Atomic scale defects critically limit performance of semiconductor materials. To improve materials, defect effects and defect formation mechanisms must be understood. In this paper, we demonstrate multiple examples where molecular dynamics simulations have effectively addressed these issues that were not well addressed in prior experiments. In the first case, we report our recent progress on modelling graphene growth, where we found that defects in graphene are created around periphery of islands throughout graphene growth, not just in regions where graphene islands impinge as believed previously. In the second case, we report our recent progress on modelling TlBr, where we discovered that under an electric field, edge dislocations in TlBr migrate in both slip and climb directions. The climb motion ejects extensive vacancies that can cause the rapid aging of the material seen in experiments. In the third case, we discovered that the growth of InGaN films on (0001) surfaces suffers from a serious polymorphism problem that creates enormous amounts of defects. Growth on (
) surfaces, on the other hand, results in single crystalline wurtzite films without any of these defects. In the fourth case, we first used simulations to derive dislocation energies that do not possess any noticeable statistical errors, and then used these error-free methods to discover possible misuse of misfit dislocation theory in past thin film studies. Finally, we highlight the significance of molecular dynamics simulations in reducing defects in the design space of nanostructures.