Electronics: The fourth element

[arendt]

Source: Nature

The memristor might provide a new path onwards and downwards to ever-greater processor density. By fabricating a cross-bar latch, consisting of one signal line crossed by two control lines5, using (two-terminal) memristors, the function of a (three-terminal) transistor can be achieved with different physics. The two-terminal device is likely to be smaller and more easily addressable than the three-terminal one, and more amenable to three-dimensional circuit architectures. That could make memristors useful for ultra-dense, non-volatile memory devices.

For memristor memory devices to become reality, and to be readily scaled downwards, the efficient and reliable design and fabrication of electrode contacts, interconnects and the active region of the memristor must be assured. In addition, because (unlike with transistors) signal gain is not possible with a memristor, work needs to be put into obtaining high resistance ratios between the ON and OFF states. In all these instances, a deeper understanding of the memristor's dynamic nature is necessary.

It is often the simple ideas that stand the test of time. But 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. Some will undoubtedly trivialize the realization of this ubiquitous nanoscale concept, whereas others will embrace it only after the demonstration of a well-functioning, large-scale array of these densely packed devices. When that happens, the race towards smaller devices will proceed at full steam.

Read more: http://www.nature.com/nature/journal/v453/n7191/pdf/453042a.pdf

---

If you subscribe, there is a full scientific paper in the same issue (published TODAY).

arendt

[pattmarty]

1. This looks important, but what we need on DU is ..............

....DU'rs with technical expertise to post here with explanations that (most of)us lay people can understand.

[arendt]

2. To be honest, I don't think even people w tech skills really know if this is...

a real breakthru or just another press release that will be dead in six months.

You should search wikipedia on "Moore's Law" to understand that the fifty year history of ever more transistors for ever less money is coming to an end. Scientists are frantically looking for ways to get out of the old way of simply shrinking a known system (silicon, photolithography). They have been looking at really strange stuff, like "quantum dots", where you measure whether or not one electron is present in a trap made of a few dozen atoms.

So, a totally new direction like this will definitely be investigated.

But, today's transistors are already being fabricated at 45 nanometers design rules - that is, one dimension of a transistor is on the order of 400 atoms. They can't go any smaller than that. An entire transistor today has only about 1 million atoms.

If these bulk memristor devices can be scaled smaller than 45 nm without loosing their properties, then this may work.

But the devil is in the technological details of fabricating the stuff at this scale in an affordable and reliable way.

arendt

[pattmarty]

3. Thx.

[salib]

8. Also, the issue of "no gain" is more important than the snippet lets on

A transistor makes a great amplifier (i.e., it produces a signal boost), so that nearly indistinguishable effect from those squiggly lines on a phonograph, that weak radio signal, or that microscopic magnetic effect on your hard disk, is amplified to be useful, and more importantly, transmitted around to those people (and devices) which need it. Now, transistors are also great switches (really another take on the amplification effect) and thus great memory devices. The bonus is, you can read it AND transmit the result with "high gain", or amplified, and thus it can get to those other destinations (you lose signal over distance/time so the initial one has to be strong).

The "memristor" being described in the article is, on the other hand, a "no gain" device, so whatever you read must subsequently be amplified (i.e., with a transistor). Superconductive devices might help (no resistance loss of signal, but still great interference loss), but that is only conjecture. It is likely that this will prove to be useful in terms of minimizing power use for large amounts of high speed storage (those transistors have to be constantly powered to work, and they lose their "memory" and have to be reset each time you "read" them), and that assumes the signal amplification issue does not wipe out any savings on power, or as related to power, size.

[arendt]

11. Good point. As I said, this could be a flash in the pan.

Still, today's DRAM is a no-gain device (just a LEAKY charge bucket). You still have to amplify (or at least use some exotic signal processing like partial likelihood (jargon failure) to read.

I don't see "no gain" to be as big a problem as the atom-level scale needed to surpass existing technogies. I mean, its usually the case that the bulk properties a compound/alloy/semiconductor/etc. do not "emerge" until you reach at least nanometer scale. If this memristor depends on some exotic solid state properties (the "nanomaterial" referred to), then there isn't much room to shrink below the existing 45 nm design rules without the bulk properties vanishing.

arendt

[EnviroBat]

4. Careful, this is what led Sky Net to become self-aware.

After that, it decided to obliterate all human life on the planet.

Better brush up on your Terminator there Skippy.

[arendt]

5. Cool. Maybe I could scare people into becoming tech literate...

eat your Biology book, junior, and grow up to be real smart - or the terminators will get you. :-)

arendt

[EnviroBat]

6. Zactly!

:P

[Indenturedebtor]

7. Posted this yesterday

Lots of discussion to browse here:

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=102x3290850

Different article though! Thanks for posting the nature link!

[arendt]

9. Right. I saw your OP. Thought that mentioning this new article would get lost...

at the bottom of your thread, so I started a new thread for the new article.

It was your OP that made me pay attention to this when it came up in Nature. Thanks.

arendt

[ogsbee]

10. This could indeed lead to a God machine or skynet, it seems to me

A computer that listens intelligently to all phonecalls, reads all emails and stores them for all eternity. We know the military already wages information warfare against their own people -- paid retired officers, embedded reporters, etc. All in an attempt at full spectrum dominance in the information war (read the phrases used in the retired officer propaganda project). Why wouldn't they want to push that as far as possible? The elite want to ensure their godlike status. (How many generals end up fantastically wealthy, anybody know?)

[OhMcNo]

12. More (Moore?) info at Ars...

You can read more about it here...

http://arstechnica.com/news.ars/post/20080501-maintaining-moores-law-with-new-memristor-circuits.html">Maintaining Moore's law with new memristor circuits

[arendt]

13. Thank you! Ars Technica is a great site. n/t