Current Research
- Rational design of new phase change materials, and improvement of existing ones.
- Using computer simulations to explore novel compositions along the GeTe-SbTe pseudo tie line for optimised device properties.
- Developing Gaussian Approximation Potentials (GAP) for GST.
- Modelling the response of phase change materials to radiation.
Annealing is a slow process for most compounds, however, for some it can occur very rapidly, on sub-nanosecond timescales. This extremely fast phase transition can be exploited, together with differences in physical properties between amorphous and crystalline phases. Similarly, some phase change materials show a large difference in electrical resistivity between their amorphous and crystalline states, and therefore application of suitable melting, annealing and reading currents to small regions of the material allow them to be used for electronic storage.
Phase change materials have already found commercial application in the recording layers of rewritable optical discs. A short, high-power pulse of laser light is applied to a microscopic region of the disc surface and causes local melting of the recording layer. On removal of the pulse the region cools very rapidly, leaving an amorphous mark in a crystalline matrix. Application of a lower-power laser pulse to an amorphous region then induces annealing, rendering the region crystalline. The phase change materials used are chosen to have a high optical contrast, and a reflective layer placed behind the recording one, so that passing a low-power laser over the disc surface and observing the intensity of reflected light allows the state of a region, and hence the value of the bit it represents, to be read. In addition, memories based on phase change alloys can in theory be faster than existing technologies, leading to high performance, low latency solutions for both general data storage (e.g. solid state disks, flash drives, memory cards) and performance applications (e.g. system and video RAM). The phase change memories are non-volatile and therefore, in contrast to current high-performance memories, will consume substantially less power both when active and when left in standby.
Ab initio molecular dynamics simulations provide detailed information about the system at every timestep, and therefore can be used to probe the microscopic changes which occur during the phase transitions in such materials. It has been shown that computer simulation of small models of Ge-Sb-Te (GST) alloys during various melting, quenching and annealing regimes accurately reproduces observed experimental behaviour.