High-Temperature Flexible Sensors: The Qualification Gap
A new polyacrylonitrile fiber-mat sensor combines temperature and pressure perception on a flexible robotic gripper. Its real significance is not a single maximum-temperature claim, but a manufacturable route toward multimodal sensing in heat—provided that coupled-load calibration, packaging, drift, and yield can be qualified.
MEMSWork Intelligence Dashboard
| Market status | Peer-reviewed laboratory prototype with robotic-gripper integration; not a qualified commercial product |
|---|---|
| Primary constraint | Maintaining calibrated, mechanically intact temperature and pressure readout under simultaneous heat and load |
| Demand driver | Contact-aware automation, hazardous-environment robotics, equipment monitoring, and battery-safety instrumentation |
| Time horizon | Near-term research and application-specific pilots; production timing remains uncertain |
| Technologies to watch | Partially carbonized PAN, high-temperature electrodes and adhesives, array calibration, thermal-mechanical decoupling |
| Strategic signal | The defensible asset may be the qualified sensor stack and process window, not the sensing material alone |
| Next trigger | Independent cycling, drift, batch-uniformity, environmental-sealing, and application-standard data |
The decision question: what did the prototype actually prove?
On September 29, 2026, researchers at Xi’an Jiaotong University reported a flexible sensor made from treated polyacrylonitrile (PAN) fiber mats that measures temperature and pressure using separate thermoresistive and piezoresistive elements. The peer-reviewed device operated from 25–300 °C for temperature and 0–480 kPa for pressure, with 0.1 °C temperature resolution and room-temperature pressure response/recovery times of 53.2/8.8 ms.
Those are meaningful results for flexible electronics. They do not yet define a product operating envelope. The same paper reports that, when temperature exceeded 180 °C while pressure exceeded 150 kPa, simultaneous loading caused the PAN mat to lose structural integrity and reduced the rate of current change. The commercially important question is therefore not “Can a flexible sensor survive 300 °C?” It is “Can the complete sensor stack deliver traceable temperature and force data across the combined conditions, duty cycles, surfaces, and failure modes of a specific application?”
What changed technically
PAN is normally an electrical insulator. The researchers used laser processing or furnace treatment to partially carbonize the fiber mat, creating conductive pathways while retaining a porous, compressible network. In their process study, material treated below 600 °C remained insulating; treatment between 600 °C and 1000 °C produced useful conductivity; and exposure above 1100 °C destroyed the fibrous structure as the carbonized PAN oxidized and volatilized. A sample treated at 700 °C for 30 minutes gave the best pressure sensitivity in the tested set, reaching 0.78 kPa−1.
The architecture matters as much as the chemistry. Two PAN elements provide independent readout: one uses resistance change with temperature; the other uses the formation of conductive contacts as the porous mat compresses. The elements sit on a polyimide substrate with interdigitated gold, silver, or copper electrodes made using photolithography, exposure, and magnetron sputtering, while a screen-printed ceramic adhesive holds the mats in place. That combination links a scalable fibrous material to familiar MEMS and printing processes rather than requiring an entirely new manufacturing ecosystem.
Published evidence and remaining qualification gaps
| Evidence layer | Reported result | Decision-relevant interpretation |
|---|---|---|
| Independent ranges | 25–300 °C; 0–480 kPa; 0.1 °C resolution | Strong laboratory envelope, but not a guaranteed simultaneous rectangle |
| Dynamic response | 53.2 ms loading; 8.8 ms recovery at 100 kPa | Fast enough for many manipulation tasks; application bandwidth still needs system-level validation |
| Mechanical cycling | More than 10,000 cycles at 150 kPa; repeatability error below 0.96% | Useful durability signal, not yet a lifetime or harsh-environment qualification |
| Coupled heat and load | Loading response of 105.3 ms at 100 °C and 113.6 ms at 200 °C; structural decline above 180 °C and 150 kPa | Thermal-mechanical cross-sensitivity and integrity define the true application envelope |
| Array scalability | 8×8 array; 64 cells of 6.5×6.5 mm in a 90×90 mm area | Patterning is plausible, but resistance spread, calibration effort, routing, and yield become product issues |
The array experiment is especially instructive. The authors patterned 64 sensing cells and reported thicknesses from 1.15 to 1.25 mm and sheet resistance from 0.43 to 1.50 kΩ/□. That range may be acceptable for a calibrated research demonstrator, but it also flags a manufacturing problem: a product must convert material variability into predictable channel-to-channel accuracy without excessive calibration cost.
Why the qualification gap matters more than the maximum temperature
High-temperature sensing is not a single market. A rigid ceramic pressure sensor can prioritize stability and bandwidth: a 2025 nitrogen-doped ITO device, for example, reported operation from 25–900 °C, 236 mV/mA/MPa sensitivity at 900 °C, and linearity below 3.74% full scale. A flexible thermocouple on an aerogel blanket has separately demonstrated brief measurement from liquid-nitrogen temperature to 1200 °C, while its stated long-term range was −30–700 °C and sensitivity reached 226.7 µV/°C. These devices solve different problems.
