I first started working in data science and data modelling as a postgraduate researcher in the Dynamics and Control (DAC) group of the Electronic Engineering Department in NUI Maynooth. During this time, I completed a PhD thesis is entitled “Virtual Metrology for Plasma Etch Processes“, starting October 2006, and finished in March 2011. Phds are long, but there were some great times!
My research focussed on state estimation and control for a plasma etch process used in semiconductor manufacture by the microchip fabrication plant, Intel Ireland, which is local for me. The aim of this work was to create mathematical and statistical models that relate readily available in-line measurements from fabrication tools to the final results of the fabrication process. There were essentially three main topics addressed in the thesis:of m
- The estimation of variables such as plasma etch rate (how fast material is removed from a wafer surface during plasma etch), using machine learning techniques applied to optical and electrical signals. With such measurement estimates, faults can be detected in a timely fashion, and processes can be run in a more controlled manner, reducing downtime for plant tools and increasing overall throughput.
- Spatial temperature estimation for a plasma rocket engine called VASIMR. In October 2007, I took part in the Irish based FÁS Science Challenge where he was awarded a scholarship that funded research work for six months with the Ad-Astra Rocket Company, Costa Rica. In the period Oct 07 – Apr 08, I worked with Ad Astra in Costa Rica applying virtual metrology techniques to the VASIMR engine, a plasma based rocket propulsion system for interplanetary travel.
- Real-time control of plasma etch using virtual metrology. During the final year of my postgraduate research, I developed a real-time virtual metrology-based model-based control system for plasma electron density and etch rate in an industrial etch chamber. The control system used a predictive function control algorithm to stabilise the plasma electron density.
- Machine Learning, data reduction, and data visualisation of high-dimensional data sets.
- Virtual metrology for semiconductor manufacturing.
- Linear and nonlinear modelling techniques.
- Real time predictive and run-to-run control of plasma processes.
- Development of the VASIMR rocket engine.
- Application of virtual metrology techniques in high volume manufacturing environments.