Optical investigation of non-thermal processes in phase change materials

Lead Research Organisation: UNIVERSITY OF EXETER
Department Name: Physics


Storage of information on optical disks and within random access memory (RAM) chips is central to both present and future information technology. Future storage devices must have larger capacity, shorter access times, smaller physical size and use low power. Existing optical storage formats (e.g DVD-R/W) are based upon a process of structural change within the coating on the disk. By heating the surface of the disk with focused laser pulses, a chalcogenide alloy is driven between amorphous and crystalline phases. Although the technology is well established, our understanding of the structural phase transition is surprisingly limited. A recent study of Ge2 Sb2 Te5 (GST), the most commonly used material, revealed that the material does not necessarily melt in a conventional sense, and suggested that optical excitation of specific electronic states drives a non-thermal phase transition. Electrically addressed phase-change RAM (PCRAM or Ovonic memory) has the potential to replace existing FLASH-based memory, which faces difficulties in scaling to smaller sizes (higher capacities). In contrast PCRAM scales well and is inherently bistable. PCRAM cells rely on a reversible transition between the amorphous and crystalline phases to 'write' data. The transition is accompanied by a dramatic change in electrical resistance that may be easily read out. When a write pulse is supplied to the amorphous material, threshold switching to a low conductivity state is observed before the structural phase change, or memory switching , occurs. The origin of the threshold switching mechanism remains controversial and the intrinsic timescales for threshold switching have, as far as we are aware, never been measured.We propose to use time-resolved femtosecond optical measurement techniques to investigate the phase transition in GST and other chalcogenide alloys. Using conventional optical pump-probe measurements we will determine which phase the material occupies at different times during the switching process and investigate its dependence upon the duration and wavelength of the exciting optical pulse. We will hence understand whether a tailored optical pulse may induce more efficient writing and erasure of an optical disk. Highly optimized sample materials will be supplied by RWTH-Aachen, Plasmon Data Systems Ltd, and ST Microelectronics. We will examine the response of the alloy to the polarization of the optical pulse. The observation of optically induced birefringence may provide a better understanding of the non-thermal nature of the transition and lead to additional applications in optical communications technology. Stroboscopic time resolved optical measurements will also be performed upon prototype PCRAM cells to understand the dynamics of the threshold and memory switching processes in real devices. The ultrafast laser will be synchronised to fast electrical pulse generators that deliver set and reset pulses to the cell, allowing us to determine the instantaneous electronic state of the chalcogenide alloy during both the threshold and memory switching processes. We will also use a short but intense laser pulse to assist the switching processes. By varying the wavelength of the pulse we will investigate the importance of specific electronic transitions in the non-thermal breaking of bonds. Optically-assisted electronic switching of PCRAM devices has, to our knowledge, not been previously attempted. The measurements will be understood, interpreted and guided by multi-scale modelling. Ab-initio Density Functional Theory (DFT) calculations will be performed to determine the crystallographic strucutre, optical properties and electronic band stucture of the material. Physically realistic macroscopic models for predicting the performance of real device (electrical and optical memories) will be developed by 'bridging the gap' between ab-initio atomic scale modelling and existing phenomenological crystallisation models.


10 25 50