Some metals can change their internal structure almost instantly, and that motion can power tiny machines. This study finds a complete switching cycle in just five nanoseconds. These structural changes already drive devices that move, cool through motion, and recover waste heat.

Faster changes would benefit all of them, but the fastest motion and its basic limits have rarely been examined. The study finds that one change finishes in about ninety picoseconds, and an almost complete back-and-forth cycle finishes within five nanoseconds.

The change may be even faster because the measurement itself sets the upper limit. The results also point beyond heat alone: stress matters, and many atoms must move together because the internal structure changes so much.

The central idea is like a crowded dance floor changing formation: the people cannot all move independently when the shape must change. Many have to shift together, and that shared movement can set the pace. To watch this happen, the study heats a thin film with an extremely brief laser pulse and checks its crystal structure with X-ray pulses.

By changing the delay, it follows the transformation as it unfolds. The measurements also connect each heating level with the film's changing temperature, making it possible to ask when ordinary temperature is enough to describe the change and when it is not. The crystal structure is tracked directly and without damaging the film, revealing how the change develops over time.

The scan follows the complete journey from one structure to the other and back again, rather than looking at only one moment. The forward change from martensite to austenite is complete in less than two hundred picoseconds across all tested heating levels. For the return from austenite to martensite, higher heating makes the transformation take much longer, beyond the available delay range.

The reason is straightforward: stronger heating leaves the film hotter, so more heat must escape before the original structure can return. Here is the surprise: the calculated film temperature stays below the temperature needed to begin the new structure, yet the complete change is still observed at the lowest heating level.

That contradiction means the film cannot be described by one temperature alone. Its electronic, magnetic, and atomic vibrations must be considered separately. The plotted points form two distinct bands: one near the fastest measurable limit, and another several steps higher, with only a slight upward tilt in each.

The smallest-scale simulations identify the main speed limit: the large distortion makes the change collective, involving many atoms, so it is slower than changes involving smaller distortions and fewer atoms. The larger structure of the material and stress must also be considered, because the local atomic movement does not tell the whole story.

The almost complete cycle takes five nanoseconds, far shorter than the two hundred nanoseconds reached in a recent study. That is already much shorter than cycle times in current shape-memory applications, so present speed limits should not soon prevent faster devices.

The result points toward miniaturized high-speed systems. By tracking the structural change, stress, and atomic temperatures at the same time, the study shows where familiar heat-based explanations still work and where they break down. For future experiments, the study proposes continuing this multi-observable approach at even shorter time scales, bridging materials science and physics in understanding martensitic transformations.

The material can switch extraordinarily fast, but its speed is limited by many atoms having to move together, while heat and built-up stress compete to control the change. That matters for faster miniature devices.