How Small Can We Make a Refrigerator?

Physicist Dr. Billy Hubbard tells us how the world's smallest 'refrigerator' may lead to ways to build smaller electronics during this Pulsar podcast brought to you by #MOSatHome. We ask questions submitted by listeners, so if you have a question you'd like us to ask an expert, send it to us at

Don’t miss an episode – subscribe to Pulsar on Apple Podcasts or Spotify today!


ERIC: Is your refrigerator running?

Today on Pulsar, we'll explore how a team of scientists built an incredibly tiny cooling device that may not replace your noisy kitchen fridge, but could someday improve how we control the temperature of things on a microscopic scale.

Thanks to Facebook Boston for supporting this episode of Pulsar. I'm your host, Eric, and my guest today is Dr. Billy Hubbard, a physicist and the CEO of Nanoelectronic Imaging. Billy, thanks for coming on Pulsar.

BILLY: Great to be here, and great to talk to you. Thanks for having me on.

ERIC: So at the Museum of Science, we're always asked about the extremes, the biggest dinosaur, the fastest rocket. And you were part of a team that built the world's smallest cold thing, kind of like the tiniest refrigerator on record in the form of a thermoelectric cooler.

So why don't we start with what that device is and how it works?

BILLY: A thermoelectric cooler is a device where you kind of take two materials, you push them together, and you apply electric voltage to it. And that generates a temperature gradient. And that's something called the Peltier effect.

So with a thermoelectric cooler, you have kind of a convenient way in a lot of situations to run an electrical current and generate a decent amount of cooling.

At present, it's kind of using a few niche applications like kind of your in-car drink cooler that you can buy at Brookstone and things like that.

And it's also pretty convenient in certain scientific instruments, but it's nowhere near as efficient as your refrigerator's cooling system, for example, which is based on compressing fluids.

ERIC: So there's other better cooling methods for our everyday making-stuff-cold needs?

BILLY: Exactly, yeah. So right now, even though it's kind of big and bulky and noisy, it's kind of the best thing we have. So obviously, just making things efficient is always a good thing to do and kind of good for the world.

But what makes thermoelectric coolers particularly interesting is there's something called the Seebeck effect, which is the opposite of the Peltier effect.

So Peltier effect is you apply a voltage and you get a temperature gradient. In the Seebeck effect, you apply a temperature gradient and that makes a voltage. So in other words, you can use heat to generate electricity.

And where that gets kind of very save-the-world is when we generate electricity, we tend to lose a lot of the energy to heat. So we can harvest a lot of that lost energy due to heat. We can get much, much more efficient generation of electricity.

ERIC: And capturing that lost heat would lead to a huge increase in efficiency.

BILLY: Right. Exactly, exactly. So why did we want to make this so small?

Other than for bragging rights, there's some pretty compelling theory out there that says if you take one of these thermoelectric coolers and you're able to make it smaller, let's say, even in one dimension - so instead of a cube, you bring it down to a flat sheet - supposedly, you can get much, much more efficient thermoelectric cooling, and therefore, much more efficient thermoelectric generation.

That's what we ended up doing. We basically made thermoelectric cooler, which is more or less two dimensional.

ERIC: So exactly how small is it? And how much did that break the record by?

BILLY: The dimensions are a few micrometers by about a micrometer. And for reference, human hair is about 100 micrometers wide. The thickness is about 100 nanometers or less. And it's constraining in that one dimension that we believe will make these things more efficient.

So the thickness is about 1,000 times thinner than a piece of human hair. By volume, we're about 10,000 times smaller than even the next best candidate.

ERIC: A question we got from someone on social media about this was if it's so small, how do you know what temperature it is?

BILLY: One thing that we're able to do is you can still kind of see the device in an optical microscope. And we can actually cool down and see little water droplets form on it, like you have like a cold soda can.

There's actually a lot of interesting physics going on there when you actually cool something like this down, where the droplets show up.

That kind of convinces us that we are cooling, and it gives us kind of a range of what temperature it's cooling at, just because we have an idea of what the dew point is in the room. It's pretty tough to actually accurately measure temperature on small length scales like that.

So if you think about it, how do you measure temperature? If you take a typical thermometer you would, like, put in your mouth, you basically have the thermometer go in there. It equilibrates to your temperature, and then you get a reading based on that.

If you have this tiny, tiny cooler, you're going to get all the heat from that thermometer, it's going to go right into that cooler. So contact thermometer would never work.

ERIC: So changing what you're measuring by measuring it?

BILLY: Exactly, exactly. You might bring the temperature of the thermometer down by a nanodegree when you're getting actual cooling in the junction. So there's other ways to measure stuff, optical, for example.

You can shine light on something and see how it comes back and what's changed. But like I said, the area we're looking at is something on the order of a micrometer by micrometer.

And so you're really starting to get into the diffraction limit of optical microscopy there. So our group - and this is work I did through my thesis with Professor Chris Regan at UCLA and the Department of Physics. So that group actually wasn't a thermoelectric group.

And it wasn't necessarily even a making-tiny-things group, although it is that to some degree. Our focus really was on transmission electron microscopy and differing imaging modes of TEM.

ERIC: So can you talk a little bit more about transmission electron microscopy?

BILLY: TEM is a really critical tool that maybe a lot of people aren't familiar with. It's an electron microscope that has the ability to actually image individual atoms. And that's actually pretty easy to do on a modern system.

