Dr. Gary W. Rubloff
Research Topics
Process diagnostics, sensing, metrology, and control
Semiconductor manufacturing processes and equipment
Systems modeling, simulation, and optimization
Engineered learning systems

Recent Results



Current class


Research group

Contacts & maps


Collaborators' corner

Consumer reports

Counter by digits.com

All materials at this site are copyrighted 2003 by the University of Maryland
1996-2003, all rights reserved

Fall 1999: Programmable Reactor Design









Return to Teaching

In fall 1999, the design project was aimed at exploring a new concept for semiconductor manufacturing processes which could substantially impact the semiconductor industry's state-of-art, delivering a meaningful technical and systems assessment as well as a development roadmap for the concept.

The opportunity for a new concept was driven by the recognition that the cost and complexity of competitive semiconductor manufacturing is substantially determined by the cost of the manufacturing process equipment used in the fab, and also that today's equipment technology depends on process reactor designs which are fixed once the reactor is constructed.

The new concept central to the design project was the notion of a programmable reactor, in which process parameters could be varied as a function of position using software control and hardware sensing and actuation devices.  This means that the process consequences could be varied as desired as a function of spatial position in the reactor, or more specifically across the semiconductor wafer.  With the current evolution from  8" to 12" wafers and the enormous value represented by a processed wafer, substantial benefit can be anticipated if improved spatial control of the process consequences across the wafer were realized.

In particular, two major advances were envisioned.  First, a spatially programmable reactor would ensure that across-wafer uniformity adequate for manufacturability could be achieved at any desired process design point (parameter set). This would mean that process performance need not be compromised to meet the uniformity demands of manufacturing.  Second, a programmable reactor design would allow more rapid research and development cycles, allowing multiple process design points to be investigated on a single wafer by intentionally programming in across-wafer nonuniformity, as well as by reducing the experimentation needed to optimize the tradeoff between process performance and uniformity.

The class investigated a spectrum of semiconductor processes, focused primarily on chemical vapor deposition as an early target for the programmable reactor concept, and addressed a variety of key materials and systems issued involved in the development of the concept.