Skin is the largest organ in the human body. It handles touch perception, temperature regulation, and barrier protection all at once, and it does this so well that engineers have spent years trying to build something similar. Thermoplastic polyurethane, or TPU, has become one of the most useful materials in that effort. It powers a new generation of flexible electronic sensors that stretch, heal, and respond to touch in ways that get closer to what real skin does.
This article breaks down how TPU-based bionic skin works, what makes it effective, and where the technology is going.
Why TPU works so well for flexible sensors
TPU is a block copolymer. Its molecular chain alternates between soft segments (polyester or polyether polyols) and hard segments (diisocyanates and chain extenders). That structure creates a microphase-separated material with a rare mix of stretchability, toughness, and tunable mechanical properties. In other words, exactly what you need for something that has to behave like skin.
Four things make TPU a strong pick for bionic skin.
1. You can tune its mechanical properties
By adjusting the soft-to-hard segment ratio (usually 40% to 80% soft segment content), manufacturers can set TPU’s tensile strength anywhere from 10 to 60 MPa and its elongation at break from 200% to 800%. That flexibility window means the same base material can work for a stiff joint sensor or a soft, skin-conformable patch. Researchers have made TPU porous membranes with tensile strengths around 1.3 MPa and elongation at break of 460%, which is enough to handle the mechanical demands of real wearable use.
2. It plays well with conductive fillers
TPU’s soft segment chain mobility lets conductive fillers like carbon nanotubes (CNTs), MXene, graphene, and silver nanowires (AgNWs) disperse evenly and form stable 3D conductive networks. This matters because sensing performance depends on how well the conductive network responds when the material deforms.
3. It supports many processing methods
TPU works with a wide range of fabrication techniques. You can melt-process it through injection molding, extrusion, or fused deposition modeling (FDM) 3D printing. You can also use solution-based methods like casting, electrospinning, or dip coating. This versatility means researchers can prototype in the lab and move to industrial production without changing the base material.
4. It is biocompatible
Medical-grade TPU has passed ISO 10993 biocompatibility testing for cytotoxicity, skin irritation, and sensitization. It is already used in catheters and wound dressings, so it has a proven safety record for skin-contact and even implantable applications.
How TPU bionic skin detects the world
The core job of any electronic skin is converting external stimuli (strain, pressure, temperature, humidity) into measurable electrical signals. TPU-based bionic skin uses several sensing mechanisms, each with its own strengths and trade-offs.
Piezoresistive sensing: the most common approach
Piezoresistive sensing is the most widely used mechanism in TPU bionic skin. The principle is simple: when force deforms the sensor, the conductive network changes shape or microstructure, causing a measurable shift in resistance. Three sub-mechanisms drive this.
Conductive network mechanism. Conductive fillers dispersed in the TPU matrix form continuous pathways above the percolation threshold. When the material stretches, the spacing between fillers increases and some pathways break, raising resistance. Release the strain and the network rebuilds. Researchers have hit gauge factors (GF) of 13, 73, and 363 across different strain ranges (0 to 40%, 40 to 80%, and above 80%) using TPU/MWCNTs@MXene composite foams. Other teams pushed GF values to 6,834 with strain ranges up to 604% using dual-layer conductive fibers that combine silver nanoparticles and CNTs.
Microcrack mechanism. When a rigid conductive layer deposited on TPU stretches, microscopic cracks form at the surface. The edges of these cracks act like switches, making and breaking contact. The result is dramatic resistance changes at tiny strains. One team built silver nanoparticle networks on TPU fibers that reached a GF of 9.52 x 10^4, with a response time of 21 milliseconds and a detection limit of 0.05% strain. Another group made superhydrophobic conductive fabrics with a GF of 22,500 across a 0 to 98% strain range.
Percolation tuning. Near the percolation threshold (around 3.5% carbon black content by weight), even tiny deformation causes sharp changes in conductive pathways. Researchers used screen printing to create periodic microstructures on TPU/carbon black films and got pressure sensitivity of 5.205 kPa^-1 in the 0 to 100 kPa range.
Capacitive sensing: precise at low pressures
Capacitive sensors work on the parallel-plate capacitor model. They detect changes in electrode spacing, effective area, or dielectric constant. They do well in low-pressure regimes (below 10 kPa) and offer better linearity and lower power consumption than piezoresistive types. One research group built a sensor with electrospun TPU nanofibers, ionic liquid, and bismuth oxybromide nanosheets as the dielectric layer. It reached 327.8 kPa^-1 sensitivity across 0 to 50 kPa and held up through nearly ten thousand cycles.
