Silicon-based single-electron memory using a multiple-tunnel junction fabricated by electron-beam direct writing

Abstract

Multiple-tunnel-junction-based nonvolatile single-electron-memory devices are promising for fast write/erase operation and long retention time. Fabrication of multiple-tunnel junctions with a predetermined number of barriers and islands is a major problem in realizing such devices. We have fabricated multiple-tunnel-junction-based single-electron devices by an electron-beam direct writing technique in a silicon-on-insulator layer. Using this technique, it is possible to fabricate multiple ultrasmall islands and tunnel barriers at a predetermined position, which is very important for reproducible device characteristics. Single-electron-memory devices based on multiple-tunnel junctions are fabricated. In these devices, a multiple-tunnel junction connects the gate electrode and a storage island. The Coulomb blockade across the multiple-tunnel junction acts as an energy barrier. Single-electron-memory operation is observed at 20 K. Retention time of at least 4 h has been observed.

Nanowire Single-Electron Memory

“…In 1993, Yano et al demonstrated controlled room-temperature storage of individual electrons in such nanoscale floating gates. 1 Multiple tunnel junction (MTJ) memories [2][3][4][5] are expected to have considerably faster write speeds than flash memories due to a reduced tunnel junction resistance. The memory function in MTJ memories is attained by trapping of electrons in metastable states.…”

Electron-beam direct writing using RD2000N for fabrication of nanodevices

“…Details of the electrical characteristics of these devices have been reported separately. 13 The retention time of electrons in the storage island, estimated from the gate voltage sweeping speed, is at least 4 h at 20 K.…”

Can single-electron integrated circuits and quantum computers be fabricated in silicon?

“…Extraordinary ingenuity has been applied to demonstrate basic single-electron logic circuits in silicon [2,3], but the problem of random polarization o sets on each Coulomb island continues to block e orts toward higher levels of integration. While single-charge phenomena may ÿnd important memory applications within present technology [4][5][6], advanced computation is di cult to imagine in the presence of random o sets.…”