This paper is a review of recent work using heavy-ion collisions to learn about the phase structure of quantum chromodynamics (QCD) at high baryon density. The author emphasizes that to interpret collision experiments reliably we need good microscopic input for transport models — the computer simulations that follow particles and their interactions through a collision. The review focuses on several specific ingredients that strongly affect those simulations: how small nuclear clusters form, how particles containing strange quarks interact, the dynamics of collisions at a few GeV per nucleon, and how the nuclear force depends on the proton–neutron mix (isospin).

Heavy-ion collisions are important because by changing the collision energy and the types of nuclei used, experiments can reach a wide range of temperatures and baryon densities. These experiments are now carried out at many facilities, including the SPS, RHIC, and the LHC, and facilities for unstable beams such as FRIB, RIKEN, and RAON can explore different proton–neutron ratios. To extract physics about the QCD equation of state (EOS) — the relation between pressure, density, and temperature — researchers compare measured quantities to dynamical simulations. Depending on the energy, those simulations use either hydrodynamics (which assumes local equilibrium) or microscopic hadronic transport codes such as UrQMD, SMASH, GiBUU, and JAM that track individual particles and collisions.

The topic matters for several reasons. At low net baryon density lattice QCD calculations show that hadronic matter turns smoothly into a quark–gluon plasma. At higher densities, theory leaves open the possibility of a first-order phase transition and a critical point that experiments might find. Observations of neutron stars and gravitational waves provide complementary constraints on dense, neutron-rich matter. But to connect those astronomical constraints with laboratory collisions, and to judge whether any observed fluctuation or flow signals come from a true phase transition, we need reliable collision modeling that includes the microscopic physics mentioned above.