CONTEXT

Information storage and more specifically non-volatile memories represent strategic issues in the field of microelectronics. The huge non-volatile memory market is currently led by the Flash technology (Flash cards, Solid State Drives). However, the intrinsic drawbacks of this technology (long writing time, 5-10 μs, use of high voltages, 5-10 V and complex design) will limit its development in the near future. Several alternatives, such as the Phase Change (PCRAM), Magnetic (MRAM) or Resistive (RERAM) Random Access Memories, are currently considered to overcome the main limitations of the Flash technology. In particular, RRAM’s, based on an Electric-Pulse-Induced Resistive Switching (EPIRS) effect, seem to be an appealing solution as they offer a better scalability, shorter writing times (10-100 ns) and a good endurance.

TECHNICAL DESCRIPTION

The patent proposed here describes the use of a new family of compound for RERAM applications. Non-volatile reversible electric-pulse-induced resistive switching were indeed recently uncovered on AM4X8 (A = Ga, Ge ; M = V, Nb, Mo, Ta ; X = S, Se) single crystals at the Institut des Matériaux Jean Rouxel (IMN). In this chalcogenide compound, short low level electric pulses (100 ns, 500 V.cm-1) applied on a simple MIM device (Metal / AM4X8 /Metal) yield a non-volatile resistive switching between a high and a low resistance state. These resistance states can constitute 0 and 1 states of a binary memory bit, as a reversible switching between both states can be achieved with another electric pulse. Writing / erasing cycles are observed at room temperature with a retention time of several months for some compositions. Moreover, these main memory characteristics may be easily tuned through a proper choice of composition within the large number of compounds and solid solutions attainable in the AM4X8 family. The resistive switching observed in the AM4X8 system corresponds neither to a phase change nor to any of the phenomena (thermal, electronic injection or ionic diffusion) proposed so far to explain the resistive switching effect in materials envisioned for RRAM applications. Our initial work suggests that it could correspond to a new mechanism involving a Mott insulator to metal transition at the nanoscale, which could enable a high scalability down to 10 nm.

BENEFIT

– short switching time, less than 100 ns
– potential low-power consumption
– potential scalability if deposited in thin layer (see patent 09-R15981-FR, 100901)
– long retention time (tested for more than several months for some compositions)

For further information, please contact us (Ref 00946-01)