Inside view

20 June 2013

Why memristors?

As semiconductor devices continue to scale down, existing nonvolatile memory technologies are approaching their physical limits. The need for breakthroughs in nonvolatile memory technologies is one of the driving forces of the rapid development of memristors during the last few years. Compared with transistor based memory technology, memristors are viewed as potential candidates for future memory devices, due to their expected smaller size and lower power consumption.

The memristor (short for the ‘memory resistor’) was orginally defined by Leon Chua in 1971 as the fourth fundamental circuit element forming a non-linear relationship between electric charge and magnetic flux linkage. In other words, the resistance of a memristor is dependent upon the history of the charge or current that has moved through the device.

A broader definition of memristors - argued by Chua and others - includes 2-terminal non-volatile memory devices based on resistance switching, called resistive random access memory (RRAM or ReRAM). Memristors present hysteretic I-V characteristics, in which the resistance of the device changes based on the amount of charge that passes through the device, and exhibit an apparent high resistive state (HRS) and low resistive state (LRS). The reversible switching between HRS and LRS can be achieved by the polarity or the amplitude of the applied voltage and has applications in digital and analogue electronics.

Why the delay?

Though memristors were envisioned by Chua in 1971 and memristor emulator circuits were implemented, simple two-terminal implementations were not realized for many years, partly because researchers believed that a memristor must involve a direct interaction of a charge with a magnetic flux. It is reasonable for one to assume that hysteretic I-V characteristics for two terminal devices have been observed by many researchers since 1971, but classification of this effect as memristive did not occur until approximately 2008.

In 2008, HP labs presented a simple physical implementation and demonstrated that the hysteretic I-V characteristics of their two-terminal device could be classified as memristive behaviour. This physical model boosted the development of the fabrication of memristors and various types of memristors are now under research. Interestingly, there is a substantial amount of criticism and contention surrounding memristor technologies. This centres around arguments related to the definition of what a memristor is, if the observed hysteretic I-V characteristics of present memristive devices can be considered to be the same as that of a memristor, and what the exact physical mechanisms are that are occurring in modern memristors. Regardless, memristors and RRAM represent technologies that are suitable to realize highly scaled, densely integrated and low energy memory systems.

How did you test your memristor?

We have fabricated non-volatile physically flexible memristors using ink-jet printed conductors on a flexible plastic substrate. The devices exhibit bipolar resistive switching (BRS) behaviour and have high resistance state (HRS) and low resistance state (LRS) which is explained by the migration of oxygen between (higher resistance) copper oxide and (lower resistance) silver oxide. To explore the performance of our memristors, we tested the repeatability and persistence of the HRS and LRS. We also fabricated other control groups to prove that copper oxide plays a significant role in the resistive switching behaviour of our devices. These memristors have features of being physically flexible, compatibility with flexible electronic technologies and low cost, low-temperature fabrication and low-power operation.

Copper and silver?

We fabricated our memristors on a flexible Kapton substrate which can remain stable in a wide range of temperatures (from −273 to +400 °C). Copper and silver are abundant in nature and inexpensive compared to other materials used to form memristors (for example, platinum). Compared to other memristor fabrication processes, our Cu/Ag memristors have a relatively simple fabrication process. We used ink-jet printing technology which can be low cost and has scalability for roll-to-roll processing as well as integration with flexible electronics technology. The structure we use is classic cross structure. Many of the previous designs of memristor were fabricated on the SiO2 substrate and using photolithographic techniques. We believe our structure can be implemented using photolithographic techniques, but we would like to push the ink-jet printing technology towards fabrication of smaller feature sizes.

What else do you work on?

We are working on advanced packaging and integration technologies, including ultra-high-density and high-speed interconnects, nanotechnology for electronic and photonic devices, integration and packaging technologies for extreme environments, which include high/low temperature as well as for biological systems such as integration into human tissue. Examples include indium-coated carbon nanotube structures as reworkable contacts for device packaging, MEMS-based spring contact structures and amorphous Si for enhancement of photovoltaic structures. We characterize these devices and systems over a wide range of temperatures, frequencies and environmental conditions to explore the impact of the nanostructures on device performance.

Further reading

This article is based on the Letter: Resistive switching characteristics in printed Cu/CuO(/AgO)/Ag memristors (new window).

A PDF version (new window) of this feature article is also available.

Journal content

Cover of Electronics Letters, volume 50, issue 24

View all content

Browse or search all papers in the latest or past issues of Electronics Letters on the IET Digital Library.