Missing Link of Electronics Discovered: "Memristor"

sciam.com ^ | May 1, 2008 | JR Minkel

[neverdem]

Memory plus resistor may add up to longer-lasting batteries and faster-booting computers

After nearly 40 years, researchers have discovered a new type of building block for electronic circuits. And there's at least a chance it will spare you from recharging your phone every other day. Scientists at Hewlett-Packard Laboratories in Palo Alto, Calif., report in Nature that a new nanometer-scale electric switch "remembers" whether it is on or off after its power is turned off. (A nanometer is one billionth of a meter.)

Researchers believe that the memristor, or memory resistor, might become a useful tool for constructing nonvolatile computer memory, which is not lost when the power goes off, or for keeping the computer industry on pace to satisfy Moore's law, the exponential growth in processing power every 18 months.

You may dimly recall circuit diagrams from your middle school science class; those little boxes with a battery on one end and a lightbulb on the other. Ring any bells? To an electrical engineer, the battery is a capacitor—a device for storing electric charge—and the lightbulb is a resistor—an obstacle to electric current. Until now, engineers have had only one other basic element to work with—the inductor, which turns current into a magnetic field.

In 1971 researcher Leon Chua of the University of California, Berkeley, noticed a gap in that list. Circuit elements express relationships between pairs of the four electromagnetic quantities of charge, current, voltage and magnetic flux. Missing was a link between charge and flux. Chua dubbed this missing link the memristor and created a crude example to demonstrate its key property: it becomes more or less resistive (less or more conductive) depending on the amount of charge that had flowed through it.

Physicist Stanley Williams of HP Labs says that after a colleague brought Chua's work to his attention, he saw that it would explain a variety of odd behaviors in electronic devices that his group and other nanotech researchers had built over the years. His "brain jolt" came, he says, when he realized that "to make a pure memristor you have to build it so as to isolate this memory function."

So he and his colleagues inserted a layer of titanium dioxide (TiO2) as thin as three nanometers between a pair of platinum layers [see image above]. Part of the TiO2 layer contained a sprinkling of positively charged divots (vacancies) where oxygen atoms would have normally been. They applied an alternating current to the electrode closer to these divots, causing it to swing between a positive and negative charge.

When positively charged, the electrode pushed the charged vacancies and spread them throughout the TiO₂, boosting the current flowing to the second electrode. When the voltage reversed, it slashed the current a million-fold, the group reports. When the researchers turned the current off, the vacancies stopped moving, which left the memristor in either its high- or low-resistant state. "Our physics model tells us that the memristive state should last for years," Williams says.

Chua says he didn't expect anyone to make a memristor in his lifetime. "It's amazing," he says. "I had just completely forgotten it." He says the HP memristor has an advantage over other potential nonvolatile memory technologies because the basic manufacturing tools are already in place.

Williams adds that memristors could be used to speed up microprocessors by synchronizing circuits that tend to drift in frequency relative to one another or by doing the work of many transistors at once.

Whether industry will adopt it remains to be seen. In an editorial accompanying the paper, nanotech researchers James Tour and Tao He of Rice University in Houston note that "even to consider an alternative to the transistor is anathema to many device engineers, and the memristor concept will have a steep slope to climb towards acceptance."

But the memristor concept is a promising one, they wrote, adding: "It is often the simple ideas that stand the test of time."

[JRandomFreeper]

Still sounds like a PIN diode. What are the capacitance effects with biasing?

/johnny

[ThePythonicCow]

Still sounds like a PIN diode.

But but it is NOT A DIODE!

It does not pass current in one direction more readily than in the other direction.

Furthermore, I don't know if it has any bias affects either way (change in resistance or capacitance based on which way and how much current is flowing through it) but I haven't seen anything in the explanations so far that would lead me to conclude it has bias affects.

It is a reprogrammable resistor, requiring just a brief flux of current to reprogram it.

[fishhound]

Thanks for posting this.

Bookmark

[JRandomFreeper]

Ummmm. You can trick (some) resistors and capacitors into behaving like diodes, depending on the frequency.

Structure and action seems like it's a PIN diode. And a PiN diode isn't a regular diode.

The whole thing looks like a tempest in a teapot. I would expect smart guy physicists to be working on the froth between the electron shells.

But I'm just a cook.

/johnny

[ThePythonicCow]

Structure and action seems like it's a PIN diode.

The action of a PIN diode, or any diode, is that current flows one way easier than the other. This device does not have that action. Current flows either way just as well.

The structure of a PIN diode involves some p-type and n-type semiconductors in a particular configuration. This device has no semiconductor components at all.

It's structure and action are clearly and entirely not those of a PIN diode.

Period.

