CAMELS (ChAlcogenides MEmories for multiLevel Storage)

EU Research FP6
DEIB Role: Partner
Start date: 2005-09-01
Length: 36 months
Project abstract
Chalcogenide-based Phase Change Memories are considered the best candidates to replace Flash memories, beyond the 45nm node.
However, in order to maintain the present growth rate in terms of storage densities, and to support the emerging, memory-intensive applications, ways must be found to further increase the storage capability of Phase Change Memories, introducing multilevel storage (two or more bits/cell).
The final targets of the project include:
The project started in July 2005. The research activity currently in course includes both the experimental characterization and the numerical modeling through device simulation codes of the PCM cell, in order to evaluate the capability of multi-bit storage.
However, in order to maintain the present growth rate in terms of storage densities, and to support the emerging, memory-intensive applications, ways must be found to further increase the storage capability of Phase Change Memories, introducing multilevel storage (two or more bits/cell).
The final targets of the project include:
- a demonstration of multilevel programmability on a Multimegabit test device;
- a demonstration of the multilevel performances of the optimised material developed in the project.
The project started in July 2005. The research activity currently in course includes both the experimental characterization and the numerical modeling through device simulation codes of the PCM cell, in order to evaluate the capability of multi-bit storage.
Project results
The experimental activity has been focused on the reliability characterization of memory cells with the standard material (Ge2Sb2Te5), in order to clarify the retention properties of programmed levels as a function on the initial state. The purpose is to identify the position and programming method of multiple levels for maximizing the multilevel cell reliability. Results indicate that reliability is limited by two physical mechanisms with opposite behavior: drift, that is the structural relaxation of the material in its amorphous phase, which leads to an increase of resistance, and crystallization, which leads to a decrease of resistance.
The modeling activity has investigated cell optimization in terms of geometry and material, in order to minimize the programming current, hence the overall size of the cell. The optimum geometry for minimum programming current has been evaluated at fixed set-state resistance. The value of the current has been analyzed by a sensitivity study as a function of the material properties, in terms of electrical and thermal conductivity.
The modeling activity has investigated cell optimization in terms of geometry and material, in order to minimize the programming current, hence the overall size of the cell. The optimum geometry for minimum programming current has been evaluated at fixed set-state resistance. The value of the current has been analyzed by a sensitivity study as a function of the material properties, in terms of electrical and thermal conductivity.