You can kind of go in and align for not too long, and see atoms. And it's pretty incredible. And it finds itself in a lot of applications that are kind of crucial to real world stuff.

So for example, biological imaging, we know what the structure of a lot of proteins and viruses and things like that are because of TEM. So you know those images of coronavirus you see with all the little points sticking out of it?

We know what that looks like because somebody imaged the coronavirus in a TEM. So on the biological side, TEM is one of the most crucial tools to the semiconductor industry.

So when you're making a microprocessor, you have these billions of transistors on a chip. And you need them to all be uniform and really high quality. And the way they do this is they inspect and run test structures on a huge scale.

The central part of that is the TEM, because it's really the only microscope that can see these tiny, tiny transistors.

ERIC: It sounds like it's a super useful tool in a lot of ways. How did your team utilize it?

BILLY: So TEM is great and looking at what kind of materials you have there. You can see if it's a metal or an insulator. You can get an idea of the chemistry.

Interestingly enough, where TEM kind of falls short, is it really can't tell you anything about what's going on thermally or electrically. So if you're looking at a wire in TEM, you couldn't tell if you were running a current through the wire.

You couldn't tell if that wire was charged up. And you couldn't tell if that wire was heating. Our group at UCLA was focused on figuring out how to map that kind of stuff in the TEM.

Coming back to these thermoelectric coolers, a few years ago, we developed a technique called plasmon energy expansion thermometry. And that's a way to map and measure temperature getting down to nanometer length scales.

It's a little complicated, but in short, we have a spectroscopic way of mapping and measuring what's called plasmon energy in certain materials. And that plasmon energy is based on the density of that material.

And just like in a mercury thermometer, if you can measure the density change, you can measure the temperature change by the coefficient thermal expansion.

ERIC: So stuff changing size based on how hot or cold it is. But instead of a tube of mercury reacting to the air temperature in a room, you're measuring these tiny changes in the microscopic thing itself to take its temperature.

BILLY: Exactly. So instead of using the lines on the thermometer to see how much it expands, we're using a very expensive and complicated piece of equipment.

Where that gets really cool is when you define temperature, you're usually talking about a really big collection of particles, the average energy of all those particles. And so when you get down to the nanoscale, you're getting into thousands, hundreds, even tens of atoms.

And so temperature kind of loses its meaning when you get that small. And so from the standpoint of physics, that's a really interesting world to kind of start to study.

ERIC: Kind of makes you stop and think about what temperature really means. Another question we got from a listener came from David, who wanted to know what kind of material your device is made of.

BILLY: One of the most common materials that are used in the thermoelectric coolers you can buy at the mall, it's called bismuth telluride, and then some related materials like bismuth antimony telluride and different alloys.

Bismuth telluride and its alloys have this interesting property where you can kind of break them apart and they'll exfoliate into these individual few atom-thick layers.

And this is kind of similar to graphene that people might be familiar with, which is that single atomic layer of carbon that you use scotch tape to separate.

It's near and dear to my heart because I kind of got my start in research working with graphene. We basically took the same principles you would use to make a device with graphene, where you take the sticky tape, pull it apart, and put it down on your substrate.

ERIC: So very elaborate scientific process to make this thin old scale material, just stick some Scotch tape on it and rip it off real quick?

BILLY: I wish I could tell you it was some hot shot team of scientists and million dollars equipment.

But yeah, it actually - the paper that we wrote on this, about half the authors were undergrads because this is a really good project for somebody who's just getting into science to be able to kind of sit in the lab, and kind of a quick and pretty inexpensive way of making a sample.

Literally, we bought thermoelectric coolers, commercially available coolers. We broke them apart. We crunched up the little blocks in them. Our students and also myself and other grad students, who made a ton of these devices, basically just Scotch tape, put them down on our substrates.

And that's how these devices were made.

It was an interesting project because it was something that our undergrads could really get their hands on and be able to actually make samples. And it's also something that once we get down to it, it's something relatable to people. Because in the end, we're making a cooler.

ERIC: To wrap up, museum visitors always want to know about applications. So how can we use this technology?

BILLY: A lot of times in science, you're kind of doing it just to show you do it. And so in this case, we made these devices. We showed that they cooled to just about below freezing, which is great.

To be able to make these devices alone and show significant cooling is a pretty big feat. There's a lot of difficulties that go into making these devices compatible with the electron microscope that basically makes it difficult to really optimize for these cooling properties.

So we've kind of taken the first step to make and measure these devices.

And now moving forward, we or anybody else who's interested in this, can kind of try to repeat this, but maybe make them a little more efficient in terms of the substrates they're sitting on, or maybe tweak the chemistry of the flakes and things like that.

The other potential application, you use thermoelectric coolers and a lot of scientific instruments as is, and so that we can imagine is inside the TEM, for example, it's usually very difficult to cool things down to a low temperature.

You have to use liquid nitrogen, sometimes liquid helium. There's a lot of math that goes along with trying to accomplish that.

If we could instead cool things down by just applying a little bit of electricity, even if it's just a small area, if we're studying a small system, we only need that small area to be cooled down.

So that's kind of a far and away application, but that's the first step, which is important.

ERIC: Well, Billy, thanks so much for coming on the podcast and talking to us all about this small cold thing.

BILLY: My pleasure. Thank you.

ERIC: You can visit our newest cold thing at the Museum of Science in our Arctic Adventure Exhibit, including a real touchable ice wall.

Until next time, keep asking questions.

Theme song by Destin Heilman