The PAN approach is interesting because it combines conformability, contact pressure, temperature, array patterning, and a gripper demonstration in one stack. Its competitive advantage will not be “highest temperature.” It will be useful spatial and contact intelligence on surfaces where rigid transducers cannot conform—provided that packaging, calibration, attachment, and replacement are engineered together.
That aligns with a broader issue in robotic perception. NIST’s manipulation program is developing test methods for tactile-sensor and end-effector capabilities, including grasp strength, slip resistance, and contact force. A hot-environment sensor will need comparable task-level evidence: not just coupon data, but repeatable grasp performance after thermal cycling, contamination, bending, abrasion, and repair.
Commercialization implications: qualify a stack, not a material
The value chain spans fiber formulation, heat or laser conversion, electrode metallurgy, ceramic adhesive deposition, flexible interconnects, signal conditioning, calibration software, and application-specific packaging. A startup that owns only the PAN recipe may be vulnerable if integrators can substitute another carbonized fiber. A stronger position would combine process control, array calibration, replaceable sensor modules, and an application data set demonstrating stable error under coupled loads.
Battery safety illustrates the distinction. The research paper identifies thermal management as a potential use, but commercial energy-storage safety is evaluated at multiple system levels. UL 9540A examines thermal-runaway behavior at cell, module, unit, and installation levels. A flexible sensor might improve local observation, but it would remain one component inside a larger safety architecture and would need evidence that adhesives, wiring, drift, and failure behavior remain predictable during the relevant event.
For robotics companies, the make-versus-partner decision should hinge on interface ownership. Sensor developers should expose calibrated temperature, pressure, uncertainty, health status, and replacement history—not just raw resistance. Robot OEMs should specify the combined thermal-mechanical envelope, allowable drift, minimum bend radius, cleaning chemistry, cycle life, and fault response. Those requirements turn a promising material into a procurement specification.
What to Watch Over the Next 12 Months
- Coupled-load maps: calibrated error surfaces across temperature, pressure, bend radius, and dwell time, especially beyond the reported 180 °C/150 kPa integrity boundary.
- Batch statistics: cell-to-cell and lot-to-lot resistance, sensitivity, offset, and yield rather than selected-device curves.
- Packaging evidence: thermal cycling, oxidation, humidity, abrasion, cable strain relief, adhesive aging, and cleanability on realistic curved surfaces.
- Array electronics: multiplexing, local compensation, inference latency, and fault isolation for cells that drift or fail.
- Application pilots: measured improvement in grasp stability, inspection coverage, warning time, or maintenance decisions against conventional sensor stacks.
- Standards alignment: test plans tied to robot end-effector metrics, industrial safety cases, or the relevant battery and fire-test framework rather than generic “harsh environment” claims.
What Would Change the Thesis?
The thesis would strengthen if independent testing showed stable calibrated error after substantially longer thermal-mechanical cycling; if array resistance spread narrowed across manufacturing lots; and if a field-relevant pilot demonstrated a measurable operational advantage over rigid sensors or separate temperature and force components.
The thesis would weaken if thermal compensation erased pressure accuracy across the intended coupled-load range, if ceramic-adhesive or electrode interfaces dominated failures, if calibration cost scaled with every sensing cell, or if competing rigid or noncontact sensors delivered equivalent decision value with simpler qualification. Evidence that the gripper integration cannot tolerate abrasion, replacement, or wiring strain would also shift the technology from a product platform to a narrower disposable or research role.
Takeaway
The new PAN device advances high-temperature flexible sensing by joining two useful modalities, scalable patterning, and robotic integration. It also reveals the diligence question that should govern commercialization: what is the validated simultaneous envelope of the complete system? Teams that treat qualification data, calibration, packaging, and replacement as core product IP—not downstream engineering—will be best positioned to turn this research direction into dependable industrial perception.
This analysis extends MEMSWork’s earlier work on technology-development questions for soft-robotic sensors, integrated sensing for flexible automation, and the transfer test behind low-cost tactile skins.
Sources and Methodology
- Liu et al., “An integrated polyacrylonitrile fiber mat sensor for simultaneous temperature and pressure monitoring in high-temperature environments,” Microsystems & Nanoengineering (2026).
- Liu et al., “Flexible temperature sensor with high sensitivity ranging from liquid nitrogen temperature to 1200 °C,” International Journal of Extreme Manufacturing (2023).
- Wang et al., “A ceramic-based ITON piezoresistive pressure sensor with wide temperature range,” Microsystems & Nanoengineering (2025).
- NIST, Grasping, Manipulation, and Contact Safety Performance of Robotic Systems.
- UL Solutions, UL 9540A Test Method for Battery Energy Storage Systems.
Last updated: October 16, 2026. Original publication.