Significance Light-induced phase transition is a key process in material processing and property modification using an ultrafast laser. Phase transitions driven by thermal effects, such as melting and evaporation, disorder a material. As such, control of an atomic structure using a laser is still not good enough and limits the processing precision of the laser. In contrast, nonthermal effects of an ultrafast laser show great potential in the high-precision control of phase transitions. However, owing to the complex light-matter interaction processes, the mechanisms behind the transitions still need to be clarified.

In recent years, many experimental studies of ultrafast laser-induced nonthermal phase transitions and their related mechanisms are reported. Several mechanisms, especially the atomic mechanisms, conflict with each other, which hinders the control and application of nonthermal phase transitions. Therefore, it is necessary to summarize the previous results to extract the key points and guide the development of ultrafast laser-induced nonthermal phase transitions.

Progress Ultrafast laser-induced nonthermal melting of Si had been proposed as early as the 1970s. A plasma annealing model in which the chemical bonding was softened by electronic excitation from bonding states to antibonding states was proposed to explain the abovementioned phenomenon. Then, the model was further improved using a tight-bonding model to quantitatively calculate the excitation induced instability. However, limited by the detection technology, nonthermal melting was not experimentally confirmed until 2001, when Rousse et al. demonstrated the ultrafast amorphization of InSb by time-resolved X-ray diffraction. In recent years, ultrafast laser-induced phase transitions in the phase-change memory (PCM) technology have attracted considerable attention owing to their interesting physics and promising applications in memory and computing technologies. For a long time, the mechanism of the ultrafast laser-induced amorphization of PCM materials was attributed to the thermal melting effect. In 2011, first-principles calculations proposed by Li et al. suggested that the electronic excitation in the PCM material Ge₂Sb₂Te₅ could induce solid-to-solid amorphization without thermal melting (Fig. 3). Then, Chen et al. further explored the key factors and rules of the electronic-excitation-induced amorphization, including global stress and local atomic forces. With the development of experimental technologies, more evidences of ultrafast laser-induced nonthermal phase transitions have been found. For example, Mitrofanov et al. had demonstrated the ultrafast laser-induced instability of the long-range order in Ge₂Sb₂Te₅ by time-resolved X-ray diffraction and X-ray absorption fine structure spectroscopy. Fons et al. observed the ultrafast laser-induced unexpected large expansion of Ge₂Sb₂Te₅ by time-resolved X-ray diffraction, which cannot be explained using thermal effects. Recently, Tanimura et al. demonstrated that thermal equilibrium in femtosecond laser irradiated PbTe can only be established after 12 ps.

Although ultrafast laser can induce non-thermal phase transitions, the final results of the phase transitions are disordered materials, which are similar to the results of thermal melting, and the results limit new applications of non-thermal phase transitions. In 2015, Hu et al. reported the femtosecond laser-induced rhombohedral-to-cubic (order-to-order) phase transition of GeTe by time-resolved electron diffraction. In 2016, Matsubara et al. reported the transition by time-resolved X-ray diffraction. They attributed the phenomenon to the rattling motion of Ge atoms rather than the real rhombohedral-to-cubic phase transition. In addition, Kolobov et al. proposed that the excitation can lead to the random distribution of long and short bonds in GeTe, where the average effect leads to the symmetry of the cubic phase. These conflicting mechanisms are debated because a real-time atomic picture of the phase transition is lacking. In 2018, Chen et al. confirmed the real rhombohedral-to-cubic phase transition of GeTe using the time-dependent density functional theory (Fig. 5). The atomic mechanism is due to the directional driving forces induced by the change of potential energy surface upon excitation. One problem is how to distinguish thermal and nonthermal phase transitions. Since the time for thermal equilibrium is of the order of picoseconds, a possible distinguishing factor is the time of phase transition. It is reasonable to believe that sub-picosecond phase transition should be nonthermal. Another problem is how to find more materials that can have order-to-order phase transitions. According to the mechanism proposed by Chen et al., the special change of potential energy surface upon excitation is the key factor for such transitions. Therefore, theoretical prediction using first-principles calculations and high-throughput screening should be a good choice in solving the abovementioned problems.

Conclusion and Prospect Compared with thermally induced phase transitions (such as melting), nonthermal phase transitions have several advantages, such as speed, energy consumption, and controllability. Especially for order-to-order phase transitions, structures of materials can be controlled at the atomic scale. Therefore, the understanding of the atomic mechanism of nonthermal phase transitions is important in the micro-nano fabrication of materials using an ultrafast laser. Nonthermal phase transitions are also applicable in memory/computing technologies with ultrafast speed and ultralow power consumption. However, further investigations are still needed to understand the atomic mechanisms of transitions under different conditions to better control them and design systems, therefore realizing specific phase transitions.