Kenneth Wayne Leffew

Single layer fluropolymer resists for 157-nm lithography

We have developed a family of 157 nm resists that utilize fluorinated terpolymer resins composed of 1) tetrafluoroethylene (TFE), 2) a norbornene fluoroalcohol (NBFOH), and 3) t-butyl acrylate (t-BA). TFE incorporation reduces optical absorbance at 157 nm, while the presence of a norbornene functionalized with hexafluoroisopropanol groups contributes to aqueous base (developer) solubility and etch resistance. The t-butyl acrylate is the acid-catalyzed deprotection switch that provides the necessary contrast for high resolution 157 nm imaging. 157 nm optical absorbances of these resists depend strongly upon the amount of t-BA in the polymers, with the TFE/NBFOH dipolymers (which do not contain t-BA) exhibiting an absorbance lower than 0.6 μm-1. The presence of greater amounts of t-BA increases the absorbance, but also enhances the dissolution rate of the polymer after deprotection, yielding higher resist contrast. Formulated resists based upon these fluorinated terpolymer resins have been imaged at International Sematech, using the 157 nm Exitech microstepper with either 0.6 NA or 0.85 NA optics. We have carefully explored the relationship between imaging performance, resist contrast, optical absorbance, and t-BA content of these terpolymer resist resins, and describe those results in this contribution.

Application of Digital Control Techniques to a Laboratory Extrusion Process

Extruders are now routinely controlled with digital computer systems and algorithms. In this study, we have installed, debugged, and demonstrated an advanced control algorithm for individual temperature-zone control of a laboratory single screw extruder. The algorithm: ¿ adjusts the parameters of a resident first-order plus deadtime process model with a linear least squares estimator, ¿ adapts the controller parameters to achieve a prespecified closed-loop setpoint response (Dahlins method) and ¿ is in use in a polymer processing research laboratory, installed on a DEC PDP 11/23 system.

6σAND SOLID-STATE POLYMERIZATION

Abstract An industrial application of data-driven modeling to a solid-state polymerization process is presented. The goal is to predict the exit IV (intrinsic viscosity) given a set of independent variables. The problem is interesting in that it involves a distributed parameter system with numerous inputs and many hours of time delays. High-fidelity delay-free predictions allow for the adjustment of independent variables for improved quality control. The problem is analyzed from the increasingly popular six sigma perspective for a comparative assessment between modeling, advanced control, and optimization and six sigma.

Constrained Model Predictive Control of a Solid-State Polymerization Process

Abstract An industrial application of constrained model predictive control (CMPC) to a solid-state polymerization process is presented. The system has six controlled variables and seven manipulated variables, presenting possible opportunities for economic optimization. A commercially available CMPC software package is used in the investigation. The step response model used in the software is developed from experimental plant tests. Excellent performance has been achieved in terms of servo and regulatory control and optimization.

Simulation and Control of a Multi-Stage Batch Operation with Measurement Constraints

This paper presents the simulation of a multi-stage batch operation, such as sputtering. The objective of the research is to examine the effect of different sampling strategies on the overall control of part-to-part quality, not the control of the continuous processes that occur during each stage. Selection of a sampling strategy is difficult because the measurement system for the operation is constrained in such a way that sampling control variables at each stage simultaneously is not possible. At individual stages, supervisory controllers are implemented to maintain part-to-part quality. Hence, the process and its simulation are completely discrete since it is the behavior of the supervisory loops that are studied. The immediate results of this research indicate that the strategies that repetitively sample every other stage are superior to the other methods considered. Additionally, these results may direct the development of intelligent sampling strategies that will not be static and fixed, but will strategically sample stages in order to better maintain overall product quality.