OPC Modeling With Joint Optical–Resist Parameter Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional step-wise optimization strategies for OPC modeling fail to simultaneously determine optimum values for optical and resist model parameters, leading to suboptimal balance and optimization of photolithography model parameters.
Innovation Solution
An OPC modeling method that simultaneously adjusts optical and resist model parameters through stochastic combination generation, photolithography simulations, Pareto principle evaluation, and genetic algorithms to find optimal parameter combinations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a step-wise optimization strategy is used to determine optical model parameters first and then resist model parameters, then the optical model parameters can be optimized, but the resist model parameters cannot achieve true optimum values because the success of the second step depends on the first step
Solution Approach 1:
The patent combines the optimization of optical model parameters and resist model parameters into a single simultaneous optimization process. Instead of performing step-wise optimization where optical parameters are optimized first and then resist parameters, the patent uses a genetic algorithm that evaluates and optimizes both sets of parameters together based on photolithography simulation results, allowing them to influence each other's optimization and achieve true joint optimum values.
2Ease of manufacture
If step-wise optimization is performed where optical model parameters are determined first and resist model parameters are determined second, then the process is simpler to implement, but the best balance between optical and resist model parameters cannot be achieved
Solution Approach 1:
The patent implements a feedback mechanism where photolithography simulation results are used to evaluate the combined performance of optical and resist model parameters. The simulation outcomes feed back into the genetic algorithm, which adjusts both sets of parameters in subsequent iterations. This closed-loop feedback ensures that the optimization process achieves the best balance between optical and resist parameters by continuously evaluating their combined effect on critical dimension accuracy.
3Power
If separate simulation of optical and resist model parameters is performed in sequence, then computational complexity is reduced, but the optimum values for both parameter sets cannot be determined simultaneously
Solution Approach 1:
The patent changes the optimization approach from sequential parameter adjustment to simultaneous parameter optimization using a genetic algorithm. The algorithm maintains populations of both optical and resist parameters and evolves them together through selection, crossover, and mutation operations. This allows the system to explore the joint parameter space and determine optimum values for both parameter sets simultaneously, rather than sequentially, thereby achieving higher optimization accuracy.
Data Source
AI summary
An OPC modeling method is disclosed, which includes: step S1: determining optical model parameters and resist model parameters; step S2: obtaining a plurality of parameter combinations by stochastically choosing values for the parameters; step S3: performing photolithography simulations and etching wafers and calculating RMS values of differences between simulated CDs and etching CDs and BCE values of the CDs; step S4: evaluating the values according to Pareto principle and calculating Pareto optimum to N-th-best Pareto suboptimum sets to prioritize the plurality of parameter combinations in a descending order; step S5: applying a genetic algorithm with position-based crossover and/or mutation to the plurality of parameter combinations, to obtain new parameter combinations; and step S6: iterating steps S3 to S5 on the new parameter combinations until a number of iterations reaches a first predetermined value and using highest prioritized ones of parameter combinations resulting from a last iteration for OPC modeling.