Before picking up the phone and calling a technician to fix a faulty microwave oven, there are always a few simple things one should check. So far, “stop looking at it” has not been part of the checklist.
For people with a special kind of imagination, microwave oven manuals contain warnings such as “do not use it to dry animals or as a storage place for books”1. For everyone else, the daily operation is basic: close the door, set the time and press the start button. Now, physicists might be tempted to add “and don't look inside or it won't work”. Writing in Physical Review Letters, Bernu et al.2 show that the build-up of a microwave inside a resonator can be blocked by just 'watching' it without absorbing the wave's energy. The edge of a very strange quantum world is now one step closer to everyday life.
Nature states firmly that information has a price. The lowest, non-negotiable one is set by quantum physics: measuring something always disturbs something else. Usually, this disturbance goes unnoticed. To detect a microwave, it is generally converted into an electric current by an antenna, or into the heat that cooks your food, so the wave is not only perturbed but destroyed. But even if you were an outstanding experimentalist and you could measure the microwave's intensity at the ultimate quantum level without absorbing a single photon, you would still leave your fingerprint by making its phase completely random. Instead of regular oscillations, the wave would make a sudden jump. This is exactly what happened in Bernu and colleagues' experiment2.
Start injecting an empty cavity with a resonant wave: it builds up and, after a short time (say, a tenth of a second), its amplitude reaches a value α. Let's see this again in slow motion. First, a small wavelet comes in. At resonance, it makes an integer number of oscillations during a cavity round-trip. Therefore, it remains in phase with the next incoming wavelet, and they simply add up. And so on. After 100 milliseconds, the wave in the cavity is made of N identical wavelets, all oscillating in phase with the same amplitude α/N. In a sense, the intra-cavity field makes N steps each with a length α/N in a direction defined by the phase of the wavelets (Fig. 1a).
However, if one now repeats the same procedure but measures the number of photons after each wavelet injection, their phases will become totally random: the field will still 'walk' by steps of α/N, but in completely random directions (Fig. 1b). As in a usual diffusion phenomenon, N steps will typically result in an amplitude of (α/N) × √N = α/√N, so the final intensity will be only α2/N instead of α2. If measured more and more often, this attenuation factor 1/N will become smaller and smaller, and eventually the resonator will simply become empty. This so-called quantum Zeno effect, which prevents a system from evolving while observed, has been convincingly demonstrated before, but usually in conjunction with other quantum effects, such as Bose–Einstein condensation or quantum tunnelling3,4,5,6,7,8. The object of interest was never as simple and classical as the intensity of a microwave in a cavity.
For Bernu and colleagues, the key to success was an incredibly good microwave resonator. Indeed, to measure the exact number of photons, everything they interact with must be under absolute control. Photons must obey two commandments: “You shall not appear in the resonator unless I send you”, and “You shall not leave it before I'm done with measuring you”. For Bernu et al. this meant building a cavity in which tightly confined photons could bounce between two superconducting mirrors about 1 billion times, cooled to a temperature of 0.8 kelvin to prevent undesired thermal photons from appearing. In addition, keeping the cavity on resonance required its frequency to be stabilized with a relative precision of about 10−11. All this led to a clear, fivefold microwave-intensity reduction after a 100-millisecond experimental sequence of alternating-field injections and photon counting.
But how can one count photons without destroying them? In physics, an old trick to measure a weak signal with low disturbance is to observe how it affects the motion of a pendulum. Just choose the right one. Back in 1846, a pendulum the size of a planet (Uranus) was used to detect Neptune. Five years later, just next door to the Ecole Normale Supérieure in Paris, where Bernu and co-workers performed their experiments, Léon Foucault used a smaller, yet still impressive, pendulum to demonstrate the rotation of the Earth.
Bernu et al. downsized the pendulum to the atomic scale. They excited individual rubidium atoms to 'Rydberg' states, making an electron oscillate between two orbits far away from the nucleus. These atoms acted as light-sensitive atomic clocks, the 'time' being given by the phase of their oscillations. When they were sent across the resonator, the hands of these clocks were shifted by the microwave photons, each photon adding a shift of 45°. The electronic state of the atoms was measured outside the cavity by ionizing the most excited ones, and the electric current created was amplified to become measurable. This last, 'destructive' part of the measurement affected only the atomic probe, leaving the microwave photons intact.
The Zeno effect cannot prevent the irreversible decay of the microwave's intensity. In an oven with a typical power of 1 kilowatt, one microwave photon is lost every 10−27 seconds, much too fast to count the photons left inside the resonator. There is thus no need to rewrite the user manual. But Bernu and colleagues' demonstration that the coherent evolution of a system can be 'frozen' by non-destructive measurements is a significant step forward for quantum control. For instance, one could continuously modify the way of measuring the system to bring it into exotic quantum states that cannot be prepared otherwise. But above all, this work is one of an impressive series9,10,11, demonstrating three essential qualities of Bernu and co-workers: expertise, rigour and a love of fundamental physics.