Fundamental physics research appears very unlikely to pose an existential risk. Physicists have extremely accurate models of the world, and are able to estimate the likely consequences of different experimental conditions. Even where they do not know precisely what will happen, they usually have a fairly good idea of roughly what will happen.
However, one of the reasons to run experiments in the first place is to probe areas where the existing models of physics might be wrong. If the models are wrong, and the experiments are sufficiently exotic, there is some very small chance of accidental and unpredictable harm resulting. Some, for example, have concerns that strangelets created by particle accelerators might cause catastrophe. Such risks are extremely unlikely conditional on our phyiscal models being true. But the risks become more concerning given a non-negligible chance that the models are mistaken.
Ord, Toby, Rafaela Hillerbrand & Anders Sandberg (2010) Probing the improbable: methodological challenges for risks with low probabilities and high stakes, Journal of risk research, vol. 13, pp. 191–205.
A technical paper investigating ways to estimate risks from model uncertainty.