Friday, June 6, 2014

Change: Part 2 - Invention

Previously we looked at the many capabilities that are being developed to extend the usability of silicon post-Moore's law. Though all of them have value and are a credit to the energy, drive, can-do attitude and creativity of High Tech engineers, they are all, in the end, stopgaps - expensively developed extensions of a technology that has reached its effective end of life.

In this editorial, we'll examine some new technologies that go beyond High Tech's normal operational mode of innovation and step up to full blown invention. We'll also discuss what the point is of pursuing these technologies.

To invent, you need a good imagination and a pile of junk. - Thomas Edison

Graphene

Just a few days ago, a news item about Apple, Google and Samsung hit the wires that, to my great surprise, attraction almost no notice:

http://www.extremetech.com/computing/182657-the-next-apple-samsung-battle-will-be-fought-over-graphene-not-in-the-courtroom

The fact that the two Area 51 companies (Apple and Google) were joining Samsung (a company whose history shows that it is clearly trying to join the Area 51 ranks) in a race to gobble up graphene intellectual property came as no great shock to me. These are visionary firms who have a keen eye set on the distant horizon. Few companies indeed can legitimately claim to have such a far - reaching and effective strategic outlook. This raises the inevitable question: why are they interested in this somewhat exotic and rather recently discovered material?

At the molecular level, graphene looks like a sheet of interconnected carbon aromatic rings - or, to put it more irreverently, chicken wire. The material is, as one would surmise from the intrinsic nature of carbon aromatic rings, a highly efficient conductor of electricity. It also has shown excellent thermal conductivity and remarkable tensile strength.

Carbon aromatic rings are both highly stable and highly reactive, which is why they are so central to organic chemistry. An entire sheet of intimately linked rings such as is found in graphene thus presents enormous opportunities and, naturally, special headaches for researchers attempting to develop the next semiconductor technology for the chip sector. Some examples of the unique nature of this material include the following: 
1. Electrons and holes propagate only thru the single atomic layer of graphene, unlike silicon. 
2. Charge carrier mobility is, barring outside effects, extremely high. Furthermore, graphene is dopable; thus, carrier mobility can be controlled.
3. Resistance is not linear like in the copper interconnect of silicon, but observed at discrete intervals thru the lattice itself. 
4. Electrons and holes behave as if they have no mass. 
5. The properties of the sheet might be manipulated in such a manner that conduction of charge carriers can occur without loss. Stated differently: the material could possibly be shaped to form its own intrinsic wire routing/interconnect between circuit elements.
6. Graphene sheets are rollable and bendable. Though this currently seems to negatively affect other intrinsic properties, there is active research underway to overcome these degradations for applications such as rollable video displays.
7. Various techniques have been successfully explored to instantiate elements that are recognizable as FET or bipolar transistors.
8. There are a number of properties intrinsic to graphene which lend themselves to many of the other capabilities being researched (quantum computing, silicon carbide, carbon nanotubes, spintronics and others) so effectively that it currently appears to be one of the preferred mediums for developing, implementing, advancing and supporting such technologies.

In short: current graphene research suggests that it may be something of a 'miracle' material, capable of implementing circuitry with density, speed and power characteristics vastly superior to anything silicon could ever hope to offer. 

Yet that's only the tip of the iceberg regarding graphene's potential. There are research efforts underway to explore graphene's innate affinity for being employed in conjunction with a broad range of composite materials and fluids.

Let that sink in for a moment. Think about what might be the potential range of applications for a 'smart film.' One can immediately grasp that a substance with the kind of flexibility graphene may have would make today's silicon-based proliferation of High Technology look like chicken feed.

Of course, there is one minor obstacle: no one has yet found a way to fulfill these wondrous capabilities on a full manufacturing line. Nevertheless, the Area 51 companies are not known for making foolish technology choices. It's not unreasonable to conjecture that it's not a matter of whether graphene will completely reshape the world, but when. 

If you google "graphene applications", you will be inundated by an astounding variety of ideas that seem drawn right from the pages of science fiction. Rather than review them, though, let's stay focused and look at some of the other post-silicon technologies that are being researched - a number of which have a high possibility of being associated directly with graphene.

