The elephantine thing

Elephants are reputed to never forget. If this fable were true, then perhaps, even though their brains are relatively small, the elephant must have found a way to pack lots of memory into its brain's limited space. Which is exactly what reader Sander Olson tells us that a Swedish company, ThinFilm, is doing. Not increasing the packing density of pachyderm gray matter, but as in that famous line from The Graduate, ThinFilm sees immensely dense memory futures-in 'plastics.'

Plastic memory 

Rather special plastic, of course, where a very thin sheet of the polymer is sandwiched between two tiny grids of electrodes. At each intersection of the checkerboard of electrodes (one set on top of, and the other below the polymer sheet), they claim that a bistable memory cell pops into existence.

A voltage applied to a given cell can modify the organic nature of the polymer at that spot, changing it from one state to another. And that state can be read at a later time. Because this 'state change' is to the chemical nature of the polymer at that location on the sheet, the state is non-volatile - this memory, unlike today's typical RAM, won't lose its mind when the power goes out. Which would make it very 'pocket device friendly.' 

So-we have a 'one' or a 'zero.' Binary storage. Without, you'll note, any silicon in the memory array! In fact, ThinFilm believes that this technology can be produced using 'roll to roll' manufacturing techniques, similar to how a newspaper is printed, which could lead to huge economies of scale and dirt cheap memory.

But what's particularly interesting about this idea, is its density. Today, a typical S-RAM memory cell takes up between four and six square micrometers of area. By comparison, ThinFilm says that their polymer cells would each occupy but one quarter of a square micrometer - quite a difference. 

ThinFilm believes that they can shrink the memory to this extent because there are no active (transistor) elements within the memory matrix, as is the case with today's memory - ThinFilm's memory cells are defined only by the passive matrix of crisscrossed electrodes. The active elements that are needed to address the memory cells and to perform the read and write operations can be located at the edge of the matrix or, alternatively, under or above the memory matrix. Which also opens new packaging structures, where the memory could be placed on top of or below another chip. This would result in far higher packing density than we're used to: consider that a gigabit of contemporary S-RAM requires from 1.5 to 6.5 billion transistors, while ThinFilm's polymer memory would require only a half-million active elements! And if that isn't dense enough, these 'memory sheets' can be stacked one atop the other, for 3D density. 

What would such dense plastic memory mean to us? According to ThinFilm, a single credit card-sized memory device built with this technology could store 60,000 DVD movies; or 126 YEARS of MP3 songs; or 400,000 CDs, or 250 million high definition digital pictures. Now wouldn't this change a lot of rules…

Of course, claims such as these sound almost too good to be true, and they may be -- when it comes to bleeding-edge research, sound skepticism is a good idea. But ThinFilm's claims may have some extra credibility, considering that Opticom ASA owns 87% of ThinFilm, and Intel owns the rest. One could well imagine that Intel would want to be involved in a technology that had the potential to take us beyond the age of silicon memory. (And remember -- as entrenched as we are today in the world of silicon, a change to a radically different memory technology could well happen - in fact it's already happened, as we moved from relay memory, to vacuum tube memory, to magnetic core memory, to silicon …)

BioMolecules?

As interesting as ThinFilm's ideas may be, they're not the only player in the organic memory game. According to an article, a group of Italian researchers have altered the structure of a special protein, 'Green Fluorescent Protein,' to create single molecules that each provide one bit of optical memory! In effect, these individual molecules can be addressed by one of two specially tuned lasers; one laser forces the molecule into a 'dark state,' while the other returns it to its 'bright state.' Since another laser can 'read' which of these two non-volatile states the molecule is in, we have an optical biological molecular memory cell! And it may be easy to construct large 3D arrays of these memory molecules, according to National Enterprise for NanoScience physicist Vittorio
Pellegrini,

'Since the active element of the device is a protein, it is possible to develop volumetric memory devices in which the proteins are forced to self-assemble…' 

It's NOT Just Memory… 

Of course it isn't only 'memory' that's on the verge of the Lilliputian. As we've followed in recent issues, computing elements themselves are teasing us with revolutionary new ways of shrinking, even to the point of complete transistors composed of but a single molecule, and of transistors made out of carbon nanotubes. Now, as others have pointed out, Ehud Shapiro of Israel's Weizmann Institute has developed a specialized DNA computer so small, that a trillion of them fit into one test tube!

At this point, these DNA computers, the first whose input, output, and software are all DNA molecules(!), are specifically designed to process other DNA molecules, such as for DNA sequencing. But DNA computers have already been experimentally used to solve more traditional mathematical problems. According to Shapiro "The living cell contains incredible molecular machines that manipulate information-encoding molecules such as DNA and RNA (its chemical cousin) in ways that are fundamentally very similar to computation." 

Over and above DNA's computing capabilities, it holds the potential for huge amounts of very tiny storage - densities 100,000 times greater than today's hard disks! Another way to look at this is that one cubic centimeter of DNA can hold the information stored on one trillion CDs. Think of the MP3 player you could have… 

Is It Soup Yet? 

Revolutionary -- I can't yet judge the likelihood that ThinFilm's technology, or fluorescent molecular memory, or DNA computing and storage, might actually make it into the marketplace. And of course it may come to pass that none of these particular advances ever do. But even if one or more of these ideas fail, they still represent good demonstrations of how brilliant scientists and engineers just insist on finding new technologies, and new techniques to explore. Which is why I feel very confident that one (or more) of these or other revolutionary advances will indeed hit a jackpot. After all - history is full of our doing exactly that!

Evolutionary -- And then again, even if NONE of these revolutionary advances yield fruit, we have fascinating, merely 'evolutionary' changes ahead, highlighted by Intel's recent announcement of a 'TeraHertz transistor'. This tiny transistor can switch on and off one trillion times each second - far faster than the transistors in today's Pentiums, and they consume far less power (and generate far less heat.) Dan Hutchinson, president of VLSI Research, commented on Intel's ability to build things on its chips, 'one atomic layer at a time:' "They've solved some of the electrical problems that looked like brick walls." 

Which really shouldn't surprise any of us. That, after all, has been, and will remain, the name of the technology game! Intel expects this new type of transistor to make its way into microprocessors within four years, and they expect that by the end of this decade, this will lead to commodity chips with,'Twenty-five times more transistors in processors than in current ones, running at ten times the speed, yet with no increase in power.'