In this thesis we propose a conceptual cluster tool design for two independently moving wafer handling robot arms inside a radial configuration cluster tool. We develop a Petri Net model which describes the tool behavior. Ways to model both 1-cyclic and 2-cyclic schedules are developed. It is shown that the Swap sequence is not unconditionally the optimal sequence. We identify three dominant types of sequences. The workload concept is extended. With this, conditions for optimality are introduced and optimality is proven.
We develop an MIP formulation for the cases where the conditions for optimality are not satisfied. The MIP formulation finds optimal sequences amongst 1-cyclic schedules. It is shown that 1-cyclic schedules cannot always find the optimal schedule. Hence, a MIP formulation to find 2-cyclic optimal solutions is proposed. An efficient bounding mechanism still needs to be developed for the latter. We show experimental results for both.
We show that using two independent arms can significantly reduce the cycle time of the tool and increase chamber utilization compared to the existing regular cluster tool architecture.