In 1971, Leon Chau of UC Berkley did more than wishful thinking and postulated a fourth element: the memristor [1]. The theoretical device — which some engineers have termed the ‘missing element’ — could remember the changes in current passing through it by changing its resistance. What’s remarkable about Chau’s mathematical postulate is that this past May, researchers at HP Labs, Palo Alto, announced they had actually developed memristors. And, as expected, several patents assigned to HP dealing with memristors suddenly appeared on the USPTO website. Apparently, scientists at HP have been working with memristors for some time now.
According to Chau’s publications, press releases from HP, and patent applications, a memristor behaves like a non-linear resistor with memory — a behavior due to hysteresis. Electrical hysteresis is commonly associated with ferromagnetic materials, as depicted by the double-s-shaped hysteresis loop. While hysteresis is often an annoyance in magnetic tape and hard drive circuitry, it has many practical applications in electronics. The use of a Schmitt trigger to eliminate contact bounce in switches is one practical application of hysteresis.
According to patent applications, HP’s memristor design is based on a sandwich of titanium dioxide films that loosely resembles the construction of a capacitor. However, unlike capacitors, one layer of titanium film is an insulator while the other layer is doped so that it is a conductor. Wires on either side of the sandwich are used to pass a charge through the film. Current through the film lowers the resistance of the film by a factor of 1,000. In this respect, the memristor seems a lot like an active device to me, but it’s nonetheless classified as a passive circuit element.
Some of the first applications of memristors will likely be high-density, energy-efficient memory devices that compete with Flash memory devices. However, as with any new technology, it’s difficult to predict what will come of the memristor. Because the memristor has qualities of both analog and digital circuitry, one application identified by HP is the development of memristor-based neural networks [2]. Perhaps memristors will make it possible to implement affordable, semi-autonomous robots with a semblance of intelligence.
If you’re like me, you probably can’t wait to order an assortment of memristors from Mouser or Jameco. However, we’re not there yet. We’ll probably have to be content with a SPICE simulation or other virtual experience for months, if not a year or more.
One of the unfortunate aspects of memristors — from the perspective of an electronics experimenter eager to create the first ‘must have’ device using memristors — is that just about every imaginable application is patented. Take a look at [url=http://www.uspto.gov]http://www.uspto.gov[/url] or http://www.freepatentsonline.com. One of the most straightforward patent applications — United States Patent 20080090337 — is for an electrically actuated switch. The application provides a good background on the mathematics and physics of a memristor. A second patent application — United States Patent 20080079029, ‘Multi-terminal electrically actuated switch’ — provides additional information, including specifics on materials and doping.
It’s not likely that memristors will make time travel or anti-gravity possible. However, it’s a new building block that will undoubtedly have myriad practical applications in areas ranging from computing and entertainment to automotive electronics. And you can be certain that Nuts & Volts will be featuring projects that illustrate the utility of the memristor — just as soon as I can get my hands on one. NV
REFERENCES [1] Chua, L. Memristor — The missing circuit element. IEEE Transactions on Circuits Theory, Volume 18, Issue 5, Sep 1971 Page(s): 507 – 519.
[2] Johnson, RC. First proof of circuit theory’s ‘missing link?’ EETimes. Monday, May 5, 2008. Page(s): 1-6.
