A new material points toward programmable heat

Heat usually obeys a frustrating symmetry. A surface that absorbs thermal radiation well from a certain direction or wavelength tends to emit it in the same way. For engineers who would like to steer waste heat, tune infrared sensors, or build optical memory, that reciprocity is a boundary: the material is useful, but not freely programmable.

A team led by Koichi Okamoto and Shunsuke Murai at Osaka Metropolitan University has now reported a laboratory design that pushes against that boundary. In Laser & Photonics Reviews, Ye Ming Qing, Yi Shen, Jun Wu, Murai, Zhaogang Dong, and Okamoto describe a metagrating that combines a magneto-optical indium arsenide waveguide with ridges of the phase-change material GST. ScienceDaily summarized the result on July 7, 2026: the device can make heat radiation behave differently depending on direction, switch between states, and remember a setting after power is removed.

The important word is radiation. This is not a pipe carrying hot water or a fan moving warm air. It concerns infrared light, the electromagnetic glow that every warm object emits. If a structure can absorb incoming infrared energy from one side while preferentially emitting or blocking it in another direction, it begins to act less like a passive coating and more like a tiny thermal circuit.


The mechanism has two parts. Magneto-optical materials change their interaction with light in a magnetic field, breaking the usual left-right sameness of the optical path. GST, a germanium-antimony-tellurium phase-change material already familiar from photonic-memory research, can switch between structural states with different optical properties. The paper calls the combined structure a dynamic, non-volatile, nonreciprocal absorber: dynamic because it can be tuned, nonreciprocal because the direction matters, and non-volatile because the phase-change layer can keep its state without continuous electrical power.

A useful detail is the near-normal angle. Earlier approaches to nonreciprocal thermal radiation often needed light to arrive at extreme grazing angles. That geometry throws away performance, because the effective area seen by the incoming light becomes small. The new metagrating was designed for strong contrast much closer to straight-on incidence, which is why the authors see it as a step toward practical emitters, sensors, and thermal-photonic components rather than only a physics demonstration.

The memory aspect is just as interesting. In ordinary electronics, a switch that forgets its setting when power disappears has limited value. The GST layer gives the thermal device a latch: once a state is written, the structure can preserve it. That does not mean a consumer gadget is imminent, but it explains the computer-chip analogy in the university release. Information is not being stored as ordinary electric charge; the setting is held in an optical and thermal material state.


There are clear limits. The work is a designed device and research demonstration, not a household heat controller. It uses specialized materials, a magnetic-field configuration, and nanostructured geometry that would have to be fabricated reliably before real products could appear. The current claim is also about controlling radiative heat transfer, not magically eliminating heat or replacing insulation, heat pumps, or cooling systems.

Still, the direction is hopeful because the need is real. Data centers, satellites, infrared cameras, compact sensors, and energy-conversion devices all struggle with heat that must be moved, measured, or hidden from the wrong channel. Better thermal control often saves energy indirectly: a sensor that sees more cleanly needs less correction, and a device that routes heat more precisely may need less brute-force cooling.

The study also shows how materials science is becoming more like programming. Instead of accepting a surface as fixed, researchers are designing structures whose geometry, magnetic response, and phase state create a menu of behaviors. Heat remains physical and stubborn, but it is becoming less anonymous. In the best case, future engineers may not merely ask how much heat a device makes. They may ask where that heat should go, when it should be visible, and which state the material should remember next.