Resolving semantic heterogeneity and giving users a unified view to access distributed data sources is one of the focuses in scientific data Management research community. For the integration and share of scientific data in materials science, a semantic model for materials scientific data is designed based on ontology technology. The semantic model covers the key concepts in materials science and their semantic relationships. With the semantic model which is considered as the global conceptual model, a semantic integration method is proposed, where mappings between the semantic model and local ontology as well as mappings between local ontology and data schema are built to make seamless semantic interoperability possible. Furthermore, the architecture and implementation details for the mappings are presented. Finally, the experimental results demonstrate the work of this paper, and show that the semantic model and integration method are useful in heterogeneous materials data environment.