Nanotubes

Carbon nanotubes are strongly related to graphene, as they also are composed of a chicken wire pattern of aromatic rings. They can be formed by rolling a portion of a graphene sheet into a tube shape. These tubes can naturally combine with each other to form very long 'ropes.' The resulting structure has incredible physical strength, as well as similar electrical and thermal properties to its graphene 'parent.'

Tubes can be nested within one another, bent into the shape of a donut and have other carbon molecules attached to their surfaces. Each of these configurations offers unique properties for a wide selection of applications. Because of their strength, thermal conductivity and malleability, carbon nanotubes seem ideally suited for nanomachine development.
Unlike graphene, nanotubes have been known and studied in isolation for decades. There has been some limited commercial use of nanotubes, primarily as structural reinforcement for various consumer products. An interesting application being researched for semiconductors is the use of carbon nanotubes in combination with graphene sheets as molecular - sized 'girders' to provide structural support.The formation of FETs in a nanotube is also a well-established capability. 

It has proven problematic to assemble nanotubes modified into 'gates' in a coherent manner to form true LSI circuits. Stanford recently made something of a breakthrough in this area, however, with the announcement of a 4004-equivalent microcontroller constructed out of nanotubes:
http://www.technologyreview.com/news/519421/the-first-carbon-nanotube-computer/

Unfortunately, this remains a laboratory exercise and is clearly nowhere near being ready for prime time. Nevertheless, a material with such attractive properties is bound to attract a great deal of development effort and the future of carbon nanotubes may be just as promising as that for graphene.

Quantum Computing

Though the phrase 'quantum computing' evokes the idea of reducing logic gates from the gross molecular to the sub-atomic level, the comparison is not really accurate. For one: conventional microelectronic logic output is discrete, whereas quantum computing results are measured as a probability distribution of potential particle states. Stated differently: in quantum computing, you don't really get a 'clean' zero or one. This difference is critical, as it has heretofore driven research in quantum computing toward solving a variety of multivariate mathematical problems that are normally in the domain of supercomputing.

There are distinct advantages to the Quantum approach over conventional digital logic for supercomputing tasks. Certain problems and algorithms lend themselves particularly well to the overlapping states permitted a quantum bit (qubit) in its probability distribution, which is why there is a great deal of interest in military circles, particularly for cryptography. 

The suitability of graphene for the formation of 'quantum dots' makes it one of the preferred mediums for implementing quantum computing functions. An electron-hole pair can be confined in a graphene crystalline structure to the point that the principles of superposition and entanglement can serve to turn the quantum dot into a qubit source.  For those who are inclined to delve more deeply into this fascinating (but somewhat mind-warping) topic, here are some links to explore (WARNING - you may want to keep a bottle of ibuprofen handy:)

http://www.physics.org/article-questions.asp?id=124
http://physics.about.com/od/quantumphysics/f/QuantumEntanglement.htm

Scientists pursuing this field of research are evidently not rattled by the ghost of Werner von Heisenberg peering over their shoulders and in fact seemed to be spurred on by the challenge of turning his Uncertainty Principle from a barrier into a useful tool. Nevertheless, it may be some time before we see the entire functionality of a leading edge laptop or desktop integrated into a device the size of a shirt button or smaller.

Lighting the Candle

There are other fascinating areas of research being undertaken in universities, corporate R&D centers and public sector laboratories around the world that have not been covered in this editorial series - spintronics, which shares with quantum computing the concept of electron spin as a potential source of digital information (in particular for advanced memory architectures), storage systems of huge capacities based on biological systems, and a variety of other somewhat out-of-this-world fields of study. I find myself simply unequal to the task of covering all of these areas of development and have hardly done sufficient justice even to those which I have reviewed herein. Suffice it to say that if even one of the dozen technologies being researched bares full fruit, it will change the world to an extent that even we who have been part of the Silicon Valley experience will be staggered and overwhelmed by it.

I have an old friend, a dear friend with whom I continue to stay in touch regularly despite our geographic separation. He's one of the two or three smartest and best people I've ever known, and his perspective on things, while not always agreeing with mine, is always thought-provoking and helpful in expanding my perspective on things.

For purposes of this discussion, let's call my friend by the convenient pseudonym "Steve." I was discussing this editorial with Steve a few days ago as well as some of the others I've penned over these last several weeks. Steve's take on the topic at hand was that he shared my interest and enthusiasm in technology hardware innovation and invention, but thought that I might have overlooked a rather vital area of creativity in High Tech that is and likely will continue to be a mighty contributor to its growth prospects - software.

Only an absolute moron would question the validity of the observation. Software innovation has resulted in the rise of New Media giants such as Google, Facebook and Twitter, server and network virtualization, and a whole universe of new and valuable software products for the consumer and enterprise markets. The growth of vast server farms and datacenters for internal and external business uses, along with personal digital devices ranging from smartphones to desktops, all networked together across the globe, is a vast and fertile field for further creativity and development of useful software products and capabilities.

It is my contention, however, that the entire software industry is approaching the same existential threat of stagnation that is beginning to sink its claws into the hardware portion of High Tech. Let me explain this thru an illustrative example.

Let's say that I was a video game developer who has a history of developing and releasing wildly successful game titles. The key to my success is a graphics engine driving the game that rasters in great detail and beautiful colors, in addition to allowing a player to participate in a deeply immersive game world with all sorts of play options, features, character representations, item customizations, world object interactions and so on.

I decide to write a new game that includes more of everything - significantly more realistic world detail & physics, deeper textures, rastering that increases the polygon count by an order of magnitude, support for online interaction between 1000 players simultaneously and so forth.
I purchase top of the line systems from Dell, HP, Lenovo, Microsoft (who makes the XBox platform) and Sony (with the Playstation platform) and try to run the game during development and testing. Unfortunately, I find that many of the game features are constrained by fundamental limitations in the hardware.

I decide to approach the OEMs directly to work with them so that my game title can be properly supported in the market. They inform me that the only system that can support the capabilities and features demanded of my software would be a design that employed four of the highest performing graphics cards on the market working in parallel thru a customized data bus architecture, a system memory configuration twice as fast and four times the size of anything currently available with its own custom bus protocol, two specially binned versions of Intel's best CPU and a cryogenic cooling system for the platform. After adding up all the unit costs and amortizing special engineering work over volume, the retail price for each system adds up to $25,000.

After such a disappointing rebuff from the systems houses, I decide to approach a variety of chipmakers and ask them why they aren't developing products that would support such system requirements at historically normal prices. To a man, there answer is simply "We can't."

It is for this reason that chip technology needs to break out of its current Silicon straightjacket. As the foundation of all High Technology, chips are the lynchpin.

It is nearly a certainty that novel software products will be developed over the next several years that will find the current system datacenter, server and networking infrastructures with its speeds, feeds and storage to be more than adequate. But limitations in chip technology will soon start to "confine the canvas." For software to always have the freedom to improve, progress, innovate and invent, the canvas needs to be as close to infinite as it can be made.

Without major progress at the chip level, an increasing number of  software applications will stall and wither on the vine. Anything much more complex than current offerings will become too expensive and demanding to implement on a commercially viable consumer product such as a tablet, desktop or smartphone. This applies to quite a few software categories, as many of the most interesting software directions employ significant computational power, memory resources and large databases that also demand high performance and low latencies.

Long is the way
And hard, that out of Hell leads up to Light. - Milton, Paradise Lost


I think there is a great struggle ahead of us in the High Technology industry. While wonders are being developed in labs and prepared for eventual low cost, high volume manufacturing, High Tech companies - whether in chips, systems, software or a combination of the three - are entering a period where innovation will be stifled, value-add will be hard to create and maintain, and a zero-sum game of Survival of the Fittest will stress participating firms in a very Darwinian sense. What can we and our organizations do to survive and perhaps even prosper while we await the Next Big Thing?

That, dear readers, will be a topic for future posts.  ;-)