Main navigation | Main content

HOME » PROGRAMS/ACTIVITIES » Annual Thematic Program

PROGRAMS/ACTIVITIES

Annual Thematic Program »Postdoctoral Fellowships »Hot Topics and Special »Public Lectures »New Directions »PI Programs »Math Modeling »Seminars »Be an Organizer »Annual »Hot Topics »PI Summer »PI Conference »Applying to Participate »

Talk Abstract

Modeling Issues in Control of Microelectronics Fabrication Processes

Modeling Issues in Control of Microelectronics Fabrication Processes

Processes to produce microelectronics devices are becoming more complex
while global competition is demanding that these processes be brought to
market more quickly and more reliably. The design of controlled
microelectronics processing systems such as Rapid Thermal Processors
(RTP), High Pressure Vertical Furnaces (HPV), Vapor Deposition Chambers,
etc., involve a large design parameter space, and therefore
trial-and-error process optimization is difficult and time consuming.
With the imminent scaling to 300 mm diameter wafer size, the industry has
realized that empirical approaches are not cost effective. In addition,
the controller design of such systems is often separated from the
equipment design, possibly resulting in a system operating far from its
maximum capability. While detailed non-linear physical simulations,
integrated with optimization and control design algorithms, have the
potential to help explore the design space, the long turn-around time
required to answer every ``*What if . . .*'' question makes this approach
impractical at the present time.

Another difficulty is the inclusion of the feedback control for closed-loop performance evaluations. The dynamic and/or sampled-data requirements of the feedback control system are often very difficult to incorporate into existing finite element and dynamic analysis software packages. Furthermore, real-time robust feedback control design only requires models that are accurate enough to meet the performance tolerances, e.g., simplified or reduced-order models together with a measure of the model error. What is missing at the present time are mathematical methods that can operate on the detailed nonlinear physical models of these advanced chambers to derive reduced-order models, together with appropriate measures of model error. Ideally, such methods would be analytic and not require extensive simulations. Lack of appropriate sensors remains an impediment even though there is a major push toward sensor development. Also, there is a lack of purely nonlinear controller synthesis methods that are implementable.

Physical phenomena in microelectronics processing systems involve chemical reactions, heat transfer, fluid mechanics, species transport, and plasma physics. Integrated physical models are required to represent these systems. Hence, progress in solving these complex engineering problems requires an interdisciplinary approach involving close interaction between systems and control theorists and their counterparts in applied mathematics, physics, chemistry, and material science. Furthermore, physical models must be validated within the context of a carefully planned experimental program.

We will illustrate the above problems by describing our experience in designing a temperature feedback controller for a Rapid Thermal Processing (RTP) chamber, and the development of an integrated model for control of an RF diode sputtering chamber for the growing of Giant Magneto-resistive (GMR) thin-films.