Owing to its attractive material properties, tungsten (W) and its alloys are leading solid-material candidates for plasma-facing components in future fusion devices. A fundamental limitation of W in such applications is its intrinsically brittle behavior, which results in low fracture toughness and manifests predominantly as intergranular fracture. This issue is further amplified by the segregation of impurities at grain boundaries (GBs), which tends to reduce the fracture toughness even more. To quantify the impact of impurities on fracture toughness, atomistic modeling is particularly valuable, as it allows the explicit incorporation of local crystalline and chemical effects into the traction–separation behavior. However, upscaling atomistic results to microscale continuum models remains challenging, largely due to the non-uniqueness of atomistically informed data. To address this challenge, we implement an approach to extract scale-independent GB traction–separation properties based on the interfacial excess framework developed by Van der Ven and Ceder [1]. As a demonstration, we investigate the fracture behavior of a large set of W twist GBs containing phosphorus (P) impurities using classical molecular dynamics simulations. The simulations employ a newly developed modified embedded atom method (MEAM) interatomic potential, specifically designed to capture the embrittling effect of P on W GBs with near density-functional-theory (DFT) accuracy [2]. The results show that scale-independent excess properties can be successfully extracted using the proposed approach, enabling a quantitative assessment of impurity effects on GB cohesion. Furthermore, these properties can be used to construct cohesive zone relations for microscale continuum modeling of GB fracture. References [1] Van der Ven A., Ceder G., \emph{The thermodynamics of decohesion}, Acta Mater, Vol. 52, pp. 1223–1235, 2004. [2] Olsson P. A. T., Hiremath P., Melin S., \emph{Atomistic investigation of the impact of phosphorus impurities on the tungsten grain boundary decohesion}, Comput. Mater. Sci., Vol. 219, pp. 112017, 2023.