Crystal-Amorphous Transformation Via Defect-Templating in Phase-Change Materials

Phase-change materials (PCM) such as GeTe and Ge-Sb-Te alloys are potential candidates for non-volatile memory applications, because they can reversibly and rapidly transform between a crystalline phase and an amorphous phase with medium-range order. Traditionally, crystal-amorphous transformation in these materials has been carried out via melt-quench pathway, where the crystalline phase is heated beyond its melting point by the rising edge of an electric pulse, and the melt phase is quenched by the falling edge into a glassy phase. Formation of an intermediate melt phase in this transformation pathway requires usage of large switching current densities, resulting in energy wastage, and device degradation issues. Furthermore, melt-quench pathway is a brute force strategy of amorphizing PCM, and does not utilize the peculiar structural properties in crystalline phase. It will be beneficial from a device perspective that crystal-amorphous transformation is carried out via subtler solid-state pathways. Single-crystalline nanowire phase-change memory, owing to its lateral geometry and large volumes of active material, offers a platform to construct a crystal-amorphous transformation pathway via gradually increasing disorder in the crystalline phase, and study it. Using in situ transmission electron microscopy on GeTe and Ge2Sb2Te5 systems, we showed that the application of an electric pulse (heat-shock) creates dislocations in the PCM that migrate with the hole-wind force, and interact with the already existing ferroelectric boundaries in case of GeTe, changing their nature. We adapted novel tools such as optical second harmonic generation polarimety to carefully study these defect interactions. These defects accumulate at a region of local inhomogeneity, and upon addition of defects beyond a critical limit to that region via electrical pulsing, an amorphous phase "nucleates". We also studied the effect of defect dynamics on carrier transport using temperature dependent transport measurements in GeTe, which transforms from a metal to a weakly localized metal to finally an Andersons insulator, upon defect accumulation, prior to amorphization. Taking lessons from these fundamental studies, we defect-engineered GeTe into insulating crystalline states as the starting crystalline states, and demonstrated orders of magnitude drop in the power densities required for switching, compared with those required for melt-quench pathway.