Background
Temperature sensing plays a pivotal role in controlling and monitoring industrial, chemical, and biomedical systems. There are some complex internal space structures such as porous and zigzag in microscale or constrained environments. Traditional temperature measurement techniques such as thermocouples, infrared sensors, and fiber optics face larger challenges in such specific environments. This is not only due to their size but also because of limitations in mechanical flexibility, spatial adaptability, or contact requirements restrictions. To overcome these challenges, researchers have explored various nanoscale thermometers, including quantum dots, lanthanide-doped nanoparticles, and nitrogen-vacancy (NV) centers in diamond. These temperature measurement methods offer high spatial resolution and a broad temperature measurement range, but they typically require continuous light excitation and real-time fluorescence detection. It makes them difficult to implement effectively in complex porous structures or confined internal spaces.

Research Progress
Xiaocong Chang and colleagues at Harbin Institute of Technology proposed a temperature-responsive microrobot (TRM). It can move within narrow and tortuous high-temperature regions, sense temperature, and carry the sensed thermal information out of complex porous environments. The TRM enables offline temperature measurement in the range of 160-240℃, providing a new approach for high-temperature detection in confined environments, and the team presented a conceptual illustration (Fig. 1).

The TRM body is composed of thermochromic functional materials, mainly consisting of transparent resin and tetraamminecopper(II) sulfate. The team analyzed its thermochromic mechanism: upon heating, tetraamminecopper(II) sulfate gradually transforms into basic copper sulfate, with the color changing from blue to green, and the transformation is nearly complete at around 200 ℃. As the temperature rises further, the transparent resin undergoes thermal oxidation to generate conjugated chromophores, causing the overall color to shift toward deep yellow. This color change is irreversible. Based on droplet microfluidics and physical vapor deposition (PVD), the team fabricated Janus-structured TRMs: one side consists of thermochromic material with tetraamminecopper(II) sulfate as the core, while the other side is coated with a nickel layer of about 300 nm in thickness via PVD. This combination allows the TRM to both sense temperature and achieve controllable motion.

The team further investigated the locomotion ability of TRMs by constructing a three-dimensional Helmholtz coil system capable of generating a tunable rotating magnetic field. They explored the effects of magnetic field strength and frequency on TRM motion, which helps optimize driving parameters in different media and geometries. They also validated the motion capability and controllable maneuverability of TRMs in complex microscale structures.

Next, the team extracted color features from microscopic images taken at different temperatures and trained a multilayer perceptron (MLP) regression model to map color to temperature. The trained model achieved an R² of 0.979 on the training set and 0.946 on the test set, with residuals approximately normally distributed, demonstrating excellent generalization accuracy and stability. Thus, once TRMs are retrieved, quantitative temperature values can be obtained directly from microscopic images, without the need for onsite power supply or real-time optical access.

Finally, the team validated the temperature-sensing capability of TRMs in quasi-realistic scenarios. In a simplified model, TRMs were guided by a magnetic field into a 200 ℃ target region, sensed the temperature, and were retrieved. The inferred temperature from the sensing model was 198.73 ℃, which closely matched the preset value. Subsequently, in experiments with a more representative non-transparent porous silicon carbide structure, TRMs moved from the entrance of the porous structure to the high-temperature bottom region, completed temperature sensing, and were retrieved. The sensing model inferred the bottom temperature to be 199.76 ℃. These experiments in quasi-realistic scenarios verified the effectiveness of TRMs in non-transparent, confined, high-temperature environments.

Future Prospects
In future, temperature-responsive microrobots offer a novel technological approach for offline detection in high-temperature, confined, and non-transparent environments: magnetic actuation, irreversible color recording, and quantitative decoding. They can be adapted to porous materials, microchannels, and complex internal structures. With the expansion of material systems, the maturation of scalable fabrication technologies, and the engineering of magnetic field equipment, this method holds promise for achieving safer and more efficient industrial-level temperature sensing.

The complete study is accessible via DOI: 10.34133/research.0760