They say that knowledge is power. Now a group of quantum physicists are working to make this adage come true—quite literally. They are investigating whether the very act of measuring the properties of a tiny engine could be used to power that engine. In this case, the information gained about a machine would serve as the fuel used to drive it.

Speculation about whether information is itself a physical entity that could one day be harnessed to power a mini-machine has become particularly pressing in recent years, as technology is advancing at a rapid pace. The race is on to build practical quantum computers—devices that aim to harness the weird properties of the micro realm to outperform conventional computers at specific tasks. In addition, physicists and engineers have been drawing up blueprints for nanoscale heat engines to power the computers and nano-refrigerators to cool their chips.

says quantum physicist, Marcus Huber, of the Institute of Quantum Optics and Quantum Information in Vienna, Austria. “People are now asking: What could the smallest refrigerators be? The smallest engines? The smallest batteries?”

With these new developments comes a novel quest to find unconventional energy sources for tiny quantum engines and fridges. But the idea of using information as a fuel to power mini devices actually has a far longer history, dating back over 150 years—before even quantum theory had been discovered. In 1867, British physicist James Clerk Maxwell, was musing about the laws of “classical thermodynamics”—the science of heat and energy transfer used to describe how standard macroscopic engines work. Classical heat engines require a temperature difference to operate. They transfer heat from a hot reservoir of gas particles, or a heat bath, to a colder reservoir, with some of this heat energy diverted into useful mechanical work in the process. But Maxwell realized something curious: It may be possible to fuel a heat engine, just by knowing information about the gas particles within it. All Maxwell needed was a clever little demon to work the machine.

Demonic Devices

Imagine, Maxwell said, that you have a box with a vertical wall down the center, partitioning it into two chambers. Fill both chambers with a gas of particles; some particles will be whizzing around quickly, while others are slower. Now suppose that there is a trapdoor in the partition wall that is guarded by Maxwell’s wily demon. The demon can gauge the speeds and locations of every particle, in both chambers, and he is nimble enough to open and close the trapdoor in such a way that he only allows fast-moving particles to pass into the left chamber and slow-moving particles to pass into the right chamber. The temperature of a gas is related to the average speed of its particles and so by selectively opening and closing the door, the demon would slowly create a thermal imbalance, as the left side of the box heats up, relative to the right. This temperature difference could then be used to drive a heat engine. Power would thus arise from the demon’s knowledge.

For more than a century, the most interesting takeaway from Maxwell’s musings was that information is not some intangible ethereal entity, but a physical resource that can, in principle, be exploited to do work. Back in 1867, there was little hope of creating such a demonic device in reality however. That changed as we entered the twenty-first century, when this somewhat esoteric thought experiment began gaining practical interest as physicists could now build devices that work—like Maxwell’s demonic engine—by manipulating individual particles.

But there is a big difference between these real quantum engines and Maxwell’s hypothesized classical ones. In the nineteenth century, Maxwell had assumed that his demon could just happily sit back and use his knowledge of all the particles’ positions and speeds to decide when to open the trapdoor and when to keep it shut. But physicists now know that individual particles must obey far stranger quantum rules and that, in the quantum world, the very act of measuring a particle to ascertain its speed or location, can irrevocably change it.

says quantum physicist Andrew Jordan of Chapman University in Orange County, California. If you’re sitting in the stands watching a football match you can follow the position of the ball thanks to the sunlight shining down and bouncing off the ball and into your eyes. The stream of minuscule light particles, or photons, hitting the ball repeatedly won’t affect the ball’s trajectory in any significant way. This is not true on tiny scales, where shining a stream of photons onto a single particle to measure its location, say, could disrupt it, changing the very thing you are trying to observe. “When you learn about something in the quantum world, you must also disturb it,” Jordan says. “That’s the paradigm shift.”

Jordan has been working with quantum physicist Alexia Auffèves, from the CNRS International Research Lab MajuLab, in Singapore, and others, in the relatively new field of “quantum thermodynamics” to better understand the consequences when you put together the classical notion that information is a physical entity and the quantum idea that measuring a system changes it. “The radical option in quantum thermodynamics is basically to remove the heat baths and just to use quantum resources—to use quantum measurement alone to power an engine,” Auffèves says. In fact, she notes that “thermodynamics” may not be the best term to use to describe engines that do not exploit temperature differences. “We would rather talk about ‘quantum energetics’,” she says.

Energy Cost

So how do you set about building a working engine fueled only by quantum measurement? Step one has been to observe the energetic cost of making the measurement. With a team led by quantum physicist Kater Murch at Washington University in St Louis, Auffèves and Jordan have been examining qubits, the building blocks of quantum computers. In our everyday classical computers, information is stored as one of two binary digits or bits, 0 or 1. The power of a quantum computer hinges on the fact that their equivalent qubits can take the value of both 0 and 1 simultaneously, potentially enabling quantum computers to zip through certain calculations incredibly quickly.

In a series of different experiments, Murch has been able to use different light sources to measure the state of a qubit—and also to confirm that indeed energy is always lost by the light in the measuring process. “This can be observed by seeing that the frequency of the light has changed,” says Auffèves. “If you imagine that, for instance, you sent in light that was blue, it ends up red, so it carries less energy.”

Huber, who was not involved in the experiment, says that the team’s work to estimate energy usage is vitally important, if quantum technologies are ever to become mainstream. He notes that when we want to calculate the energy cost of a car, we could take photos of the car’s changing position, and also time how long it takes to move from one place to another. “The little battery usage of my digital camera is nothing compared to the work that is performed by the car’s engine, so I can safely talk about the energy use of the car without speaking about the energy use of the measurement apparatus,” says Huber. “But if I start measuring a single photon, then the situation is flipped on its head and most of the energy now is in the measurement apparatus—which is a situation we don’t even have the language to talk about, and why this is such an exciting area of research.”

Measurement-Powered Engines

Having established that measuring information about a qubit costs a specific amount of light energy, the second step is to investigate if part, or all, of that energetic price can be paid to power an engine. “The fact that there is a cost to measurement can be understood as the fuel of the engine because you lose energy,” says Jordan. “Then the question is: now that your system has gained energy, can you then use that energy—or recycle it—to do some useful work, for example, pushing a particle?”

In collaboration with quantum physicist Benjamin Huard’s experimental group at École Normale Supérieure de Lyon in France, Auffèves has built a quantum version of Maxwell’s demonic engine. The first part of the experiment involves measuring the system using photons and, like Maxwell’s demon, Huard’s team then uses the results of the measurement as feedback telling them how to act next. If the system is measured to be in a high-energy state, for instance, the set-up allows the system to drop to lower energy, in the process spitting out a photon, which can be harvested for doing work. If the system is measured to be in the low-energy state, by contrast, then the set-up blocks photons from being absorbed. “So we send some input light with photons, and the output light carries more photons than the input light,” says Auffèves. “That’s really an engine. “

The experiment provides proof that quantum versions of Maxwell’s demon are possible—even if this particular engine has no immediate practical use. “What’s really fascinating about this experiment is that quantum measurement can be seen as a fuel that you cannot find in the classical world,” says Auffèves.

But those practical uses could be coming eventually. “You won’t power a locomotive with these, but I would not preclude that maybe my children, or my children’s children, will be using these kinds of engines to power an atom,” says Jordan. “The exciting thing about new discoveries is that you don’t know where they’re going to go.”

Zeeya Merali is a London-based science journalist and author of the popular physics book, A Big Bang in a Little Room.

Interested in learning more about Maxwell’s Demon? Check out Paul Davies’ book, A Demon in the Machine.