The coldest place on Earth: Inside Bluefors dilution refrigerators

Ever wondered how to make something cooler than anything else in the known universe?

That’s the question that greets visitors to the website of Finnish cryogenics company, Bluefors. Their answer: take that something and place it inside one of the company’s dilution refrigerators. Once active, the central compartment of the fridge will chill anything inside to (a fraction of one degree Kelvin above) absolute zero. Absolute zero - the lowest possible energy state for matter; the coldest it is possible to get - would (theoretically) be achieved at -273.15 degrees Celsius. Bluefors’ dilution refrigerators will get you to below 0.01 degrees Celsius of that.

In both senses of the word, this is indeed very cool. But when the first dilution refrigerators were produced a little over 50 years ago, they were expensive answers to questions that very few people (mostly in low-temperature physics laboratories and materials research centres) were asking. With the advent of Quantum computing, the company has found a new, growing and lucrative market.

“Until about ten years ago, dilution refrigerators had [only] been found in rather specialised low-temperature laboratories,” says Dr. Rob Blaauwgeers, Bluefors CEO. “There wasn’t much commercial use for them, even in the scientific community. When we started the company, we just wanted to make the refrigerators more available, more reliable and easier to operate. A lot of my PhD work was on just managing dilution fridges and keeping them ‘alive’. We didn’t have Quantum computing in mind. Quantum computers weren’t even on the radar.”

The applications for Bluefors’ fridges at the time were, Blaauwgeers admits, pretty niche: specialist science departments, helium researchers and various detector projects that required “very low thermal noise.”

“We weren’t thinking about producing very big volumes for commercial applications,” he says. “But as we started the company, the field of Quantum computing started to grow - and they had the need for low temperatures. We really grew with that field. The focus of the company changed.”

As a master’s student at Leiden University in the Netherlands, Blaauwgeers didn’t initially see a future for himself in low-temperature work. Originally, he wanted to be a theoretician. But the program at Leiden (a key player in the development of the world’s first dilution refrigerators) critically (and serendipitously) required students to perform units of practical work to complete their degrees. Blaauwgeers decided to work with the low-temperature physics group studying the properties of helium - an experience that radically changed his trajectory and put him firmly on the road to founding Bluefors.

“It was very hands-on,” he says of his master’s work. “Traditionally this was a field of physics where you couldn’t buy equipment commercially, so a lot of it was just built by the people themselves… I fell in love with the engineering of it: it changed my career at that point and I decided, ‘this is what I really want to do’.”

Blaauwgeers went on to complete a PhD in low-temperature physics, where he worked experimenting with and upgrading the dilution refrigerators at the University of Helsinki’s Low Temperature Laboratory. In 2008, he and academic colleague Pieter Vorselman, a cryogenic engineer, founded Bluefors (a portmanteau of its founders’ last names), using their combined experience to build commercial dilution fridges for what was then still a meagre market Blaauwgeers recalls one early success supplying one of the company’s machines to a project that needed near-zero Kelvin temperatures to measure background radiation in space using telescopes. Initially, he envisioned the company making perhaps ten dilution fridges per year. Today it makes over a hundred - and its principal customers are building Quantum computers.

The Quantum computing companies to which Bluefors caters require extremely low temperatures because of the extreme sensitivity of qubits to environmental disturbance (including thermal radiation). But controlling the qubits inside the fridges presents a challenge: how does a person (at room temperature) interact with a qubit which (at a little under -217 degrees Celsius) decidedly isn’t?

To answer that question, Blaauwgeers first explains the construction (and in effect, the barriers between operator and qubit) of a Bluefors dilution refrigerator. 

“In the simplest terms, it’s just pumping a liquid around a space that gives you cooling. That’s something people can relate to; it’s how your refrigerator at home works,” he says. “But then, of course, when you go to very, very low temperatures, everything freezes. So there are no liquids to pump anymore. But we’re very lucky, because liquid helium stays liquid at absolute zero, so you can make use of it as a coolant.

“Basically that’s a dilution refrigerator: you’re pumping liquid helium through a circulating system which has many layers of insulation. It’s really like a Thermos flask: you have layers and layers of shielding at different temperatures.”

The challenge for companies using dilution fridges to house their Quantum computers is how to penetrate those layers to tell the computer what to do, and then see what it does, without compromising that virtually static environment. Essentially: how do you drill a hole in a Thermos flask without changing the temperature inside?

“Quantum computers have to cool their qubits to very low temperatures,” says Blaauwgeers. “But on the other hand they also have to control the qubits, so you have to have measurement lines going from room temperature to the very low temperatures, and there you have the same problem: it takes a lot of engineering for measurement leads or read-out leads or wiring, in general, to go down to these low temperatures - especially as you increase the number of qubits. If you have, say, ten lines going into the dilution refrigerators, that’s OK. But it you have tens of qubits and you need several lines per qubit, now we’re talking hundreds of lines which starts to overload the system. Of course, Quantum engineers want even more qubits, so soon we’ll be talking about thousands of lines. 

“So,” he says. “There certainly are challenges!”

Scaling up Quantum computers presents enough challenges as it is. But in the case of cooling, it’s the lines to the qubits that will, for the time being, cause a parallel scaling-up of dilution refrigerators. With relatively small arrays, that’s a functional approach. But also one that, in the end, is not sustainable.

“The qubits themselves take up very little space, so of course if you want to go many orders of magnitude up in the number of qubits, you can [easily] fit them into your refrigerator,” Blaauwgeers says. “Actually, if you look at it over time, the refrigerators have only gotten larger and larger essentially to facilitate the control lines to qubits.

“Of course, that becomes a problem, too. One of the things we’re dealing with now is miniaturising these lines. That’s one way to go forward. But at a certain point that becomes quite a brute force approach and you’ll most likely have to go to multiplexing.”

“Multiplexing” in this context means being able to address, control and receive readouts from individual qubits without each one requiring a unique set of lines. It’s an answer to the looming problem of linear qubit scaling.

“If you want to control a qubit, that requires three lines into the refrigerator,” says Blaauwgeers. “So, if you have two qubits you’d need six wires. If you had three qubits you’d need nine wires... At some point, if you’re talking about thousands of qubits you need thousands of wires - which obviously becomes a problem. So instead of having physical wires, you could have a multiplex. Basically, that’s a device where you can set an address which then opens a switch to a certain qubit, [to which you] then you send the information. Then if you want to go to another qubit, you set the device to a different address. So you need many fewer lines to control the same number of qubits.”

Any line running from a lab computer to the inside of a dilution fridge is an engineering challenge. Mitigating those challenges - of noise, of heat - is an ongoing process. And yet, for a company founded just a little over a decade ago with no thought to Quantum computing, Bluefors has faith it can be done. So much so that the company has just purchased an adjacent building to convert into a production line that will fill 7,000 square metres.

“I don’t think anyone really doubts that Quantum computing is going to work,” says Blaauwgeers. “It’s more a question of how long it will take.

“We are expanding our R&D. We’re expanding everything. We want to be making 250 to 300 systems per year. So the goal for this year is to get that new factory up and running. We’re not sitting back and relaxing. We’re really preparing to ramp up production over the next year.

“We excited about what the future holds,” he finishes.