We have developed a reactive force field (ReaxFF), which is able to reproduce accurately the physical and chemical properties of a comprehensive Fe/Na/P/O system. This ReaxFF was trained systematically using a large number of quantum data of relative energy, heat of formation, partial charges, bulk modulus, and crystal cell parameters of binary, ternary, and quaternary oxides using a robust parallel and multiparameter optimization of genetic algorithm (GA) to achieve the global optimization. The results indicated a substantial improvement upon previous ReaxFFs for systems of FexOy, NaxOy, and PxOy crystals. Moreover, an excellent prediction of molecular, electronic, and chemical properties of inorganic alkali polyphosphate (IAP) was found at low and elevated temperatures. An application of this developed ReaxFF in thin film lubrication of IAP confined between hematite surfaces showed a good agreement with experiments which showed that sodium played a vital role at IAP–hematite interfaces. The tribological performance of the sliding interface has been improved due to the formation of a hierarchical structure of the tribofilm.