My research is in molecular dynamics simulations and quantum chemistry. A lot of progress has been made in the field in the last few decades; however, there is a roadblock right now that no one really knows how to get around called the fermion sign problem. If we want to simulate atoms or molecules as accurately as we can, the number of resources we require scales exponentially with the number of particles. This is basically due to the Coulombic (electrostatic) interactions between the electrons in the atom. Accounting for these interactions gives many materials their interesting properties, but we've only been able to come up with approximations for solving these systems.
Quantum computing would allow these systems to be solved in polynomial time. The advantage would be enormous; in some cases, atomic simulation results would match experimental ones exactly†, and that is a very exciting notion.
(†glossing over a huge number of details here)