Department of Chemical Engineering An Object-Oriented Modular Simulation Framework for Semiconductor CVD Processing Systems Jing Chen, Raymond A. Adomaitis Objectives Data • Design a set of computational tools for solving lumped and distributed parameter models generated in process design and simulation • Develop flexible and reusable modular components for CVD systems • Apply the technique of object-oriented design to reduce developing cycles of designing new systems and lower costs of CVD process design and simulations • Facilitate distributed simulation through the application of Java and XML technique to heterogeneous data archiving, analysis and presentation Why OOP and Modular Case Study -- Programmable CVD System • Archive data in XML format • Facilitate distributed simulation and information sharing Simulator input Ar Web applications Simulator validation Process diagnosis Simulation of gases composition profile along segment Approach • Standalone modules which could be physical parts or virtual parts describing transport or reaction mechanisms Processes/Applications • Determine the modules: • one segment class : describe the transportation inside a segment • one inter-flux class : describe the diffusion of gases among segment ALD PECVD Other process Programmable CVD MOCVD BREAK DOWN • Compose the system: • 3 objects of the segment module • 1 object of the inter-segment flux module Gas Lamp ShowerHead Surface Reaction Wafer GasPhase Reactions SubHeater Qik • Solve the system with non-linear solver DepRate Standalone Modules Framework Systems • Specified by users through input of modularized components • Acting as a coordinator for information exchanging and distribution among modules in the system z=L Nik Systems micro valve Steady-state CSTR model of WCVD Other userdefined system z = zf Orifice …... Conclusions Modular components Relations • Computational tools include two main packages: Solutions Computational tools • Solvers offer different algorithms for solving AE/ODE/DAE systems • MWRtools used for discretizing PDE systems and solving the systems by weighted residual methods MWRtools DAE nonlinear solving Manager • The application of OOD and modular approach greatly improves current modeling and simulation capability Solvers ODE Parser z=0 wafer surface at z = -h Energy bal’n on wafer Computational Tools Systems As Mass bal’n in gas phase Channel Discretizing Experimental data Property estimator Af W Data Physical/ thermal/ kinetics data Details on the system and modeling equations can be found in “Spatially Programmable Chemical Vapor Depostion” -- Choo, Adomaitis, et.al., 2003 Parameter identification Modular Components WaferTemp • Modular approach: • Allow different variable names among modules • Allow distributed simulation • May solve modules with different algorithms • Easy to change model structure H2 WF6 Distributed applications Web applications • Include information on properties of equipment or process and have subsystem equations associated with • Features of OOP: • Encapsulation • Inheritance • Polymorphism Property Data Experimental Data • Separates data applications and presentation PDE systems ODE AE DAE er nv co tin g • The development of the object-oriented simulation library offers a set of flexible and adaptable tools for solving large systems consisting of different application models • Modularized components can be easily integrated to form a new CVD systems. The system can be solved in the sequential or simultaneous approach depending on the specific applications • The application of Java and XML technology to wrap and manipulate data facilitates data distribution among dissimilar applications