Challenging the physical limits of a computation

Researchers from Delft University of Technology, in collaboration with the University of Zaragoza, have now erased and stored one bit of information encoded in a quantum nanomagnet while measuring the amount of energy dissipated in the process.

Why do our computers and electronic devices heat our laps and drain our batteries? After all, we might think, they shuffle and manipulate only bits of information, and information seems to be a virtual entity. In reality, there is a small caveat: when information is erased, an amount of energy is unavoidably dissipated in the form of heat.

How much? A millionth of a billionth (10^-15) times the energy it takes you to lift a grain of sand to a height of one meter. Tiny, but not zero! This energy minimum is given by the second law of thermodynamics and, more specifically, by one of its corollaries: the Landauer principle. In fact, the transistors of our computers and phones consume millions of times more than that. This is because we operate them fast, in particular, faster than their internal dynamics. In thermodynamics, speed and energetic efficiency oppose each other.    

Researchers from Delft University of Technology, in collaboration with the University of Zaragoza, have now erased and stored one bit of information encoded in a quantum nanomagnet while measuring the amount of energy dissipated in the process. They obtained two important results: first, they found out that they could operate the system at the Landauer limit and that the limit applies invariably to a quantum system. “This extension to the quantum realm is an important result considering that, until now, the limit has been tested in relatively large and fully classical systems” says first author Rocco Gaudenzi. The second result is that they realized they could increase the speed of the simple computation and still maintain the dissipated energy close to the Landauer limit.

“The trick is to take advantage of the fast internal dynamics of the system”, says Rocco. “The bit of information is encoded in the quantum spin of a single molecular nanomagnet, which has a very fast relaxation time compared to any classical system due to its ‘low inertia’”. Another important ingredient is quantum tunneling: “When a transverse magnetic field is applied, the spin can tunnel between the ‘up’ and ‘down’ states through the classical barrier that separates them.” These two ingredients allow their quantum spin register to be fast and efficient. In the paper, the researchers define a new figure of merit to judge the real cost of a computation: not the speed or the energy efficiency alone, but rather their product, the energy-time cost. Considering this figure, their system performs orders of magnitude better than any existing one.

“During my first year here in Delft, I was working with molecular nanomagnets, investigating their magnetic properties in molecular junctions. Later on in my PhD, while talking to a colleague, I suddenly realized that they would be perfect to test the Landauer limit and improve the speed of classical computations.”

The results have been published in Nature Physics.