RESET CURRENT AND RESET TIME SCALING IN AN ALTERNATIVE PHASE CHANGE MEMORY (PCM) DEVICE

Abstract

The key issues in Non-volatile storage medium are long term data retention, fast speed and scalability. As Non-volatile Flash memories approach their physical limitation for scaling and reliability, several alternative memory concepts are being proposed and investigated based on new materials. The phase change memory (PCM), based on the reversible structural phase transition in Chalcogenide material is one of the most promising memory technology, gaining increasing importance due to long term data retention, fast programming speed and easier scalability among innovative non-volatile memories. After the short description of the PCM cell operation, in the present work, 3D - Alternating 1)irection Implicit (ADI) thermal modeling scheme has been adopted for faster and accurate RESET current (IRESET) and RESET time (tRESEF) predictions of an alternative Phase Change Memory (PCM) device. For an incremental time step At, an implicit method is used for 3-D ADI scheme for spatial discretization of material co-ordinates, alongside appropriate transient thermal boundary conditions. As such, thermal boundary conditions are derived through energy balance approach across the interfaces of dissimilar materials. Direction implicit discretization method thus allows tn-diagonal matrix symmetry in 3D - ADI solver, and permits linear run time, unlike the normal Crank-Nicolson (CN) scheme for numerical simulation of second-order parabolic heat diffusion equation. Through well calibrated 31)- ADl modeling, it has been demonstrated that both RESET current and RESET time in conventional PCM cell can be dramatically scaled to nano-Ampere and femto-second, respectively through incorporating an additional ultra-thin Silicon Nitride (Si3N4) layer as a thermal concentrator, inserting between the top semiconducting GST and the bottom electrode contacts. Furthermore, our 3D simulations successfully envisage that pico-second heat relaxation is achievable with conductive Copper bottom electrode, with reset pulse being withdrawn. In the second part of the work, proposed PCM device having Silicon Nitride thermal concentrator and conductive copper bottom electrode will be optimized for improved RESET performance through results obtained from the 3D-ADI modeling, and further calibration of ADI data with COMSOL multi-physics modeling framework. It has been shown that, scaling of nitride heater thickness increases RESET drive current at the expense of heat confinement. Spatial heat gradient inside Nitride (dT/dx) and the ratio of Nitride peak to GST interfacial temperature (TpIK/TTNy) are found to rise slowly for thin Nitride as when compared to the thick Nitride device when the RESET bias is held constant. Through ADI III numerical simulations we furthermore confirm that non-linear resistive heating inside Nitride ensures RESET time to scale for thinner Nitride, instead. Furthermore, on the other hand for isotropic cell area increase it is ascertained from the ADI simulations that RESET current and RESET time both increase monotonically with increase in cell area, and so as the reverse heat relaxation time followed by cell RESET, which thereby impedes speed of PCM. In this regard, we have found fast heat relaxation is attainable with proper selection of conductive bottom electrode contact material in proposed PCM cell. Finally, we propose an in-plane read operation for said PCM device to maintain sufficient read current window between the RESET and SET phases and demonstrate its working through proper adjustment of RESET pulse width and GST-thickness up to a critical limit. Hereinafter, different PCM device layouts will be outlined for possible reduction in active cell area.