In 2008, Tony Tether, director of the Defense Advanced Research Projects Agency (DARPA), tapped Stanford M.D. and electrical engineer Gregory T.A. Kovacs to lead DARPA’s Microsystems Technology Office (MTO). The MTO funds engineering in five general areas: electronics, photonics, microelectromechanical systems (MEMS), computer architectures, and algorithms. But as Kovacs has repeatedly said, the role of DARPA is more about integrating these units into interdisciplinary projects. Take, for example, the Hybrid Insect Micro-Electro-Mechanical Systems (HI-MEMS) program. Researchers funded under that program are tasked with the creation of moths or other insects that have electronic controls and energy-harvesting devices implanted inside them, making them self-powered remote-controlled spies.
Other projects are more pure electronics engineering, such as the Trust in Integrated Circuits (TIC) program, which aims to verify the contents of any microchip assembled offshore. This is a punishing task, as only a few hundred transistors out of 2 billion could theoretically wreak havoc; finding them is a classic needle-in-a-haystack problem.
Not all the projects are chip-scale: Integrated Sensor Is Structure (ISIS) aims for the construction of a 150- to 300-meter-long stratospheric airship to deliver real-time surveillance of the battlefield below and the horizon all around.
Kovacs is where the buck stops and starts: He decides which projects are funded at MTO and which are slated for the trash heap. IEEE Spectrum’s Sally Adee asked him about his job.
IEEE Spectrum: With the confluence of the economic crisis, the ailing semiconductor industry, the new administration, and the sudden departure of Tony Tether, on 20 February, what is it like to be at MTO these days?
Greg Kovacs: Despite the downturn, folks are excited and energetic at DARPA. We just had 600 people at the MTO Symposium in San Jose. The theme for the meeting was ”Building the Future From the Inside Out.” That’s how we see ourselves going forward—enabling systems that couldn’t be built before or could barely be built. It’s microscale stuff enabling big things.
Spectrum: Describe MTO’s mission.
GK: Well, we make the small things that enable the large systems. Almost everything we make you could hold in your hand. We do transistors, electronic circuits, sensors; we do the MEMS and the [nanoelectromechanical systems]; we do photonics. We operate over a broad range of scales, from single photons to giant lasers. We’re probably the geekiest DARPA office. We have more horn-rimmed glasses and pocket protectors than any of the other offices.
Spectrum: What are your criteria for accepting new program managers and the ideas they bring? How do you decide what an MTO project is?
GK: They pull a lot of areas together—it’s not just about one technology by itself; it’s about looking through our portfolio of projects to address a bigger question. For example, one of our projects looks at the ability to navigate space and time without external reference. That project, Navigation-Grade Integrated Micro Gyroscopes (NGIMG), pulls all five MTO areas together to build machines you couldn’t build at any scale before.
It’s one of the more complex machines we’re building. We want to have one in our hands within a year or two. It’s a device the size of a small digital recorder, and it could fit in a soldier’s backpack or pocket. It provides precision navigation in the places GPS doesn’t work.
Spectrum: Where’s that?
GK: Consider a scenario where GPS has never worked to begin with, like in a cave or underwater. We want to build very compact systems—maybe their great-great-grandfather was a giant rack on a submarine or a big chunk of hardware on a satellite—that will give you complete knowledge of your position in space and time without any external fix.
In Afghanistan, many man-made caves are left over from the conflicts with the Soviets. [American] soldiers are crawling around on their hands and knees in the kind of claustrophobic space most people would never want to go into.
Spectrum: They actually have to go into those things? That sounds like a horror movie.
GK: Right, and they can go into a cave to root out some bad guys, and they can know where they are even if they have been in the dark in a cave for a couple of hours—know where their buddies are and maintain precise time synchronization.
Spectrum: So until now that kind of gyro wasn’t possible?
GK: It was very hard to build. People started trying to do it in the 1970s. But even if they could have built it, it would have been really big, and it would have been used to guide flying missiles.
But the point is, you couldn’t carry sizable equipment [into the cave]. Unless it’s the size of a couple of sugar cubes, it’s got no hope of helping. We now have versions of that that are very advanced and very small—small enough to be carried.
Another example is the ocean. It’s impenetrable to the radio frequencies that GPS uses. You have no geographic fix at all, and at that point, all you know is your depth. Sure, a compass will work, but if you’ve ever dived, you’re relying on a compass in murky water. It’s not very accurate.
