The semiconductor manufacturing industry, inherently capital and technology-intensive, faces growing complexities in production scheduling due to increasing process sophistication and stringent energy requirements. Diffusion Furnaces (DFs), which are essential in semiconductor doping processes, are critical bottlenecks due to their long processing times, high energy consumption, and the need to accommodate incompatible job families with specific time constraints. This study proposes a dynamic scheduling model for DFs aimed at minimizing total costs, including penalties for tardiness, time window violations, and electricity usage under a Time-of-Use (TOU) tariff structure. A mathematical model is formulated for non-identical parallel DFs, incorporating machine eligibility restriction (MER). The proposed mathematical model is empirically validated using randomly generated small size numerical problem and the LINGO solver. The results obtained from LINGO Solver for the numerical problem confirms the correctness of the model proposed for the research problem considered in this study. While the proposed model achieves optimal total cost, they exhibit computational intractability for large-scale problems. This research contributes to the advancement of efficient scheduling strategies for DFs in semiconductor manufacturing, with potential for significant cost reductions and improvements in energy efficiency.