Professor Michael Spencer recently met with ECE Director Clif Pollock to discuss the devaluation of hardware and how a software-focused generation is changing the way students learn. They also discussed the ways in which the department can contribute to research improving high voltage electronics.
Pollock—I’m impressed with what you’re doing inside your Advanced Materials and Devices Lab in Phillips Hall, in particular how you and the school have discussed and chosen a path in electronic materials and devices.
Spencer—Hardware currently goes against the grain of where many people are investing. The conversation we’ve been having inside and also outside of Cornell is about the role of physical electronics in a world which is dominated by software. There’s a lot of short-term gratification because you do some interesting things with software and create an app that will define something new. But it doesn’t have the permanency of a physical electronic. It doesn’t often change the paradigm.
Pollock—Yeah. Software development doesn’t require anything close to the investments needed in time and capital that go into device development.
Spencer—I’m reading this really interesting book by Howard Gardner and Katie Davis called “The App Generation.” They’re asking: Are these apps a boon for us to move to the next level, or are they somehow restraining creativity? One of the suggestions is that students from this App Generation take fewer risks, and they think all of the answers are on the Internet. It explores this idea of Googling everything, and I think it’s interesting when you start thinking about what that means for engineering students.
Pollock—You’ve mentioned software and apps. We use CAD software all the time but it’s like your apps, it can do a certain task and nothing more. Does that constrain us?
Spencer—To the extent that the software forces you to use the package by creating standard constructs, I think it does place constraints upon us. If you’re analyzing a laser, you have to have a multi-level system. It has population inversion, reflectivity at the end and then the software is going to crank out efficiencies and so forth. But you’ve already said that the only way you’re going to make a laser system is with a multi-level system with reflectors at the end. It precludes you from thinking about the possibility that it might be a Plasmon laser or something that doesn’t use a multi-level system. It potentially constrains you in one specific direction because it boxes you into this way of thinking.
Pollock—Things have changed in the span of our collective careers. When I walk down the hall today I see students sitting in front of a computer. When we were students, our desk was where we stored our books and we spent our time in the lab. Now it’s reversed.
Spencer—When I was a student, we would spend a lot of time in the library stacks. We would look things up. But it also became a kind of place where you could run into people and have interactions. The very act of looking things up required you to physically move from place to place. And so, I think there was more of a forced social engagement rather than now when you can put on a pair of headphones, sit down at a computer, and access almost any document you need.
Pollock—So are we better or worse off? In spite of all the hands-on experience I developed, today’s students do a lot more design work than I ever did. With computers, they can design with a thousand times more complexity than I ever could.
Spencer—Yeah, we’re getting a lot of really clever designs. But how much stuff is really shifting the paradigm? There are some very fundamental things happening, but the questions is: Are there more or less fundamental things happening now than before? Is it constrained by something else or is it truly the way people are approaching things?
I have a question for you, Clif: What is your vision for the department? I know one time you said people were always questioning you about what we’re doing in energy. Where do you see the impact?
Pollock—One ECE area in energy that intrigues me is high voltage electronics. You can’t buy an analog chip nowadays that isn’t programmable because it’s full of switches, which are cheap and easy to make. So, when the power network becomes switchable, I think all the smart grid stuff is going to expand dramatically. But, until we have the technology to switch and do it quickly and cheaply, I’m not sure it’s going to happen.
Spencer—The question about the grid—Cree* now has demonstrated a silicon carbide at 20-kilovolt device and it costs a lot of money. If you have a grid device, even if you make it work, how many are you going to need?
Pollock—Not too many. Maybe a dozen in a town like Ithaca?
Spencer—Maybe a little more than that. But it’s not like an automobile or solar inverter where there are a lot of them. What company is going to go into that market? And that’s at the highest energy.
My first grants were in solar cell technology. Solar cell power nowadays is really coming from standard silicon devices. The price of solar energy is now dependent upon the cost of polycrystal silicon and China is controlling the cost. They’ve made it artificially low.
There’s not a lot of fundamental research, in my opinion, in an area that’s going to make an impact on something that is so cost driven. It’s almost getting to be a commodity. In terms of cell technology, I just don’t see us being able to do much to move the needle.
In terms of grid technology, that’s also an interesting question. Cree has been doing higher and higher voltage devices, slowly. It’s a slow development. The big market where they’re now finally producing devices is in silicon carbide. The big market has always been automotive, where the industry has never really been committing.
Pollock—You’re referring to the electric car?
Spencer—Yeah, that’s the big one. The power device market is in two places—600 volts and 1,200 volts. Twelve-hundred volts is right about what you need to power an electrical car. That’s where the high-power switches really have an advantage of silicon. The problem is the automotive market place is incredibly difficult to penetrate. So the only place where the power switches have been able to survive is solar inverters. That’s it so far.
Pollock—So this is a DC to AC conversion?
Spencer—DC to AC, so the power company buys back your energy. The other one that may save them is downloading power to quick-charge cars. For those remote stations, you need a high-power semiconductor device to be able to charge many amps very quickly.
But probably the biggest energy efficiency argument is this one: In our computers there’s an energy cost of going from zero to one. The switching cost. Our brain switches about two to three magnitudes of order less than that. Where do you see this anywhere else?
So if you actually go to the Google server, the power required for that is a building about the size of The Statler Hotel. It’s all wasted heat. So if you can solve the switching problem, you actually make a huge contribution to the energy situation.
Pollock—And that’s by just reducing the power. There’s a lot of opportunities for us. ECE could have as big of an impact there
as anyone.
*Cree, Inc. is a multinational manufacturer of semiconductor light-emitting diode (LED) materials and devices, with its headquarters in Durham, North Carolina. Most of its products are based on silicon carbide.
