Abstract: Channel layout optimization is a typical constrained spatial arrangement problem widely found in manufacturing and engineering applications. To address channel layout problems with multiple objectives and strict spatial constraints, this study proposes an optimization framework based on a genetic algorithm. A weighted-sum method is adopted to convert the two objectives, namely maximizing the number of flexible items and minimizing the clustering degree of items within channels, into a single-objective optimization problem. Furthermore, a three-layer chromosome encoding and decoding strategy is designed according to the structural characteristics of channel layouts, enabling the generation of high-quality solutions while balancing multiple objectives. The effectiveness of the proposed framework is validated through its application to alignment mark placement on semiconductor photomasks. Comparative experiments show that, for small-scale problems, the proposed method achieves solution quality comparable to the commercial solver Gurobi (with an average gap of 1.63%) while reducing computation time by 99.56% (a 230-fold speedup). For large-scale problems, compared to the particle swarm optimization algorithm, the proposed algorithm improves solution quality by 19.24% and reduces computation time by 87.00%. These findings demonstrate that the proposed method offers high efficiency and strong applicability for channel layout optimization.

Key words: channel layout, genetic algorithm, semiconductor manufacturing, photomask