Technology and AI

A cooling prototype ran on waste heat with no electric actuator at all

By Staff Writer  |  31 August 2026

Two ultra thin metal films, one that shrinks when heated and one that cools when the load on it is released, have produced measurable cold with no electric motor driving them. The work is a feasibility study rather than a product, and the team says so.

The result was published on Friday 28 August by researchers at the Karlsruhe Institute of Technology and the University of Tsukuba. The paper is titled Heat-driven elastocaloric cooling with shape memory films, and carries the digital object identifier 10.1038/s41560-026-02122-6.

Some background is needed to see why it is interesting. Almost every cooling machine in the world works the same way it did a century ago: an electrically driven compressor moves heat from one place to another using a refrigerant, and a great many of those refrigerants are themselves greenhouse gases. Solid state cooling is the alternative that has been circling the problem for years. Certain metal alloys cool down when a mechanical load that has been applied to them is released, which is a way of making cold without any gas at all. The catch has always been the actuator: something has to apply and release that load, and until now that something has been an electric motor.

The heat does the work the motor used to do

The team coupled two nickel titanium films with opposite jobs. The first exploits the shape memory effect, so when it is heated it contracts and turns thermal energy directly into mechanical work. That movement passes straight into the second film, where repeated loading and unloading changes the crystal structure and produces cold. Heat has replaced the motor.

The crucial innovation is that we combine two complementary functions of shape memory alloys, with one film converting heat into mechanical work and the other film converting this work into cold. This way, we're establishing a new approach to drive solid-state cooling, thereby opening up exciting possibilities for the use of waste heat and solar energy.

Dr Jingyuan Xu, head of the Young Investigator Group of the ZEco Thermal Lab at the Institute of Microstructure Technology, Karlsruhe Institute of Technology

The numbers are small and are presented as such. At an actuator temperature of 86 degrees Celsius the prototype produced a temperature difference of 4 degrees at component level, with a change of nearly 13 degrees in the refrigerant film itself. It also ran reliably when fed by an external source at 130 degrees, which is the test that matters for the claim that real waste heat will do.

The decisive moment for us was when we were able to measure the cold that had indeed been generated by a heat-driven system. This showed us that the principle doesn't just work in theory.

Yi-Ting Hsiau, lead author and doctoral researcher at the Institute of Microstructure Technology, Karlsruhe Institute of Technology

What it is not, yet

Four degrees at component level will not cool anything anybody owns. The claim being made is that the principle works, and the team states in terms that the setup was designed as a feasibility study and is not optimised for cooling capacity.

The next step is connecting films in parallel to raise that capacity. The applications the institute names are worth noting because they are narrow and physical rather than sweeping: cooling processors in computers using the waste heat those processors themselves produce, and cooling electronics in vehicles using heat from the drive train. The institute puts cooling and heating together at almost half of global energy consumption, which is the reason anybody is funding this line of work at all.

For those planning schemes where the cooling load has become a planning question rather than a mechanical one, none of this changes a design due next year. It is worth a note in the file for a different reason. Every scheme of this kind currently treats heat as the output to be got rid of. A machine that treats it as the input is a different arithmetic, and somebody is going to have to price it eventually.