Productive Nanosystems Panel: Applications

Chris Phoenix is providing live blog coverage for us on all the presentations from an important conference on Productive Nanosystems: Launching the Technology Roadmap...

Panel abstract/topic:

Work toward productive nanosystems results in new commercial applications at virtually every step. The increasing ability to control matter to atomic precision enables major leaps in power generation and storage, computation density and efficiency, high performance sensors, and materials for aerospace that outperform past achievements by surprising factors. This panel will explore the possibilities from near-term and practical to longer-term and visionary.

Panelists:

Malcolm R. O'Neill, former CTO, Lockheed Martin; and Chairman, Board on Army S&T, The National Academies
J. Storrs (Josh) Hall, Research Fellow, Institute for Molecular Manufacturing
Papu Maniar, Advanced Materials and Nanotechnology Manager, Motorola
Thomas Theis, Director, Physical Sciences, IBM Research
Moderator: Pearl Chin, President, Foresight Nanotech Institute

[This is a near-transcript -- yes, I do type that fast.]

We're starting with presentations. First, from Josh:

We'll remember the 19th century for the Industrial Revolution. Newcomen steam engine, built just about 300 years ago. The size of a 3-story house, consumed massive amounts of coal. Eventually, with the contributions of James Watt, they began to take over from water wheels. It took almost 100 years to break into a significant paradigm shift position in technology (from vacuum-driven to high-pressure-steam driven) which enabled locomotives and required machine tools. There was a synergistic effect from more than one segment of the economy which created a never-before-seen economic mode.

The racing chain saw (used in lumberjack competitions) has the same horsepower as the Newcomen steam engine, but it's handheld. Horsepower per pound forms more or less an exponential trajectory for 300 years, through steam engines, gas piston, gas turbine... the curve projects molecular power mills at 10^15 watts per cubic meter in 2050. [There's a missing engine technology in the curve, starting about now; if I get a chance I'll ask Josh what it might be.]

Moore's Law. Cray-1: $7M for 133 Mflop. We got 52,600 more op/$ in 30 years; a dual quad Xeon costs $5k for 5 Gflop... and we use it to play Solitaire.

In 2030, will I be able to afford things that cost $7M today? Airliners, factories, hospitals. Can I carry in my pocket things that weigh ten tons today? Houses, trucks, construction equipment. Josh pulls a memory stick out of his pocket--ten times as much memory as a ten-ton memory bank that served a college in 1976.

Nanofactories may enable this. (Don't forget the 2,500 ton aircraft carrier from David Forrest's talk.)

Malcolm:

Aerospace is interested in atomically precise manufacturing: Lockheed is spending $20-25(?) M per year on nano. APM promises a fundamental change in how we think about making things:
Smaller volume, lower weight, potentially lower cost, stronger lighter materials, higher energy propellants, higher performance reliability capability and quality.

Define APM: Lots of different definitions. Lots of implications. We can make anything we want to make with any properties we can get out of the best materials. When going to Mars, a factor of 50 weight is the difference between success and failure.

Payoffs in current products, and also yet-to-be-invented products. Lighter weight. Molecular sensors. Smart clothing. Beating Moore's law. (And more...)

In the DoD environment, it's hard to think outside the box. "To some of my friends, graphite epoxy is just black aluminum." Then you need lab tests andprototypes. Then you need demonstrators. 'Show me.' Then upgrade parts of existing systems. Finally, new baseline designs. But this is slow, because a bad piece of equipment can cost lives and national security. So APM technologies will come through commercial applications to the military. In national security, our computers are typically a generation behind.

Long term payoffs means partnerships are needed: industry in close alliances with universities and government labs, primary developers with small startups, to make sure manufacturing gets proper attention and reliability, environmental requirements, etc., are being met.

Tom:

Six years ago, he was asked to tell design automation people what would happen in 50 years in nanotech. So: first 10 years, business as usual, dealing with ever-increasing complexity. Increasing use of synthesis and self-assembly. Organic electronics in niches. Increasing integration of heterogeneious functions (sensors).

10+ years (from six years ago--prediction): Chemically synthesized nano-building blocks replace semiconductor logic and memory devices. Result: Increasing emphasis on redundancy, test-and-repair, and self-repair. We may see little bits of this.