The catch: capacitive sensors are vulnerable to electromagnetic interference and need complex readout circuitry, which limits how practical they are in messy real-world wearable setups.
Piezoelectric and triboelectric sensing: generating their own power
Both piezoresistive and capacitive sensors need external power. Piezoelectric and triboelectric mechanisms avoid that problem by generating their own signals.
Piezoelectric sensing works because certain materials, like polyvinylidene fluoride (PVDF) or zinc oxide nanowires, produce electrical charges under mechanical stress. Combined with TPU’s elastic matrix, these composites can detect dynamic forces with response times around 30 milliseconds. A particularly interesting application is a self-powered bionic material with a branched nanofiber scaffold that senses and also promotes wound healing through electrical stimulation.
Triboelectric sensing exploits contact electrification and electrostatic induction. TPU’s high electronegativity in the triboelectric series makes it a good friction layer for triboelectric nanogenerators (TENGs). One team built a stretchable TENG-based electronic skin that delivers 202.4 volts of open-circuit voltage and 6 mW/m^2 of power density from a 2 cm x 2 cm sample. It also had pressure sensitivity of 78.4 kPa^-1 and a 1.4-millisecond response time.
Temperature and humidity sensing
Human skin does more than feel pressure. It senses heat and moisture too. Adding temperature and humidity detection to electronic skin is not just about mimicking skin. It has practical value: temperature sensors can flag fever or inflammation, and humidity sensors can catch signal drift caused by sweat buildup at the skin-sensor interface.
TPU’s natural hydrophobicity gives a stable baseline for humidity sensing. By adding hydrophilic fillers like MXene, researchers can add controlled moisture responsiveness. One team built an all-nanofiber humidity sensor using TPU/MXene that reached -91% linear sensitivity with a 2.3-second response time across 11 to 95% relative humidity.
Multimodal decoupling: separating mixed signals
Real-world movement generates strain, pressure, temperature, and humidity signals all at once. Without decoupling, a stretch might look the same as a temperature change in the raw data. Two main strategies address this.
Structural decoupling physically separates sensing functions into different layers or regions of the device. Sandwich-structured composites, for instance, can isolate pressure sensing from humidity protection layers.
Material matching decoupling uses different material systems that respond selectively to specific stimuli on the same substrate. One approach uses nickel oxide for temperature sensing (based on negative temperature coefficient thermistor principles) and TPU@polyaniline for pressure sensing (capacitive principle). This gave independent temperature and pressure readouts.
Microstructure design: engineering sensitivity from the bottom up
Microstructure is where TPU bionic skin gets its performance edge. By engineering the physical structure at microscopic scales, researchers can amplify signal responses substantially.
Surface microarrays: microdomes, micropyramids, and microcolumns
Building ordered microstructure arrays on the sensing surface creates gradient changes in contact area under compression. This approach is the most mature in terms of manufacturability, with clear quantitative relationships between geometric parameters and sensing performance. Mushroom-inspired microstructures, for example, have been used to make bioelectrodes with better skin friction coefficients and higher signal-to-noise ratios than commercial gel electrodes.
Porous and foam structures
Introducing 3D interconnected pores into the TPU matrix cuts density and elastic modulus dramatically, which boosts sensitivity. This is currently the most active research area. Methods include salt templating, nonsolvent-induced phase separation, and freeze-drying, each offering different pore sizes and distributions.
One notable idea came from polar bear hair. Researchers built a hollow, porous structure through coaxial wet spinning and got a GF of 102 at 300% strain. Another team made a gradient porous pressure sensor using freeze-drying that covered 0 to 300 kPa with 2.609 kPa^-1 sensitivity.
The fundamental problem with porous structures is a tension between two goals. Higher porosity means higher sensitivity but lower mechanical strength. At large deformations, pore walls collapse and performance drops for good.
Biomimetic hierarchical structures
Nature has blueprints for solving the sensitivity-versus-strength problem. Three bio-inspired approaches have produced striking results.
Nacre-inspired “brick-and-mortar” structures. By alternately electrospinning TPU fiber membranes and spraying MXene/CNTs conductive ink, researchers reached a GF of 5.8 x 10^4, a sensing range of 0 to 535%, and a detection limit of 0.15%.
Mussel-inspired multilayer structures. Using Ti-O-C covalent bonding between MXene and reduced graphene oxide layers, a team pushed GF past 8.43 x 10^4 with a 0 to 200% range and 70-millisecond response time.
Spider slit organ-inspired structures. Combining TPU fluorescent fiber membranes with MXene/CNTs conductive networks, researchers got a maximum GF of 3.92 x 10^7. That is the highest reported value. It also had a 5-millisecond response time and a detection limit of 0.001% strain. The design added force-induced luminescence for visual sensing and TENG self-powering.
