High Frequency Applications of Memristors

This project aims to experimentally and theoretically study, classify and describe, the behaviour of memristors at high and radio frequencies. It also aims to develop device models and design methods for memristors, which will enable their use, both as stand-alone devices, and as add-on elements to CMOS silicon Electronics. The aim is to enable the design of high frequency nanoscale switches, filters, delay elements and mixers. A secondary aim is to study whether the memristors can also be used as memory capacitors/inductors, amplifying devices and also as pulse generators as has been proposed by several authors, and not yet experimentally confirmed.

Memristors are nanoscale non-volatile programmable resistors which have increasingly, over the past decade, been attracting attention as a promising microelectronic technology for memory and neural network applications. The potential of memristors as an add-on devices to silicon CMOS integrated circuits is currently investigated at Imperial College with EPSRC support under the FORTE programme grant.

The initial objective of this work is to generate a satisfactory memristor model which accurately describes the dynamic behaviour of memristors. Such a model will eventually serve as a scaffold to build a frequency sensitive lumped model useful for circuit design, especially at higher frequencies and fast transients. Despite the proliferation of memristor circuit models in the literature, these tend to be static and not account for dynamics, including energy storage and delay effects. An accurate lumped circuit model is required both locally and by the wider memristor research community.

The models developed in the first activity will be calibrated and tested by designing, prototyping and evaluating small scale CMOS/memristor hybrid integrated circuits within the activities of the FORTE programme.

A number of unexplored until now niches will also be explored: memory capacitance, memory inductance and local activity. The electrical response of memristors at high frequencies will be studied both experimentally and theoretically. Initial measurements of the frequency dependence of memristor resistance, capacitance and inductance will extend up to 6GHz in order to evaluate the extent to which memory effects also extend to capacitance and inductance as has been proposed theoretically; if the answer to this research question is affirmative, memory impedance effects will be studied more extensively, at a wide range of operating and ambient conditions, so that the extent to which they can be exploited in circuit applications can be appraised. Finally, a number of authors have proposed that it is possible to use the memristor as an amplifying element, because its static voltage-current characteristic bears similarities to features of the current-voltage characteristic of a tunnel diode which enable it to be used as an amplifier. Since tunnel diodes have successfully been used as very high frequency amplifiers, it has been proposed, but not experimentally confirmed that memristors may also be useful as amplifying devices.

This project's ambition is that its outputs facilitate the use of memristors in large scale analogue, neural network and Radio Frequency integrated circuit design, and this way contribute to consolidating the presence of the UK as a major player in the emergent memristor technology scene.

EPSRC Themes relevant to this project:
Microelectronics Design
Microsystems
RF and microwave devices