, where a sensor output directly updates a synaptic weight and triggers a logic decision, remains an open challenge. Future work targets integrated circuit topologies and three-dimensional stacking to close this gap, moving toward a complete artificial peripheral nervous system for soft robotics, intelligent prosthetics, and human-machine interfaces.
For authors and communications departmentsclick to open
Lay summary
Prefilled posts
Nanowerk Newsletter
Get our Nanotechnology Spotlight updates to your inbox!
Thank you!
You have successfully joined our subscriber list.
Become a Spotlight guest author! Join our large and growing group of guest contributors. Have you just published a scientific paper or have other exciting developments to share with the nanotechnology community? Here is how to publish on nanowerk.com.