These biomimetic structures are the cutting edge, but their fabrication complexity and poor reproducibility keep them in the lab for now.
Fabrication methods: from lab bench to production line
How TPU bionic skin is made has a huge effect on its performance. Four families of fabrication methods dominate the field.
Electrospinning: the research favorite
Electrospinning produces ultrafine fibers that form 3D porous networks, which is ideal for breathable, skin-conformable sensors. Four variants are in use.
Solution electrospinning dissolves TPU in solvents like DMF or THF. It produces the finest fibers (220 nm to 1.1 um) and, with filler compositing and microcrack design, can reach GF values between 428 and 83,982. The downside is organic solvent use, which raises environmental and safety concerns.
Melt electrospinning operates at 205 to 208 degrees Celsius without solvents. That makes it greener and more biocompatible. It deposits material four times faster than solution spinning but produces thicker fibers (4 to 8 um), which limits sensitivity.
Coaxial electrospinning creates core-shell fibers that keep conductivity above 250% strain. It allows flexible combination of unconventional conductive materials like liquid metal, but it needs precise control of core/shell flow rates and produces fibers larger than 2 um.
Freeze spinning creates oriented porous structures with excellent linearity (R^2 = 0.998) and elongation at break up to 618%, though sensitivity stays lower.
3D printing: customization freedom
3D printing enables complex 3D structures that traditional methods cannot make. It is particularly useful for personalized medical devices. Researchers have used fused filament fabrication with conductive TPU/carbon black filaments to make strain sensors with 217% elongation at break and 14 MPa tensile strength. Direct ink writing has been used to print silver interdigitated electrodes, which boosted gauge factors in small-strain regions.
The main limitation is that there are not many commercially available conductive TPU filaments, and balancing conductivity with mechanical performance is still hard.
Dip coating and spray coating: simple and scalable
These methods deposit conductive layers onto porous or fibrous substrates quickly and cheaply. They are the most scalable approaches and have been used for large-area sensor fabrication. However, batch-to-batch consistency in coating uniformity and interface adhesion is still a manufacturing bottleneck.
In-situ polymerization and chemical vapor deposition
These techniques grow conductive materials directly on or within the TPU matrix, creating strong chemical bonds at the interface. One team built a “sandwich” structured TPU/silver nanoparticle/graphene composite with an initial conductivity of 1.4 x 10^5 S/m and 1,000% stretchability. Oxidative chemical vapor deposition has been validated as a solvent-free route for depositing polypyrrole coatings, which eliminates residual solvent toxicity. That is a critical advantage for moving from wearable to implantable applications.
The trade-off is high equipment cost and low deposition rates, which keep these methods in the laboratory.
Where the technology stands now
TPU-based bionic skin has come a long way across three areas: sensing mechanisms, microstructure design, and fabrication processes. Piezoresistive sensors now cover a sensitivity range from GF = 10 to 10^5. Capacitive and triboelectric mechanisms provide low-power and self-powered alternatives. Temperature, humidity, and multimodal decoupling strategies have started to address signal cross-talk.
Microstructure design has moved from simple templates to biomimetic hierarchical structures. Porous structures bring elastic modulus down to skin-matching levels, and nacre-inspired designs break through the sensitivity-range trade-off.
Fabrication methods span from laboratory electrospinning to industrial dip coating, with each method filling a specific niche.
What is still hard
A few bottlenecks remain before TPU bionic skin reaches wide commercial use.
Sensitivity versus stability. Ultra-high sensitivity sensors (GF above 10^4) mostly rely on microcrack mechanisms, which have narrow linear ranges and poor cycling stability. Real wearable applications need sensors that hold up over thousands of deformation cycles.
The lab-to-factory gap. The high sensitivities achieved in research are hard to reproduce consistently at industrial scale. Batch consistency is the critical bottleneck for commercialization.
Long-term reliability. Most studies test single functions in short-term setups. Data on signal drift, sweat interference, and biocompatibility over extended wear periods is still thin.
The road ahead
Humanoid robotics, personalized medicine, and the Internet of Things are all growing fast. TPU-based bionic skin is likely to move from single sensing elements toward integrated systems that combine perception, actuation, energy harvesting, and communication. The most promising path is probably not optimizing one method but mixing multiple fabrication techniques, multiple microstructure strategies, and multiple sensing mechanisms together.
Nobody has built artificial skin that can feel, heal, and power itself as well as the skin we were born with. TPU is getting closer, but there is still a real gap between the lab and the clinic, and between the clinic and the factory floor.