Empty space isn't truly empty, according to quantum physics — and a magnetar with a magnetic field hundreds of trillions of times stronger than Earth's has now produced the strongest evidence yet, polarizing X-rays in a pattern researchers struggled to explain without a 90-year-old quantum effect.

A magnetar called 1E 1547.0−5408 has produced the strongest observational case so far that an intense magnetic field can change how light travels through otherwise empty space. The evidence comes from X-rays whose polarisation stayed unusually high and followed the star’s magnetic geometry as it rotated.

Rachael Stewart, Hoa Dinh Thi and colleagues report the measurements in a peer-reviewed Nature paper published on 5 August 2026. Their interpretation invokes vacuum birefringence, a quantum-electrodynamic effect first described by Werner Heisenberg and Hans Heinrich Euler in 1936.

This is one study, not settled consensus. IXPE directly measured the polarised X-rays, but vacuum birefringence is the explanation inferred from those data and from models of how the radiation crossed the magnetar’s atmosphere and magnetic field.

What IXPE measured

NASA’s Imaging X-ray Polarimetry Explorer, or IXPE, observed 1E 1547.0−5408 for more than 140 hours between 26 March and 5 April 2025. NASA’s NICER instrument measured its X-ray spectrum and timing, while Murriyang, CSIRO’s 64-metre Parkes radio telescope, followed the star’s radio pulses.

The magnetar completes one rotation in about 2.09 seconds. By sorting the incoming photons according to rotational phase and energy, the team could watch both the degree of polarisation and its direction change across each turn.

Across IXPE’s 2 to 8 kiloelectronvolt band, the phase-averaged polarisation degree was 46 ± 4 per cent. It rose to 59 ± 5 per cent in the softer 2 to 3 kiloelectronvolt band. At some rotational phases, the measured value reached 82 ± 15 per cent and remained at least about 40 per cent through the part of the cycle associated with the radio beam.

Those uncertainties matter. The result is not that every X-ray from the star was aligned, nor that 82 per cent is an exact fixed property. The more useful observation is the sustained, coherent pattern across energy and rotational phase.

A magnetic field beyond terrestrial reach

Magnetars are neutron stars whose magnetic fields exceed 10¹⁴ gauss. The Nature paper uses a dipole-field estimate of about 2 × 10¹⁴ gauss for 1E 1547.0−5408, equivalent to roughly 20 billion tesla.

Earth’s surface field is about 0.5 gauss on average. Using that comparison, the magnetar’s estimated field is about 400 trillion times stronger. It also exceeds the quantum-electrodynamic critical field of roughly 4.4 × 10¹³ gauss, the scale at which strong-field quantum effects become difficult to ignore.

The star is therefore a natural physics experiment whose magnetic conditions cannot be sustained in a terrestrial laboratory.

Why quantum theory says a vacuum can refract light

In classical electromagnetism, an ideal vacuum has no material structure and does not care how a light wave is polarised. Quantum electrodynamics gives the vacuum a more complicated description. It is the lowest-energy state of electromagnetic and charged-particle fields, and those fields can respond to an external magnetic field.

In the Heisenberg-Euler description, quantum corrections make the electromagnetic response slightly nonlinear. The two natural linear-polarisation modes of a photon then acquire different effective refractive indices when they travel through a sufficiently strong magnetic field. That is vacuum birefringence.

The familiar analogy is a birefringent crystal, but the comparison has limits. There is no crystal lattice around the magnetar and IXPE did not see one beam split into two visible rays. The effect changes how the orientation and phase of the X-ray polarisation evolve while the photons move through the magnetosphere.

A common explanation describes virtual electron-positron pairs flickering in and out of existence. This is useful intuition, provided the pairs are not mistaken for a literal hidden gas. They are part of the quantum calculation that gives the vacuum its field-dependent response.

The radio data restricted the geometry

A large polarisation percentage by itself is not a clean detection of vacuum birefringence. Radiation can leave a strongly magnetised neutron-star atmosphere already highly polarised, and a small hot region on the surface can preserve much of that alignment.

The geometry determines how much survives when an unresolved telescope adds light from different parts of the star. The result depends on the angle between the rotation axis, magnetic axis and our line of sight, as well as the size and position of the X-ray-emitting region.

That is why simultaneous radio observations mattered. The radio polarisation angle made a characteristic sweep as the magnetar rotated, allowing the researchers to constrain its large-scale magnetic geometry independently of the X-ray fit. The X-ray polarisation angle followed a broadly similar rotating-vector pattern.

The team then simulated the X-ray intensity and polarisation with and without vacuum birefringence. Within its tested framework, the model that included the quantum effect reproduced the observed Stokes parameters more successfully. Switching the effect off produced stronger phase-dependent variations than IXPE saw, while the best alternative geometry sat uneasily with the radio constraints.

A recent SpaceDaily analysis examines the measurement and model chain in greater technical detail.

Why the interpretation remains contested

A separate peer-reviewed analysis led by Roberto Taverna, published in The Astrophysical Journal, examined the same magnetar and reached a more guarded conclusion. Its authors agreed that the polarisation-angle pattern and energy dependence hint at quantum-electrodynamic effects, but argued that high polarisation from a small hot spot does not by itself provide compelling evidence for magnetospheric vacuum birefringence.

The disagreement turns largely on geometry and modelling. The Taverna team derived an inclined configuration from the X-ray data alone in which strong polarisation could reach the observer without requiring the magnetospheric effect. The Nature team used the coordinated radio observations to favour a nearly aligned geometry and found that its vacuum-on simulations gave a substantially better account of the combined pattern.

Neither paper disputes that IXPE detected an unusually strong X-ray polarisation signal. The question is how uniquely the signal selects vacuum birefringence once uncertain surface emission, hot-spot shape, plasma effects and non-dipolar magnetic structure are allowed.

That distinction is why “strongest evidence yet” is more defensible than “proof”. NASA’s own account uses the word “may”, while the Nature authors describe their result as a marked advance and call for further observations and theory.

What would strengthen the case

More X-ray photons would narrow the uncertainties, particularly above 4 kiloelectronvolts where the present measurements are less precise. Repeating the same phase-resolved pattern at another epoch would test whether it follows a stable magnetic geometry rather than a temporary state of the magnetosphere.

Observing other radio-emitting magnetars would be more decisive still. A relationship that recurs between independently measured radio geometry and the X-ray polarisation predicted by vacuum birefringence would be harder to reproduce with a special hot-spot arrangement on one star.

Theory also has to test a wider range of surface compositions, field twists, higher-order magnetic structure and plasma propagation. The open question is whether the vacuum-birefringence signature remains necessary across those plausible alternatives.

For now, 1E 1547.0−5408 has moved the search from suggestive polarisation towards a constrained, multiwavelength test. The next observations will determine whether that model-dependent case can become a repeatable measurement across more than one magnetar.