20-50 years: Increasing use of hierarchical self-organization (whatever that meant). IT systems approach biological levels of complexity. (A requirement for APM systems [I disagree--CJP].) We have no clue how to design and verify such systems. Even back then, we were designing self-repairing memories.

Chip-building is the "new industry" and will absorb whatever advances come along.

How big can information technology get? It's 10-12% of the economy at this point. Back in history, people wondered how we could become a manufacturing economy--what would people eat? Today, 3% of economy is agriculture. In future, 98% of economy may be information technology, based on nanotech, probably APM.

Papu:

Nano in mobile devices: electronics, storage, antennas, power, biometrics, camera...

Latest technologies have to be low cost, high volume, quick to commodity. You don't know which phone will take off, at which point you have to make 10,000 to 15,000 phones per day. In 2006, 2.2-2.5 million cell phones were sold per day. By 2010-2011, we'll be selling 5 million phones per day.

How we do R&D has changed: Ideation to acceptance to commercialization. Ideation takes proof of science to proof of technology. One of a kind isn't enough any more; proof of concept has to include repeatable, scalable, six-sigma. Finally, R&D equipment has to be matched to older OEM equipment (transfer). Tech prototype has to generate a product prototype; this is proof of value. Product prototype = 300-3000 units. It has to be not only manufacturing qualified, but suited for high volume manufacturing.

You used to have concept IP. Now you have technology IP, which takes it from proof of concept. And manufacturing IP, which takes it from proof of value.

Nano specific challenges: Ideation: Value is application specific. Proof of sciencde is necessary but not sufficient.
Proof of concept and transfer: Very long cycle times to make repeatable setups, because you're forcing nano processes on established equipment.
Proof of value: Value is diluted because product isn't optimized; supply chain isn't ready; risk isn't worth it. Maybe 1 out of 10 nanomaterials makes it into a product.

Nano introduction timelines: 1-3 years: Housing and displays; 3-5 years: energy, storage, RF; 5-10 years: energy, RF, wearables; 10+ years: nano circuits, flexibles.

Pearl: What do the panelists mean by near-term, long-term, visionary?

Tom: Near - evolutionary enhancements on existing tech; visionary: what would be really revolutionary is mfg technologies and machines operating at thermodynamic efficiencies. Precision is a done deal, scaleup has to be worked on, efficiencies depend on ...

Papu: Short term 1-3 years. Long-term 3-7, anything beyond 7 years we have no clue. Cell phones won't look like cell phones in five years.

Malcolm: The chart that Papu showed is exactly the way we see it in aerospace; except we buy rather than make. Investors want to see 3-5 years out. So that's were we use internal funding. Up to 10 years, government funding.

Josh: I and several other nanotech people were invited to the Foundation for the Future in Seattle in 2000. They asked me to talk about what the future was going to be like. They said, "Tell me what the year 3000 is going to be like." I was floored. A thought about design automation: When I was a postdoc in the 90's my group wrote an AI program that could design a complete pipelined microprocessor given a description of the instruction set. That kind of thing is getting better as time goes by. Design and other parts of information economy are moving at Moore's Law growth rate. Right now, APM has one foot in the digital world of the atomic precision, one foot back in the old analog world of the industrial revolution growth rate. Key challenge is to get the synergies right to move the whole thing into the digital growth rate. I discovered that doing things in the real world is a lot harder. When I wire up microprocessors, they work right the first time. When I wire up motor controllers, which are much simpler but carry 500 amps, they blow up. As we go digital, the opposite will happen, which is why digital has the accelerate growth rate that it does.

Malcolm: One of the interesting technologies I don't think you talked about was meso-atmosphere. From 50,000 feet to 100 km, where a satellite can stay in orbit. That's a tremendous range of altitude that no machine occupies. We're exploring that through nanomaterials: lightweight fibers from Akron; power generation and storage; conformal antennae; lightweight materials. That's a system, mission, capability that would be disruptive, revolutionary, all of the above. You wouldn't have to put things in orbit; you'd be above the jet stream; you could stay above one point on the ground virtually forever.

Papu: Motorola has worried about cell phones and mobile comm will change in the next 3 years. Trends we worry about: variables in general, and whether current architecture will be distributed or stay unified.

