As these martensitic transformations are diffusionless and reversible, they are a priori fast, but the fundamental speed limits remain unexplored.
From a fundamental point of view, it is therefore of high interest to explore the temporal limits of a martensitic transformation.
Among structural transformations, martensitic transformations are remarkable because they exhibit very large lattice distortions.
With lattice distortions that are about two orders of magnitude larger, martensitic transformations stand out among structural transformations.
Already at the ms time scale, the martensitic microstructure formed after fast cooling differs from a slowly cooled sample21.
Structural transformations in shape memory alloys enable a broad range of emerging applications, including high force, high stroke actuation1,2, sensing3, mechanocaloric refrigeration4,5, and waste heat harvesting6,7. The crystal structure in these alloys changes between high-symmetric austenite (A) at high temperatures and low-symmetric martensite (M) at low temperatures. As these martensitic transformations are diffusionless and reversible, they are a priori fast, but the fundamental speed limits remain unexplored. Most of the above-mentioned applications benefit from a fast cycle since their power is determined by the energy per cycle multiplied by the frequency. From a fundamental point of view, it is therefore of high interest to explore the temporal limits of a martensitic transformation.
The question of fundamental speed limits is relevant to many functional materials with various types of transformations, including magnetic8,9,10, ferroelectric11,12, glassy-crystalline phase change13, metal-insulator14,15, and coupled magnetostructural transformations. Among these transformations, structural ones are slowest since they require the movement of atoms, which have a much higher inertial mass compared to electrons. Among structural transformations, martensitic transformations are remarkable because they exhibit very large lattice distortions. For example, in the Ni-Mn-Ga system examined here, the lattice of the low temperature phase distorts by about 24% with respect to the high temperature phase. This is much more than the 0.3% distortion at the isostructural transformation in FeRh – one of the first systems used for time dependence experiments16, which is still of current interest17,18. With lattice distortions that are about two orders of magnitude larger, martensitic transformations stand out among structural transformations. In particular, the large distortion requires the collective movement of many atoms19, which in turn affects the martensitic microstructure up to the macroscale20. The joint movement of many atoms implies that martensitic transformations should be slower than other diffusionless structural transformations with smaller distortions, but time-resolved experiments on this class of functional materials are sparse.
The few key findings on the time dependence of martensitic transformations are best sorted by decreasing time scale. This also allows to connect the more applied aspects with fundamental ones. Already at the ms time scale, the martensitic microstructure formed after fast cooling differs from a slowly cooled sample21. As this microstructure decides on most functional properties, this is expected to have an impact on most applications. These experiments reveal that quite many atoms are collectively involved in this transformation. During fast cooling, they do not have enough time to find their equilibrium position inside this otherwise quite well-ordered hierarchical martensitic microstructure. To probe the μs time scale, Shilo’s group developed a setup22, which is quite similar to a typical actuator application. By Joule heating NiTi thin wires and tracking the transient electrical response, they confirmed a transformation time of approximately 20 μs. These experiments show that martensitic transformations can be described by the generalized framework of moving interfaces through pinning sites, where a thermally controlled transformation from creeping to depinning and flow occurs. Their additional measurements by time-resolved X-ray diffraction revealed that the transformation at the surface can be mostly completed within 1 μs23.
To probe nanosecond dynamics, we heated Ni-Mn-Ga24 and NiTi25 films with a 7 ns laser. Synchrotron-based time-resolved X-ray diffraction revealed a complete transformation from martensite to austenite within the duration of the laser pulse. The full transformation cycle was completed within 200 ns, owing to the fast cooling of the film due to its high surface-to-volume ratio25. To examine even shorter time scales, Mariager et al.26 used a 120 fs laser pulse to heat a Ni–Mn–Ga film and reported a transformation from martensite to austenite within 200 ps measured by time-resolved synchrotron X-ray diffraction. They also detected coherent phonons in the form of intensity oscillations of Bragg reflections from the modulated crystal structure, appearing 300 fs after laser heating—close to the 270 fs time resolution of their setup. With a different setup having a time resolution of 200 fs26, they also measured a demagnetization time of 320 fs, illustrating that the spin reacts much faster than the martensitic transformation completes in this material. This difference is expected from the three-temperature model, which sketches the initial timescales when heating a metallic magnet by an ultrashort laser pulse8. The laser first heats the electronic system, and the spin system usually follows within picoseconds. Finally, the lattice temperature also increases, which can take several picoseconds.
Lantz et al. used a similar combination of magnetic and structural probes to measure the time dependence of the transition from a modulated premartensite to austenite27. Changes occur within picoseconds for both, magnetism and structural modulations, which indicates a coupling between magnetism and charge density waves, which are often considered the origin of structural modulations. However, these experiments examined premartensite in Ni 2 MnGa, which occurs at temperatures slightly above the first-order martensitic transformation28. Premartensite exhibits a much lower distortion of just 0.7%29 compared to the 24% in the transition to non-modulated martensite (NM). Accordingly, the low strain in premartensite is already accommodated within several nanometers, resulting in a characteristic tweed nanostructure28. This nanoscaled microstructure therefore requires the collective motion of much less atoms compared to a common martensitic microstructure, which ranges up to the macroscale. Thus the observation that premartensitic transformation can occur already within ps27 supports the argument that the lattice distortion has a strong influence on the transformation time. An aspect that was neglected in most previous studies, is that martensitic transformations are driven by temperature, and therefore it is important to quantify the laser-induced temperature rise in such experiments in order to understand similarities and differences to experiments at longer time scales.
