Applying the sequential neural-network approximation and orthogonal array algorithm to optimize the axial-flow cooling system for rapid thermal processes

• Published in: Semiconductor science and technology Vol. 24; no. 2; pp. 025021 - 025021 (10)
• Main Authors: Hung, Shih-Yu, Chang, Ying-Pin, Shen, Ming-Ho
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.02.2009; Institute of Physics
• Subjects: Cold working, work hardening; annealing, quenching, tempering, recovery, and recrystallization; textures; Cross-disciplinary physics: materials science; rheology; Exact sciences and technology; Materials science; Physics; Treatment of materials and its effects on microstructure and properties; Cooling systems; Velocity distribution; Fracture criterion; Rapid cooling; Rapid thermal annealing; Gas flow; Sequential method; Neural networks; Momentum approximation; Thermoelasticity; Modelling; Silicon; Thermal stresses
• ISSN: 0268-1242; 1361-6641
• DOI: 10.1088/0268-1242/24/2/025021

The sequential neural-network approximation and orthogonal array (SNAOA) were used to shorten the cooling time for the rapid cooling process such that the normalized maximum resolved stress in silicon wafer was always below one in this study. An orthogonal array was first conducted to obtain the initial solution set. The initial solution set was treated as the initial training sample. Next, a back-propagation sequential neural network was trained to simulate the feasible domain to obtain the optimal parameter setting. The size of the training sample was greatly reduced due to the use of the orthogonal array. In addition, a restart strategy was also incorporated into the SNAOA so that the searching process may have a better opportunity to reach a near global optimum. In this work, we considered three different cooling control schemes during the rapid thermal process: (1) downward axial gas flow cooling scheme; (2) upward axial gas flow cooling scheme; (3) dual axial gas flow cooling scheme. Based on the maximum shear stress failure criterion, the other control factors such as flow rate, inlet diameter, outlet width, chamber height and chamber diameter were also examined with respect to cooling time. The results showed that the cooling time could be significantly reduced using the SNAOA approach.