Ilias Iliopoulos

Use of Virtual Metrology for in-situ Visualization of Thermal Uniformity and Handoff Adjustment in RTP Critical Anneals

Applied materials rapid thermal processing (RTP) systems are unique in providing high resolution process data particularly wafer rotation angle and wafer rotation speed as a function of time. This work explores how this information can be used to predict on-wafer process results using an advanced analysis package known as WISR (wafer interdiction and scrap reduction). WISR is an advanced process control platform for the collection, storage, visualization, and analysis of process parameters from production tools. One of the main analysis features of WISR is the ability to create virtual sensors. Virtual sensors are calculated parameters derived from physical sensors that can provide real-time and statistical representation of process health. The focus of this paper is the ability of WISR to transform time series chamber parameters from the pyrometers and the magnetic levitation controller into thermal wafer images at any time during the recipe execution and provide wafer-to-wafer handoff correction. These abilities constitute a virtual sensor module in WISR known as Virtual metrology (VM). We describe the implementation of the VM-analysis and show how temperature maps and handoff corrections correlate with offline metrology in RTP critical anneals. This is an innovative new method to significantly reduce wafer processing errors, enhance yield, and minimize production cost.

Virtual Metrology in RTP with WISR

The Applied Materials RTP systems are unique in providing high resolution process data particularly lamp power and temperature as a function of radial position and time. This work explores how these data can be exploited to predict on-wafer process results using an SPC analysis package named WISR. WISR is an advanced process control platform for collection, storage, visualization, and analysis of data from process tools. The product is collecting data at rates up to 100Hz. It has the ability of notification and interdiction if data goes outside of specification. One of its main features is the ability to create virtual sensors. These are parameters that are derived from physical sensors and provide a statistical representation of the process health. The focus of this paper is the ability of WISR to transform the time series chamber data from the pyrometers into thermal wafer images at any time window during the recipe execution. We describe the data analysis that was deployed, and show how temperature maps generated correlate with metrology data from RTP processes. This is a disruptive new method to significantly save monitor wafer cost, enhance yield and prevent wafer processing errors.