3. INTRODUCTION
■ A process is controlled based on the information obtained.
■ Metrology values are available only for a small fraction of samples.
■ To overcome this, Virtual Metrology is used.
4. LIMITATIONS OF CURRENT METROLOGY
■ Metrology value depends on few sampled wafers.
■ Usage of metrology results is restricted.
■ Increased cycle time.
5. VIRTUAL METROLOGY
■ Prediction of metrology variables using information about the state of process or
product.
■ Class of methods aiming to estimate the metrology values, given the process data and
previous metrology information.
■ In practise it is difficult to measure each wafer.
■ So Virtual Metrology is used.
6.
7.
8. METHODOLOGY
■ Methodology is composed of three successive stages:
– Data Pre-processing
– VM Module Development
– VM Module Implementation
9. 1. DATA PRE-PROCESSING
■ Assuring the input quality of the data.
■ Three steps are involved:
– 1. Data Sources
■ To define family of products.
■ Selecting recipes and steps for VM module development.
■ Preliminary study of processes and metrology equipment's is necessary.
10. 1. DATA PRE-PROCESSING
– 2. Data Acquisition
■ To define raw data set including FDC data from production equipment and
measurement data from metrology equipment.
– 3. Data Consolidation
■ Includes data cleaning and statistical data analysis.
■ Identifying and removing outliers, missing values etc.. .
■ Statistical analysis include data normalization , data correlation studies.
11. 2. VM MODULE DEVELOPMENT
■ Aims to build different prediction models , compare them and validate the best model.
■ Three steps are involved:
– 4. VM Modelling
■ To choose nonlinear prediction methods.
■ Build prediction model in two levels.
12. 2. VM MODULE DEVELOPMENT
– 5. Model Comparison
■ Models compared based on robustness and prediction accuracy criteria.
■ To select the best model based to the performance.
– 6. VM Module
■ To perform the VM module with adjustments of the best model chosen in the previous
step.
13. 3. VM MODULE IMPLEMENTATION
■ To define the steps to integrate the VM module into the industrial environment.
■ This stage includes three steps :
– 7. VM Module Test
■ To perform off line test with the offline data from production.
■ To identify the problems of the model and to evaluate the results.
■ To define a prototype for offline VM Module Implementation.
14. 3. VM MODULE IMPLEMENTATION
– 8. VM Module in Production
■ Provide guidelines for full integration of VM module into manufacturing system.
– 9. VM Module Consolidation
■ To define the maintenance policies for the update of real time VM module.
15. ADVANTAGES
■ Increased production time.
■ Low scrap rate.
■ Predictive maintenance.
■ Real time tool performance.
■ Continuous information.
17. REFERENCES
■ Virtual metrology for run-to-run control in semiconductor manufacturing, Pilsung Kang
a, Dongil Kim a, Hyoung-joo Lee b, Seungyong Doh c, Sungzoon Cho a, Expert
Systems with Applications 38 (2011) 2508–2522
■ University of Padua, Department of Information Engineering, Virtual Metrology for
Semiconductor Manufacturing Applications, Prof. Beghi Alessandro, Gian Antonio
Susto, Padua, 28 June 2010