The NOvA collaboration searched for magnetic monopoles — hypothetical particles that carry a single magnetic charge — using 2,743 live days of its Far Detector. They found no events consistent with monopoles. From that null result they set an upper limit on the monopole flux of 8 × 10^−16 cm^−2 s^−1 sr^−1 at 90% confidence for monopoles moving with speed 6 × 10^−4 < β < 5 × 10^−3 and with mass above 10^9 GeV. Here β is the particle speed as a fraction of the speed of light c (so β = v/c). This search covers a low‑mass region that had not been explored before by other experiments because NOvA sits near the surface with only about three meters water equivalent of overburden shielding above it.

Magnetic monopoles were first proposed by Dirac and are also predicted in some grand unified theories. Different experiments look for them in different ways. Deep underground detectors have placed very strong limits on very slow, extremely heavy monopoles. Mountaintop experiments and particle colliders probe other mass ranges. NOvA’s strength is its large, surface detector, which can see slow monopoles that would be stopped before reaching deeper detectors but can reach this shallow site if their mass is at least about 10^9 GeV.

The NOvA Far Detector is a 14‑kiloton liquid scintillator detector made of many long plastic cells. Charged particles passing through the liquid make light. That light is collected by fibers and read out by avalanche photodiodes sampled continuously at 2 megahertz. The detector records two projected views (called xz and yz) that are combined to reconstruct three‑dimensional tracks. The experiment ran a dedicated monopole trigger that looks for slow, straight tracks spread in time across the detector.

To estimate how well they would see monopoles, the team simulated particles passing through the detector with a range of directions and speeds. They assumed monopoles travel in straight lines and do not stop inside the detector because their total energy is much larger than what they lose traversing the detector. The simulations used the Geant4 toolkit for energy deposition, with a conservative choice of 90% of the nominal energy loss values, and each simulated signal was mixed with five milliseconds of real detector data to include the ordinary cosmic‑ray activity that makes the background. Samples were generated across speeds from 10^−4 to 10^−2 in β to measure detection efficiency.