Molecular Computing,
The Next Information Systems Revolution

An overview of a family of new technologies (in 1995) that is going
to turn the computer and consumer electronics industries
upside down.

Francis Vale

By the years 2000-2005, users may have access to massively parallel, palm-held computers whose processing power will surpass a gargantuan NCube machine. These diminutive new systems may have terabytes of data storage, emit almost no heat, run for days off a single battery charge, and possibly exhibit a new type of higher order 'wet' intelligence that goes well beyond the usual notions of AI.

Perhaps even more incredible, these super high speed systems of tomorrow may not be transistor-based at all. Instead of semiconductors, these future systems may be using ultra-fast molecular-sized devices 'grown' from biological systems (biomolecular electronics); or use miniaturized molecular devices (nanotechnology).

When (not if) these radical new systems appear on the market, they are going to cause a massive dislocation of semiconductor-based computing systems and their associated industries. This coming upheaval will encompass consumer electronics, as well as most every type of content producer.

It is therefore critical that industry professionals of all stripes now begin to understand the implications and the genesis of this completely new computing environment.

This new scientific field has already grown well past its early infancy. For example, three years ago, on June 25, 1992, the space shuttle Columbia thundered into orbit carrying a most unusual payload: a purplish looking, coastal swamp marsh bacteria commonly found in the San Francisco Bay area.

This bacteria, Halobacterium halobium, owes its special color to a unique pigment found in its cell membrane called bacteriarhodopsin (bR). The bR protein captures photons of light and converts them into cellular energy for Halobacterium. For molecular computing researchers, bR has a very special, highly prized property -- bR devices can be constructed that will switch between alternate states, just like the binary logic of today's semiconductor-based digital systems.

Cooled sufficiently, a nanometer-sized section (a nanometer is a billionth of a meter) of the bR molecules will kink out of shape when struck by a green laser. But most important, the altered bR molecules can be made to snap back to their original form if hit by a red laser. Hence, bR can act as the basis for a molecular binary switch.

Such bR-based molecular storage devices could potentially store as much as 480 gigabytes of data (roughly equivalent to 20 million pages of typed text) in just five cubic centimeters that can be read, written, or erased in as little as five pico (five trillionth) seconds using present laser diode technology.

In contrast to such a bR-based system, today's semiconductor systems are a thousand times slower, and would require enclosures the size of a home refrigerator to hold an equivalent amount of data. Also, unlike 2D semiconductors, bR devices are naturally 3D in geometry. As such, bR's unique architecture might herald a new era of multi-dimensional, holographic computing.

The space shuttle experiment was conducted by Professor Robert Birge, Director of the Center for Molecular Electronics at Syracuse University. The experiment was sponsored by the W.M. Keck Foundation, and BioServe Space Technologies. Digital has also been a long time supporter of Birge's work via various hardware and software grants to the Center.

But Birge is not alone in his quest. Worldwide, there are many other researchers looking into photo-active bR-based systems, as well. Among them is a Dr. Felix Hong at Wayne State University. He is working on a proof of concept that biological materials like bR can function the same as a conventional computer chip.

Dr. Hong's research is building upon the work of others at Wayne State. This work has shown that bR can be altered via genetic engineering; thus producing highly successful mutants that are much more amenable to being used as molecular devices than naturally occurring bR. Says Hong, "This (research) raises the hope of breeding mutants with the right type of intelligence for the intended design of the molecular device "

Hong uses chemistry to control the switching of his bR devices. By modifying the hydrogen ion concentration (pH) surrounding the protein, Hong is able to easily modulate the electrical behavior of his bR device. Both Hong and Birge use light as the input to their bR systems. But their bR research differs in one fundamental way: Hong is using bR's electrical signal as the output; while Birge is concerned with bR's light output as the information carrying medium.

According to Hong, mutant bR systems could eventually open the way to the construction of "a single molecule, or a supramolecular- cluster with massively parallel and massively distributed processing capabilities." That is, genetically engineered bR may one day produce a biomolecular 'NCube machine' that will easily fit within the confines of your current palm-held computer

But researchers are also exploring non-biological approaches to produce miniaturized molecular devices. This is the realm of nanotechnology.

For example, Richard Potember at the Applied Physics Laboratory at Johns Hopkins University has patented a radically new kind of data storage device that uses a scanning tunneling microscope (STM).

STMs have been in use for some time now, but as research 'microscopes' for exploring atomic domains. An STM consists of a very sharp needle which is brought so close to the surface of a material that their respective electrons spinning within their atomic orbits actually overlap. When a small potential difference (electric charge) is applied between the STM needle and the underlying material, electrons will 'tunnel' from the needle's tip to the material, or vice versa.

Potember's group has demonstrated that it is possible to use an STM to cause a field-induced, reversible phase transition in some types of underlying materials. The transition is driven by the electric field at the tip of the STM.

The type of material being used at Johns Hopkins is a silver and copper-based metal salt called tetracyanoquinodimethane (TCNQ). These phase transitions cause TCNQ to go back and forth from a high impedance state to a low impedance state, thereby yielding specific angstrom-sized domains (an angstrom is one ten-billionth of a meter) that are detected by the STM. These reversible operations are roughly analogous to the heads of a disc drive doing read/writes on some type of underlying substrate.

This STM/TCNQ device is being designed to do read, write, and erase operations within individual domains whose dimensions are just 30 to 40 angstroms (present-sized domains on magnetic or electro-optical disc drives are typically orders of magnitude larger, at about one square micron.)

At these extraordinarily small dimensions, this molecular electronic device could offer terabytes of information storage capacity (i.e., many times more data than that held by your average city library). Potember says that once you strip away all the experimental apparatus from an STM, his STM system could fit inside the same sized enclosure as a conventional 5.25" optical disc drive. Or in the same space as one of the highly touted new super discs, like the new DVD -- Digital Versatile Disk (sic). These new DVD laser discs will at first offer 4.7GB of storage, and in later generations will hold about 20 GB. This is still an incredibly small amount of storage, however, when compared with the tens, even hundreds of terabytes these molecular devices will offer. The impact on audio and video content producers is obvious: No more data compression required. This new development will also have staggering implications on how A/V content is produced.

But like any radically new technology, there is a daunting learning and manufacturing curve that must first be overcome before these molecular devices can be mass produced. They are still five to ten years away from becoming a commercial reality. Given such projected long lead times, one could speculate that advances in semiconductor based systems, as well as in magnetic and laser storage technologies, might put them on an equal footing with any far off molecular system.

However, there are some potentially intractable problems that could severely hinder the growth path of any of these traditional computer materials. These problems include: 1) escalating costs to design and fabricate; 2) the weird world of quantum mechanics; and 3) in the possible case of semiconductor-based processors, massive heat generation.

1) the costs to design and build a 64 megabit memory chip are estimated to be on the order of a billion dollars. These costs are staggering and will get even higher with larger capacity chips. It is one of the primary reasons why IBM has joined with Toshiba and Siemens to design 256 megabit chips.

But some types of biomolecular systems, like bR, offer the promise of being economically grown in a vat (no billion dollar fabs, just a few $ million or so will do); can be quickly harvested in a normal working environment (no worry about painstaking clean room procedures); and be easily controlled via ordinary chemistry, or use off the shelf laser diodes (no need to build complex and expensive control circuitry.)

The continuing revolution in genetics engineering also offers the promise of being able to easily refine and extend the useful features of such biologically-based systems.

2) The next big stumbling block for traditional computing materials is their ever decreasing size. When something is small enough to the point where you are selectively dealing with individual atoms, quantum effects take over.

In quantum theory, the Heisenberg uncertainty principal states that just the act of observing such atomic snippets effectively alters the behavior of that which you are observing (i.e., to see it is to change it.) As a consequence, it is impossible to ever know what is precisely going on in the atomic realm.

Obviously, this can cause all sorts of problems; especially for those engineers designing semiconductor systems having vanishingly small transistors.

One possible way to overcome quantum effects is to create redundant circuits. By using their summed signals, it may ensure that you are getting predictable behavior out of the Lilliputian circuits, uncertainty principal or no. But as transistor-based systems are not naturally redundant, this feature has to be fabricated in, which adds extra costs to an already expensive proposition.

But the creators of molecular-based systems have it much easier, as many billions of atoms are stuffed into even the smallest patch of material. Thousands of molecules can thus be used to carry or encode identical information without worrying about using up all of the available storage capacity.

By taking advantage of this natural redundancy, and using averaged output, the molecular systems designer can reasonably predict that the data is being handled correctly, despite quantum effects. This particular technique is called ensemble averaging.

It is thus a classic case of either designing from the bottom up, or the top down. What nature does on the cheap and in abundance, you have to artificially design and manufacture-in when using traditional materials.

In the next issue of 21st, we will examine the third, and most interesting problem, thermal buildup; and how a radical new type of molecular device might offer an extraordinary solution.

Copyright 1996, Francis Vale, All Rights Reserved


21st, The VXM Network,