Compass dead reckoning has the same problem that a bad gyro does: The old-generation gyros drift, meaning the longer you go without a fix, the more off you’ll be from the place you think you are. So your position relative to where you expect to be is getting farther apart the more time elapses. That’s why you need to know your drift rate—how far you’ve drifted from where you think you are. These atomic gyros hardly drift, so you don’t have to worry about those details.
Spectrum: What brought you to MTO?
GK: A call from Dr. Tether.
Spectrum: How did you know him?
GK: The Defense Sciences Research Council (DSRC), which advises DARPA informally. I had been on that council ever since I was a brand-new faculty member at Stanford. We met there. But the members of the DSRC also function as the reservists for DARPA, so periodically they point to someone and say, ”You. Uncle Tony wants you.” And that’s what happened to me.
Spectrum: How has Dr. Tether’s 20 February departure affected you in MTO? Has anything changed?
GK: For the time being, we’re just carrying on in largely the way we had been. A lot of people want to say, ”Oh, things are different, things are different”—things are not that different. DARPA’s continuing in a lot of the directions set by Tony.
Spectrum: You’ll still be directing MTO?
GK: Unless I get fired. [Laughs] Whoever the new permanent director is, whenever they get here, may change things. Historically, that’s not likely. I’ve known three of DARPA’s directors fairly well, and there’s not much precedent for that.
Spectrum: When is the new director coming?
GK: We have no idea.
Spectrum: Since the economy soured, has MTO seen any change in applications for program manager positions?
GK: I don’t think there’s been a rush or a decline. But that’s partly because we’re in D.C., not Silicon Valley. There they’ve had a downturn; here, nothing’s changed in terms of program manager applications. The one thing that is different here: All our program managers are really enthusiastic. That just keeps climbing.
Spectrum: Do you ever share your technology with the other DARPA programs?
GK: That happens all the time. The other programs—like the Tactical Technology Office or the Strategic Technology Office—are building things that go fast or go boom. [Laughs] They might benefit from what we’re doing in MTO.
Spectrum: Can you give me a specific example?
GK: ISIS is now in both MTO and the Strategic Technology Office. The giant radar at the core of the ISIS project will use thousands, if not millions, of chips built as stacks of different semiconductor materials. The technology to do the stacking is being developed in the Scalable Millimeter-wave Architectures for Reconfigurable Transceivers (SMART) program.
Spectrum: Has the semiconductor-industry slump affected MTO’s mission at all, and if so, how?
GK: Not really, other than maybe giving DARPA efforts more leverage as research funds become scarce in other sectors.
Spectrum: What kinds of semiconductor-industry-specific problems are you working on?
GK: Advanced thermal management, for example. The reason your laptop is now dual core instead of being twice as fast as it was 10 years ago is that they can’t do that anymore. They can’t just double the speed. That’s because with increasing speed comes increasing heat you have to get rid of.
Spectrum: What’s a new, promising thermal-management method DARPA is working on that no one else is tackling?
GK: We’re doing a lot. We’re investigating an entirely new type of transistor, called a tunneling transistor, which would operate at lower voltages—a quarter volt instead of today’s 1 volt. That would greatly reduce the active heat dissipation, which is proportional to the square of the voltage.
We’re also looking at reducing the standby heat dissipated when a transistor is nominally ”off” by using nanoelectromechanical switches to physically disconnect a transistor when it is off, which would prevent leakage current that generates waste heat.
We also have researchers working to reduce the heat dissipated by the interconnects that link the active devices in a chip. They’re looking at replacing some of the longer metal wires with optical interconnects, which would generate far less heat while greatly improving data transfer speed.
Spectrum: You know the semiconductor industry is suffering a lot in this recession. There are a lot of worries about where the next hotbed of innovation will be found. Can DARPA start the next generation of Bell Labs?
GK: Instead of building Bell Labs, we’re doing much more conservative things, like giant radar balloons the size of a few football fields that will float up there for 10 years. You know, safe stuff.
At DARPA, we don’t have physical labs, and we probably never will. It’s out on the street, at least in some corners, that we have our secret basement labs, with all kinds of vats boiling with colored liquids. We don’t. That would be fun—believe me, if we had that I’d be down there on my lunch hour.
Seriously though, industry has moved away from the large centralized laboratories like Bell and others. A big part of the current model seems to be to drive innovation via start-ups. DARPA funds large and small companies as well as academics. We think we can help create the technologies of the future by engaging with creative folks in all of these settings.