Tom: My focus is on devices and IT: storing, processing, communicating information. Something that's about to happen in a big way, and most of you aren't aware: there's been a precipitous change, a tremendous increase in the rate of decrease of per-bit cost of solid-state memory. Later this year, Samsung will introduce the world's first large-capacity phase change memory. These can be scaled to the few cubic nanometer range without running into fundamental problems. What's already in the labs tells me this trend of decreasing cost of memory will accelerate or at least continue. We'll have attributes of products that we don't have today. Hard drive companies are all panicking and looking for other business. What needs to happen is something has to replace the silicon transistor. That's nano. That's simply not there; there's lots of handwaving, e.g. non-charge-based devices, spintronic, plasmonic; but everything that's in the lab doesn't have the capability of doing better than the transistor in terms of performance/power. Clock speeds saturated because we can't run faster without using too much power. To move forward, IT has to figure out how to make things work in a nearly reversible fashion. Today, each bit erasure dissipates the bit's energy. If we can develop reversible devices, then processing technologies can go much further, and peta or exaflop devices could fit in your pocket. If we're stuck with the transistor, then we're stuck with 1-5 GHz computers for as long as I can see.

Josh: Is anyone still working on optical computing? (Something about power being too high.)

Tom: We analyzed and never went into it. Plasmonics is the new thing: light couples to smaller waves, lets you miniaturize the devices. So far I haven't seen a device that's better than transistors.

Pearl: Any questions from audience?

Audience: Once we start working with cellular sized machines and we have nanoscale computing elements, what conceptual bridges do we have to cross to create a neural computer interface?

Josh: A lot of software.

Tom: I won't answer that, but the fundamental problem with neural anything is that we don't understand the algorithms the brain runs. We can do things that a computer can't do no matter how long we give it. We basically know that most of what the brain does is pattern recognition; we (meaning the neuroscience community) know this can be mapped to Bayesian inference; trouble is, that's NP-hard; we don't know what approximations the brain uses. And remember the brain does this with only a few (~10) logical operations. If I knew the algorithms, I could have a group implementing them; so you're asking for a breakthrough in algorithm, not nanotech.

Malcolm: [??] will do two studies in 2008; one is neuroscience; there's some very interesting work going on, trying to figure out how the brain waves couple into thought and actions; this is something the National Academies are working on.

Josh: The [??] group at MIT has an architecture that's as good as humans at recognizing dogs.

Audience: What do each of you see as the most important technological development in the next ten years in nanotech?

Josh: I don't know.

Malcolm: I'd hope it would be a fundamental understanding of the potential of APM. Without having achieved it, but at least understanding where we need to go, where to invest, where's the low-hanging fruit.

Tom: That device I described, the device that can exceed the ultimate performance of the transistor, that's most important for IT.

Papu: Mobility. With mobility, we get convergence. Power: We need a portable form factor. Also, we need a display bigger than the device. Third one: health-related diagnostics: mobile device for personal health. That goes to medical/bio sensors.

Tom: Outside IT, and maybe more important than anything in IT, getting the cost of photovoltaics below coal could be biggest. That could happen around 2015.

Pearl: Want to thank every panel speaker.

[... And that wraps up the conference! It's been fun...]

Chris Phoenix

Tags: nanotechnology nanotech nano science technology ethics weblog blog

Comments

[Hal]

Thanks for blogging this, Chris. After reading through most of these postings however I am left confused as to what this "roadmap" is. Is it a specific document? A diagram? A map? Or was this conference itself, the collection of all of the presentations, considered to be the roadmap?

[Chris Phoenix, CRN]

The roadmap is a document. It describes a bunch of nanotechnologies, and calls for developing them in the direction of molecular manufacturing. It was developed by Battelle and Foresight, with lots of researchers participating, including Drexler.

Chris

[brian wang]

Is roadmap document going to be available online anywhere ?

[Pearl Chin]

Here is pre-release version for the Vision Weekend: http://www.foresight.org/SrAssoc/2007/wiki/index.php/Atomically_Precise_Construction_Productive_Nanosystems_Roadmap

[Will Ware]

The Roadmap document is now available online:
http://e-drexler.com/p/07/00/1204TechnologyRoadmap.html