Tool reliability, cycle times, other productivity metrics must improve in increasingly complex fabs
As chipmaking fabs evolve, their systems must become more integrated, tying together everything from the components and subsystems to the process equipment to the factory itself. With the advent of 300-mm processing and its unprecedented levels of automation and control, the old unit-based approaches no longer work. Tool and fab must interact and communicate in ever-more sophisticated ways in order to achieve the productivity levels necessary to meet the immediate-gratification demands of the consumer gadget revolution. As the introduction to the Factory Integration chapter of the International Technology Roadmap for Semiconductors says: “All the factory components need to efficiently produce the right products in the right volumes on schedule while meeting cost targets.”
But this is no simple matter. The roadmap includes two related grand challenges facing the industry: how to respond to rapidly changing complex business requirements, and how to improve the trade-off between manufacturing costs and cycle times. Our two Hot Button participants in this issue, Mani Janakiram and Eric Englhardt, come from the largest IDM (Intel) and the largest OEM (Applied Materials), respectively. Mani has played a key role in the ITRS Factory Integration technical working group, while Eric has emerged as one of the equipment community’s leading voices on the subject. In the following contributions, they share their observations and opinions on such topics as cycle-time and work in process (WIP) reduction, tool availability and reliability, the unique challenges of high product mixes and small lot sizes, and the looming transition to 450-mm wafers.
MANI JANAKIRAM (manager/principal engineer, analysis and control technology, Intel): The semiconductor industry has been experiencing tremendous growth over the past few years. It has been consistently meeting or exceeding Moore’s Law by doubling the number of transistors and at the same time reducing the cost of production at equal proportions every 18 months. However, product, process, and manufacturing complexities are growing, and there is an ever-increasing need for improving productivity while successfully introducing new products.
The Factory Integration (FI) working group of the International Technology Roadmap for Semiconductors has been actively working on all productivity aspects in order to improve the efficiency and effectiveness of our factories. The efficiency part refers to improvements in cycle-time reduction, throughput, WIP reduction, and equipment availability and utilization. The effectiveness aspect refers to process and yield improvements in the form of reduced defects, as well as enhanced process capability and process integration.
Since productivity gains and reduced cost per function have been attained through continuous improvements, technology shrinks, wafer-size increases, and yield enhancements, the FI group will evaluate all these metrics in the form of near- and long-term technology challenges. In this way, current factories’ productivity can be improved, and future needs such as technology shrinks, introduction of EUV lithography, post–bulk CMOS manufacturing, and next wafer sizes (450 mm) can be addressed. While coming up with potential solutions, one key issue that must be addressed is the adherence to EFS—extendability, flexibility, and scalability. The cost and time to build a fab and cost and time to procure critical tools are so high that it behooves us to look for potential changes in the factory environment and build EFS into factory designs and equipment.
Another factory integration challenge is the emergence of a wide range of business models, from fabless to IDM to pure-play foundry to joint venture fabs and several combinations thereof. These models change the nature of factory operations, fostering a need for high-mix operations and fast product delivery while maintaining cost advantages. A high-mix fab does not translate into small fab sizes, since the economies of scale still favor the current fab size of 40,000–50,000 wafer starts per month.
In order to understand the needs of an efficient high-mix fab, we must evaluate factory operations based on bigger wafer lots, intrinsic equipment losses caused by setup, and the need for better process control and automated solutions. Several innovative solutions have been proposed, including rapid equipment installation using adapter plates. This innovation should reduce tool installation times by weeks to months. Direct tool-to-tool transportation techniques will improve throughput and reduce WIP and cycle times. Certain analytical methods have been proposed as potential FI solutions, including better capacity-demand prediction, preventive maintenance analysis and prediction, and cycle time–throughput relationships. Universities and consortia are working on some of these potential solutions under the guidance of semiconductor manufacturers.
The 300-mm fabs are maturing and have started to achieve the goal of a 30% cost-per-function reduction. The industry has also started to effectively address the issues involved with 90- and 65-nm process complexity. In order to keep up with technology and productivity needs, FI is evaluating the ramifications of a 450-mm wafer introduction around 2012. The focus here is to keep improving the 300-mm installed base and to evaluate factors such as small lot sizes for better cycle times; optimal wafer carriers for high mix; batch versus single-wafer processing for high mix; and efficient fab design from an environmental, safety, and health perspective. Related areas include reducing equipment setup losses, upgrading lot transport systems, improving reliability and flexibility of factory software solutions, and evaluating ubiquitous fabwide process control and yield solutions.
Once these solutions are learned at 300 mm, they can be applied to 450-mm wafer processing and fab requirements, so that we can continue to improve and meet semiconductor factory productivity needs and successfully address technology challenges.
ERIC ENGLHARDT (senior director, systems engineering, Applied Materials): Many changes taking place in our industry are starting to affect how fabs operate. Until recently, manufacturers have been able to treat each process step as a unit process, with unit specifications. If the specifications were met, it was likely that the integrated process flow would be successful. Today, increased process complexity and the unprecedented degree of interaction between processes at the 90- and 65-nm technology nodes have driven the industry to a more integrated approach to process development and implementation.
Equipment design and factory design have operated in the unit-optimization mode since the early days of the industry. While integrated process development methods are used to design the process equipment, there has been little need for integration across tools at the factory level. During the 300-mm transition, automation standards were introduced as a bridge between different types of tools. These were the first indicators that fab operation was changing from a unit-tool world to a more integrated type of environment.
At the factory level, device performance was largely optimized as a sum of independent processing units; tool characteristics were treated as a given parameter. As chipmakers and equipment suppliers push toward ever-increasing tool productivity, it has become clear that simply adding up unit-equipment performance no longer predicts factory performance. The interactions between equipment and the factory are increasingly important.
The ITRS factory integration working group was created to address this growing trend and is chartered with creating a roadmap that addresses the growing need to integrate solutions across equipment and across the factory. Initial work focused on creating new unit specifications for equipment and software that were aligned with the overall factory roadmap.
Looking forward, the working group as well as our company’s team are beginning to address broader requirements that go beyond the unit specifications and require higher levels of interaction between the equipment, automation, production control, and factory software. New trends include cycle-time reduction, increased flexibility, higher productivity with a more diverse product mix, and increased predictability.
The need for faster cycle times will drive a move to more single-wafer tools. It will also force reductions in lot and front-opening unified pod (FOUP) size to a new standard that may be as small as two wafers per pod. Automation systems with dramatically higher move rates and transaction rates are needed to meet the challenge. The need for greater flexibility in high-mix environments will also drive us to smaller lot sizes, processed by tools that can run multiple processes with little or no set-up time between varied lots. The demand for increased predictability will require improved equipment diagnostics, predictive maintenance, and a move from dispatching to integrated scheduling of all activities across the factory, from lot movement to specific equipment maintenance.
We in the industry have more collective experience with shrinking geometries than we do with shrinking lot sizes. Fortunately, the scope of the latter is less than that of many scaling challenges facing the industry.
Search | Current Issue | MicroArchives
Buyers Guide | Media Kit
Questions/comments about MICRO Magazine? E-mail us at firstname.lastname@example.org.
© 2007 Tom Cheyney
All rights reserved.