[mkjessup]

Looks like the biggest cheese dog in the world, but where’s the bun? Some chili to go with all that cheese would be good too.

Seriously, a fascinating article, thanks for posting! :)

[JRandomFreeper]

The PiN diode is the Texan of diodes. It acts like a diode, resistor, and sometimes, a very small uf capacitor. It has fast switching times, fast enough to confuse an Old Crow.

It also has some inductive properties, due to the lead length.

It is similar, and I call Bravo Sierra.

/johnny

[NicknamedBob]

Now if they can only work out how to make a superconducting memristor, I'll have everything I need for my flux capacitor.

Time to start working on the power supply ...

[ThePythonicCow]

It (the PiN diode) is similar

You do seem to be more familiar with the PIN diode than I am. Good.

However, I can find nothing whatsoever, other than the word 'resistor' and the fast switching times, in the description of memristors, resembling what you describe of the PIN diode.

Could you please point to anything that suggests that the memristor has any semiconductor material, that it has any non-trivial inductance, that it has any diode properties, or that it has more capacitance than one would expect from a -very- short (high or low, depending on state) resistance titanium dioxide separating two conductors?

Nor have I heard or seen any suggestion that PIN diodes have the persistent memory, without applied power, of memristors. What memory affects I can find in PIN diodes seem to be quite transient. There is a difference in conductivity of the middle layer of a PIN diode, the I-layer, depending on the bias voltage on the device, but this difference does not persist after the bias is removed, so far as I can tell.

I'm not disagreeing with you on what PIN diodes are; though I can't claim to understand them as well as you.

I have described how memristors have properties, material, structure and action that are quite different than PIN diodes.

Other than you repeated and entirely unsubstantiated claims that memristors are 'similar', I find nothing in your replies justifying these claims.

I call Bravo Sierra, in response.

In particular, the uses of these devices are quite different. The PIN diode has such as uses as controlling an RF signal, rectifiers and RF filters, while the memristor has uses such as very compact, low power, fast switching, non-volatile memory.

Also the device size is orders of magnitude different. Commercial PIN diodes seem to be provided in separate packages, orders of magnitude larger than even the initial laboratory samples of memristors.

For those who might be getting tired of our bickering, here's a reasonable (to my untrained eye) explanation of a PIN diode, from http://hyperphysics.phy-astr.gsu.edu/hbase/Solids/diod.html:

The PIN Diode

The PIN diode has heavily doped p-type and n-type regions separated by an intrinsic region. When reverse biased, it acts like an almost constant capacitance and when forward biased it behaves as a variable resistor.

The forward resistance of the intrinsic region decreases with increasing current. Since its forward resistance can be changed by varying the bias, it can be used as a modulating device for AC signals. It is used in microwave switching applications.

A PIN diode has three layers: a positively doped semiconductor, an undoped intrinsic semiconductor and a negatively doped semiconductor.

A memristor also has three layers (aha - another similarity): two ordinary conductors sandwiching a titanium dioxide with a few missing oxygen atoms.

[JRandomFreeper]

Yes, I might be more familiar. That happens over long times and close proximity.

You left out frequency in your analysis.

Ok. I'll quit bickering. All you have to do is build a functional radio transmitter/receiver with the product. I've done it with rusty razor blades, fool's gold, and a couple of other items that didn't have a substrate of ANY sort.

/johnny

[JRandomFreeper]

And put it in an Altoids box?

[JRandomFreeper]

Yep.

Meet the next requirement.

Just a cook my skinny arse..

/johnny

[ThePythonicCow]

All you have to do is build a functional radio

I'd have trouble building a radio with a Heathkit and a soldering iron ;).

Heathkit HW-8 80-15m QRP CW Transceiver

But can you build a non-volatile memory with a rusty razor blade, or a PIN diode?

[ThePythonicCow]

Oops - typo - make that:

But can you build a non-volatile memory with a rusty razor blade, or a PIN diode memristor ?

[ThePythonicCow]

C**p - not typo - I’ll shut up now ;).

[Robert A Cook PE]

Buy HP.

8<)

[Zuben Elgenubi]

From the photo in post #1: "Each wire is 50 nanometers wide, equivalent to about 150 atoms."

Is this possible? I'm surprised we can see mass at that atomic level. Boy, I've got a lot to read up on.

[Old Student]

“Buy HP.”

Looks like the most sensible comment on the thread. High-capacity solid-state hard-drives, anyone?

[Robert A Cook PE]

Is this possible? I'm surprised we can see mass at that atomic level. Boy, I've got a lot to read up on.

IBM was writing their name using single atoms (of xenon I believe) a few years ago on a substrate of another substance.

[steveo]

ah... the good old days...