This paper reviews experimental hints for a possible new light particle called X17 and the theories that could explain it. The X17 name comes from an excess seen in electron–positron pairs that looks like it could come from a neutral boson with mass near 17 MeV. The authors collect the experimental results, survey proposed theoretical frameworks, and test how an X17 would show up in precision measurements.

The experimental story centers on measurements of internal pair creation (IPC), a process where an excited nucleus emits a virtual photon that becomes an electron–positron pair. The ATOMKI group studied the reaction 7Li(p,γ)8Be using a proton beam near 1.03 MeV to populate an 18.15 MeV excited state of 8Be. They reported a significant excess of pairs at large opening angles that looks like a bump in the reconstructed mass near 16.7 MeV/c2, quoted with a significance above 6σ. A later ATOMKI-like signal was also reported in 4He. An independent group at VNU University of Science found a compatible excess with mX = 16.66 ± 0.47(stat) ± 0.35(sys) MeV and significance above 4σ. Other dedicated searches, notably the MEG II experiment, did not see a matching signal, and the NA64 beam‑dump experiment found no X17 in a bremsstrahlung search, placing strong limits on how strongly X17 could couple to electrons.

On the theory side, the authors review several ideas that can produce a 17 MeV boson while trying to respect existing experimental bounds. The most studied option is a light vector boson whose coupling to protons is suppressed — a “protophobic” vector — so it can be produced in the specific nuclear transitions without violating other constraints. In this picture the effective couplings needed to explain the nuclear signals are small, roughly 10^−4 to 10^−3. Scalar or pseudoscalar alternatives have also been proposed, but angular‑momentum considerations in the 8Be transition and limits from other particle and astrophysical searches make those explanations harder to reconcile. More general frameworks introduce a portal between the Standard Model and a hidden “dark” sector, but these typically need careful tuning to remain allowed.