NATIONAL HARBOR, Md.—Can a government agency capture the explorer's urge of Christopher Columbus, Neil Armstrong or James Cameron? That's what the Advanced Research Projects Agency for Energy hopes to do, in the words of Deputy Director Cheryl Martin, who is currently heading ARPA–E.

The mission of the agency is simple: “to advance high-potential, high-impact transformative energy technologies that are too early [in their development] for private investors,” Martin said at the agency's recent summit here. The goal is to enhance U.S. economic, environmental and energy security or, as Martin put it: “to think big, to think bold, to think differently in order to discover new paths, accelerate technology adoption and, ultimately, change the way our country produces, transports and consumes energy.”

The world is actually headed in almost the exact opposite direction. Whereas the use of renewables for electricity has quadrupled since the turn of the 21st century, the demand for coal grew 10 times faster. So the mission of ARPA–E remains critical if the U.S. and the world are serious about combating climate change and solving challenges like energy poverty. After all, the planet has to prepare for the emissions of 10 billion people by 2050, all seeking the benefits of using more energy.

The key is what ARPA–E seems to offer: a vision of a better future rather than some energy-scarce, climate-catastrophe dystopia. "We have a moral responsibility to the most innocent victims should climate change occur," said Steven Chu, the departing secretary of energy at the summit. “Decades from now, we don't want our children to ask: ‘What were our parents thinking? Didn't they care about us?’” Yet ARPA–E also seems to have begun to shift focus from more speculative efforts, such as using microbes to turn electricity into liquid fuels in favor of efficiency gains in electric transmission or enhanced use of natural gas for vehicles, perhaps in part because of the focus on a viable path to commercialization.

Thus far, ARPA–E has invested some $770 million in 285 projects. Scientific American sat down with Martin during the recent ARPA–E annual summit to find out how the agency has changed and how it assesses its accomplishments. After all, as Martin noted, "it took Christopher Columbus three funded projects to find the New World."

[An edited transcript of the interview follows.]

ARPA–E is now heading into its fourth year, what constitutes success for your projects?
In every piece of energy, whether it's a battery, wind, magnets and motors or fuels, the path to market is different. Every piece will need some kind of handoff. ARPA–E is going to finish its three years of time—So what happens at month 37? Is the technology early enough that more research has to happen? Is there another piece of Department of Energy funding for that? Or is it the scale of a first prototype in a Department of Defense installation to get feedback from a forward-operating base test? Or is it the time to say a small company does spin out of this research? Small companies won't make sense for all of these projects, in some cases you need a strategic partner [a large company that invests in the technology for its own business purposes.]

ARPA–E has had some failures along the way, including everything from algae projects to a water-based air conditioner. How do you deal with them?
If it's part of the plan, I don't know if it's a failure. There are projects you stop. There have been about a dozen of those where the science didn't work out like we thought it would. But we learned what doesn't work.

We've stopped programs from universities to big companies on mutual agreement. Life is short, let's work on something that matters. It's part of the plan; it's not personal. In an analogy with exploring, you go to some places and it doesn't go the way you thought it would, but that doesn't mean we didn't learn something.

Were there particular areas of research that proved more challenging?
It's not any particular segment. Coming from the chemical industry, we stop projects all the time.

One of the big challenges for ARPA–E may be identifying the customer for this research. For the Defense Advanced Research Projects Agency [DARPA], the customer is the military. What about for ARPA–E?
To some extent, we have the same customer as DARPA. The military is such a large user of energy, and we have good engagement with the military already. So, for example, the BEETIT Program was about energy efficiency for air-conditioning. That's a big area of focus for the military. The places we tend to have military establishments are hot and humid.

Especially for the Navy guys, the size of the hulls are pretty much set. There are more and more power demands that have got to fit those footprints. Greater efficiency needs to happen to fit these things. So a project like Virginia Tech's voltage-regulator chip that will shrink voltage regulators by 10 times gets really exciting. Technologies are also starting to be big enough to be real and be tested by an entity in the electric industry, for example the Tennessee Valley Authority testing Smart Wire Grid's power-flow controller for transmission lines.

Has ARPA–E moved from "home runs" to more incremental improvements like energy efficiency in air-conditioning systems or power converters? Or better batteries rather than “electrofuels”?
A 50 percent improvement in how much fuel you need to run an AC is not incremental, but it is not sexy. Some of these other things are wild and crazy. The question is: What's the impact? Does anybody believe that's possible?

Once you demonstrate that it's possible, then the world changes. Once we knew the world was round, well, when was it ever not? Once a runner broke the four-minute mile, then, of course, lots of people do it. We'll all start to think these things are possible.

New technologies and discoveries have enabled companies to extract more oil and natural gas in the U.S. at levels not seen in decades. Can alternative energy survive this fossil-fuel boom or is all the research going to go into things like better tanks in vehicles to enable them to drive on natural gas?
There's things about gas itself that offer a big opportunity. Some of this gas is remote. It's burned [flared] in a lot of places because it can't get to a pipeline. Innovation can help to make it all better, like biological or chemical conversion of natural gas on site.

Look at our mandate: it's economic, environmental and energy security writ large. There are pieces of the natural gas puzzle that can be helped by understanding where it is and making sure it doesn't end up as a greenhouse gas that we want to help with. There are ways it might be used productively for U.S. energy and economic security.

There is no one answer, and being tied to just one thing is never good, it's just not. There are places where it's just not easy to get gas to and having options in these spaces provides security and balance.

It's interesting you mention greenhouse gases. Why was there so little mention of climate change at this year's summit? Is global warming still a focus for ARPA–E?
It's one of our mandates. That's why we do carbon capture. I don't think it has ever gone away.

Many of the initial projects and staff have come to the end of their three-year terms, including that of the first director, Arun Majumdar. How do you maintain the institutional memory and make sure that what was learned stays learned?
It's part of the model, these limited terms. It's part of the model to bring in people who are passionate, have an idea about what they want to pursue and bring a community together to solve a problem. So, for example, we have done a lot of battery research, both BEEST [Batteries for Electrical Energy Storage in Transportation] and AMPED [Advanced Management and Protection of Energy-Storage Devices]. But we just launched RANGE [Robust Affordable Next-Generation Energy Storage Systems]. We've gone from battery energy density to the whole car, and we can now rethink battery shape and what's chemically possible. We get that because people come in and look at the space differently.

It's hard to do it well. How do you capture that knowledge from successes and failures and pass that through? I've been chatting with the military because they deal with turnover every three to four years. DARPA has done this. ARPA–E is going from a start-up to a company: How do you retain innovation as you also mature and have a history that you do have to remember? Our technical support staff is not on the same rotation—that helps a lot. It also gets into how do you document? Should we use our own Web site to report on good and bad? We don't want other people to go down paths that we feel we have data on to say, "Don't do that."

I joke with Arun: "You had it easier, because I have to remember everything you said."

So that brings us back to the definition of success. When will we know if ARPA–E is a success?

Ultimately, it's, do we actually see a large-scale impact? Do we get the Internet of energy? How long has it been since ARPANET [DARPA's precursor to the Internet]? Fifty years from now, in 2060, someone else will say if we were a success. That's the ultimate metric.