Classical physics tells us that space is empty, but the uncertainty principle allows particles and antiparticles to spontaneously appear and disappear, changing the structure of the vacuum in the process. In particular, a large magnetic field can cause the refractive index of space to vary with the polarization of light passing through it.

In the PVLAS experiment a linearly polarized laser beam is sent through 5 Tesla magnetic field in a vacuum, and any changes in the polarization of the beam over a distance of 1m are measured. Based on 44,000 such measurements Zavattini and co-workers found that the beam emerges with a slight elliptical polarization, and that its polarization vector is rotated by 3.9±0.5x10-12 radians (less than half a billionth of a degree).

These two results can be explained by supposing that photons interact with an as yet unobserved particle when they pass through the vacuum. For example, a laser photon may interact with a virtual photon to produce an intermediate particle that quickly decays back into two photons. This intermediate particle delays the propagation of photons that are polarized parallel to the external field, causing the beam to become elliptically polarized. The rotation in the plane of polarization, meanwhile, could be caused by a photon interacting with a virtual photon to produce a real particle that propagates away -- carrying angular momentum with it.

If the PVLAS results are indeed due to the existence of an axion-like particle, the experiment could place tighter restrictions on its mass and coupling strength. Furthermore, the result could lead to observable astrophysical effects in the vicinity of compact objects such as neutron stars. Indeed, Giovanni Bignami of Pavia University and co-workers recently proposed that the magnetic fields around neutron stars, which can be as high as 1011 Tesla, could bend light such that we would see multiple images of distant objects.

According to Bignami, such quantum vacuum lensing should be particularly visible during an eclipse of the double pulsar system J037-3039, although the next such eclipse is not expected until about 2020. In the mean time, the PVLAS team is continuing to scrutinize its lab-based version by repeating the experiment with a shorter-wavelength laser.