Research Topics

In order to process large amount of data with fast speed, the data bottleneck phenomenon caused by the speed difference between memory devices must be prevented. To solve this problem, researches on storage class memory of new concept to reduce the latency gap between DRAM and NAND, which are the most widely used memory devices, are actively being carried out.

For this purpose, among the various storage class memory candidates, we are studying RRAM (Resistive Random Access memory) devices showing high-integration density and low-power characteristics.

Furthermore, in order to maximize the high-integration density property of the RRAM device, a cross-point memory array consisting of the RRAMs should be implemented. However, in arrays consisting solely of memory devices, several problems such as leakage current occur. Therefore, a selector device that can act as a switch to selectively operate only the cells of interest in the RRAM array is under development.

Finally, we are developing a neuromorphic system hardware that imitates human neural networks using RRAM-based cross-point arrays developed to solve the problems of existing computing systems such as slow computing speed and large power consumption.