Lithography Sampling Scheme Using Frequency-Domain APC Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated process control methods in lithographic manufacturing are cumbersome and inefficient, struggling with optimizing metrology load, sampling rates, and time filter settings, which leads to suboptimal correction potential and increased costs due to reliance on computationally costly simulations and trial-and-error approaches.
Innovation Solution
A method that transforms measurement data into frequency domain data to determine optimal control schemes and sampling strategies, allowing for the identification of relevant contributors to errors and adjustment of time filter settings to accurately capture and correct lot-to-lot variations, thereby improving the effectiveness of the control loop.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional time-domain APC methods are used with weighted moving averages, then process control is simple to implement, but the control effectiveness is suboptimal and cannot accurately capture lot-to-lot variations
Solution Approach 1:
The patent transforms the control approach from time-domain to frequency-domain by changing the parameter domain. This allows accurate capture of lot-to-lot variations through spectral analysis while maintaining manageable complexity through systematic frequency domain processing methods
Solution Approach 2:
The patent replaces the mechanical time-domain filtering approach (weighted moving averages) with a frequency-domain spectral analysis system. This substitution enables more effective separation and correction of different types of variations through frequency-based processing
2Reliability
If metrology measurements are performed frequently to improve control accuracy, then yield improves, but throughput decreases due to time-consuming measurements
Solution Approach 1:
The patent applies partial action by performing metrology measurements on only a subset of substrates rather than all substrates. The frequency domain analysis of this partial measurement data enables effective control that generalizes to the entire batch, maintaining high yield while preserving throughput
Solution Approach 2:
The patent creates a frequency-domain representation (spectral copy) of the measurement data that captures the essential variation patterns. This spectral copy can be processed and used for control decisions without requiring additional physical measurements, thus maintaining throughput while improving control accuracy
3Measurement precision
If computational simulations are used to optimize control parameters, then control scheme accuracy improves, but computational cost and time increase
Solution Approach 1:
The patent substitutes iterative computational simulations with direct frequency domain analysis. By transforming measurement data to the frequency domain and analyzing spectral characteristics, the system directly determines optimal control parameters without requiring time-consuming simulation iterations
Data Source
AI summary
A method for generating a sampling scheme for a device manufacturing process, the method including: obtaining a measurement data time series of a plurality of processed substrates; transforming the measurement data time series to obtain frequency domain data; determining, using the frequency domain data, a temporal sampling scheme; determining an error offset introduced by the temporal sampling scheme on the basis of measurements on substrates performed according to the temporal sampling scheme; and determining an improved temporal sampling scheme to compensate the error offset.