Design for Reliability, Chapter 1: Can New Approaches in Wafer Handling Design Support the Demand for Better Process Technology? |
By Ghulam Mustafa |
Abstract Manufacturing Practices of Today The Impact of Moore’s Law Since 1965, when Gordon Moore first proposed that the number of transistors would double every two years, Moore’s Law has become the holy grail for the IC industry. Indeed one can argue that the ITRS, the guiding document for the semiconductor industry is driven by Moore’s Law and has been the engine of growth for the high tech for the past 40 years. It is the de facto enabler for the dawn of the nanotechnology era, currently at 65 nm node, with 45 nm and beyond geometries to follow in the not so distant future. Tools are being designed, developed and manufactured for ever shrinking line widths and far more exacting processes. Just imagine if the automotive industry had a similar technology growth paradigm. Rapid Development Affects Reliability Rapid delivery of high technology implies short development and test times. This means that at launch, or even after delivery, processes may not be fully understood, and not thoroughly tested. But revolutionary new tools that arrive first on the market also arrive with development issues that need attention, sometimes even when they are in production. In fact, Industry veterans believe that by the time a new product is fully tested and debugged, it is very likely obsolete. The cycle is repeated every 18 to 24 months. When a new generation of tools arrives, so does a whole new set of technology challenges. Remarkably, the semiconductor capital equipment market accepts this as norm, given that the alternative of thorough development and test results in loss of market share. We are running a continuously accelerating treadmill, driven ever faster by innovation and delivery pressures. That and the fact that at its core, the IC industry relies on semiconductor manufacturing equipment that is sometimes unreliable, poses a dilemma. To thrive in this environment, the industry has developed a number of response strategies, namely:
The High Cost of Rapid Development These approaches have worked well in the past; they come, however, with a heavy price tag. This leads us to the flip side of Moore’s Law. As chip density increases exponentially, the cost of manufacturing infrastructure also increases exponentially. This is the sad consequence of Moore's Law, the exponentially increasing force behind our high tech revolution. Complexity has a price. Moore’s Law has made the nanotechnology a reality; can we afford the flip side at an exponential cost? How can the Semiconductor Capital Equipment Industry Respond? The Historical Approach: Cluster Tools In the 1980’s, the semiconductor capital equipment industry moved toward the “cluster tool” as an efficient and logical design for process platform. The value to the equipment manufacturer was to deliver the process technology at an accelerated pace to the customer without sinking resources in developing the wafer handling part of the tool. Accordingly, a number of processes were logically organized into groups, or clusters. Wafers were then transferred from one process to another by an integrated wafer transport module. Delivering new technology meant new process chambers, not the wafer handling robot. Smaller and Faster, but Cheaper? In order to minimize footprint and maximize throughput a common approach taken by equipment manufacturers is to link the processes with a robot based transfer module. Wafer transfer is carefully choreographed from one process to the next, with careful attention to fast, efficient, clean and well aligned transfers without dropping or damaging the wafer. Most often, wafer delivery is unique to each platform, so highly customized software and alignment mechanics are applied in order to ensure performance. The complexity of the wafer movement makes it especially challenging for the operator to keep the system running perpetually at full capacity. Furthermore, since the wafer movement does not add value, the objective is to make the robot work just enough so as to have a minimal impact on the “value” side of the platform. The processes come first. Robotics comes later. As the primary and central motion generating mechanisms, wafer-handling robots tend have significant impact on tool uptime, throughput and reliability. It is possible to take one or more process chambers off-line; any failure of the robot will bring a tool down for many hours. The high speed mechanical movement and low tolerances for variances make the wafer transfer amenable to failure. To make matters worse, each platform has a different robot, which means that keeping these systems up and running can be a daunting task. Next Generation Wafer Handling: A New Paradigm? Within the parameters defined by the industry, the following questions beg answers.
Undoubtedly, the semiconductor equipment industry has thrived by embracing an unprecedented focus on the rapid acceptance of new technology. It is therefore a necessity that the industry adapts readily to change. With the cost of technology growing exponentially, perhaps the time to look at other factors, including process reliability, is now. About the Author: Ghulam Mustafa, PhD., is Director of Engineering at Crossing Automation, Inc. He welcomes inquiries at gmustafa@crossinginc.com
|
