On August 19, 2026, *Nature* published an international collaboration reporting the first possible astronomical observational evidence of vacuum birefringence — a quantum electrodynamics (QED) prediction that has waited 90 years for an answer. Using NASA's IXPE satellite, NICER on the International Space Station, and Australia's Parkes radio telescope, the team studied the magnetar 1E 1547.0-5408 in the Milky Way.
Classical vacuum and QED vacuum are not the same thing
In classical electrodynamics, vacuum is perfectly symmetric: light passes through without any change in polarization.
QED paints a different picture:
> Virtual electron-positron pairs constantly appear and vanish in the vacuum. In an ultra-strong magnetic field, these virtual particles become "aligned," creating a directional environment. As a result, light of different polarizations travels through the vacuum at slightly different speeds — the effect resembles light passing through a birefringent crystal.
This concept of vacuum birefringence was derived in the 1930s (the Heisenberg-Euler effective Lagrangian) and has been in textbooks ever since, but it lacked observational evidence — laboratory magnetic fields are far too weak. The only environments that can produce a detectable signal are compact objects with extreme magnetic fields.
Why magnetars
Magnetars are the most strongly magnetized class of neutron stars, with surface fields up to 10¹⁴–10¹⁵ gauss — 6 to 8 orders of magnitude beyond the strongest laboratory fields. The target, 1E 1547.0-5408, is a representative example.
The team combined three complementary instruments:
- IXPE (Imaging X-ray Polarimetry Explorer): NASA's imaging X-ray polarization detector;
- NICER: the Neutron star Interior Composition Explorer on the ISS, for X-ray timing and spectroscopy;
- Parkes: CSIRO's radio telescope, for radio polarization.
- The polarization angle rotates with the magnetar's spin;
- The polarization direction and the radio emission jointly trace the magnetar's large-scale magnetic field;
- Standard "surface radiation + ordinary magnetosphere" models cannot explain such a high, field-locked polarization.
- Other processes around magnetars (magnetospheric structure, surface emission anisotropy) need to be further excluded;
- Deeper IXPE observations plus improved simulations are needed to quantitatively separate the vacuum birefringence contribution from the magnetospheric one;
- Whether other magnetars reproduce the same 65–80% "vacuum-locked" polarization pattern.
- QED has its first astronomical-scale empirical support in the strong-field limit;
- The vision of "the vacuum as a particle-physics laboratory" gains a measurable window;
- The next 5–10 years are critical for "vacuum birefringence astronomy" to take shape, with more magnetars and magnetized accreting neutron stars likely to be added to the observing catalog.
- Can NICER + Parkes reproduce the "vacuum-locked" polarization pattern in more magnetars within 6–12 months;
- Whether the IXPE team will formalize this methodology into a standard procedure;
- Whether the "vacuum polarization spectrum" could become a third information channel in multi-messenger astronomy, after gravitational waves and neutrinos.
Only together could they align "X-ray polarization + radio polarization + timing" on a single source.
What IXPE saw
IXPE measured an average polarization degree of about 65% at 2 keV, approaching 80% in the 2–3 keV band at certain rotation phases:
Vacuum birefringence explains all of this naturally: the magnetic field turns the vacuum into a directional medium, and light leaving the magnetar is "neatly calibrated" by the field.
What remains unsettled
The researchers themselves stress this is a possible first observational detection, not a definitive proof. Open issues:
Still, this is the first time in 90 years that reproducible, cross-verifiable astronomical measurements have come this close to the prediction — a qualitative shift in itself.
Where this sits in the history of physics
When Heisenberg and Euler derived the vacuum birefringence correction in the 1930s, the laser had not even been invented. Ninety years later, three things finally came together: astronomical observations, space-based X-ray polarimetry plus radio telescopes, and an extremely magnetized cosmic source.
If this step holds: