Advanced software technologies in semiconductor manufacturing
02/01/1998
Advanced software technologies in semiconductor manufacturing
Cell teams show results
Mary Swedberg
Manufacturing system technologies are having an increasing impact on complex semiconductor factories and the equipment that populates them. Understanding more about this domain will be essential for a broad spectrum of manufacturing personnel. Good management doesn`t mean knowing all the answers - more often than not, it`s knowing the right questions to ask. Nowhere is this truer than in the domain of software technology. The following offers areas to consider in planning manufacturing system technology investments.
Manufacturing applications. Actual manufacturing functionality, that is, "what the system does," attracts most of the attention. Facility managers should consider what kind of systems will be needed to support 300-mm, 250-nm (0.25-?m) manufacturing, as well as how to affordably reach that point. Some novel approaches to overall factory control hold promise for addressing the increasing complexity in semiconductor manufacturing.
Standard methods exist for integrating new software products in current manufacturing system environments. Facility managers should take advantage of the increasing number of specialty applications without discarding what they have or incurring exorbitant custom integration costs. Advanced process control is becoming increasingly important with each new process generation, and is being implemented in today`s factories. Drastic reductions in "time-to-manufacturing" for new application features are also being achieved. New technologies put more flexibility in the hands of the end users themselves, rather than requiring software specialists each time a change is needed.
System technologies. "How a system is built" is far more important and complex than most people realize. There are no easy answers or quick solutions; and continuous, long-term investment is required to address these concerns.
Facility managers must consider the importance of "nonfunctional" system requirements, such as scalability, interoperability, evolvability, substitutability, and fault tolerance. Formal software development processes play an important role in creating reliable factory and equipment control systems. Prototyping and rapid iterative development techniques may be appropriate when the goal is to refine user requirements and experiment with different candidate approaches. By contrast, formal configuration management, regression testing and detailed integration, release, handoff, and support procedures are necessary when the resulting software is destined for production. Moreover, various strategies exist for migrating from the current generation of proprietary, "trail-mix" factory systems to standards-based, open-system architectures. An incremental approach adds benefit for the manufacturing end users as well as the system development/support staff at every step. Alternatively, tackling major portions of the system infrastructure first may mean the remaining steps would be easier.
In addition, manufacturing companies should anticipate and understand such emerging software technologies as Java (the latest and most popular programming language), agent technology, object request brokers, CORBA, DCOM, and other elements of advanced distributed computing systems. The use of browsers, applets, HTML documents, and the like may improve manufacturing.
SEMATECH CIM Framework fits into this picture, and semiconductor users and suppliers can easily and incrementally adopt this technology. SEMI Standards affect these management systems today and in the future.
Other issues. A basic reluctance to change is one of the key, nontechnical risks associated with applying advanced software technology to manufacturing. Facility managers can use prototype and pilot implementations to build confidence in new solutions and their delivery/ support teams by delivering incremental benefit (rather than relying on a "big bang" approach). In addition, the manufacturing staff must be trained and prepared to make the necessary transition.
A brief case study illustrates how many facility management concerns surface and can be addressed in a major project context.
ObjectSpace has been a technical partner in the Advanced Process Control Framework Initiative (APCFI) since its beginning in early 1996. The APCFI is a $10 million, 2.5-year project involving Advanced Micro Devices, Honeywell, and SEMATECH, and is partially supported by the NIST Advanced Technology Program. The project`s key objectives include:
designing and building an open framework that enables the integration of third-party, model-based process control solutions with existing factory systems;
deploying and demonstrating a full range of user-configurable control applications to validate the approach (sensor and tool data collection, fault detection/classification, and run-to-run control in single-tool and multibay environments); and
publishing the resulting architecture and interface specifications to spread the technology to other users and suppliers.
The first consulting task was to lead manufacturing technologists (operators, process engineers, and process control specialists) and software system engineers through an intensive, multicompany "Design Fest" process, which used rapid, object-oriented analysis techniques to produce the initial, high-level design. The team had to address many of the system technology concerns discussed in this article, because of the legacy application systems involved and the open distributed computing infrastructure that was chosen for the new system. A simplified version of this design is shown in the figure.
The quality of this process and its deliverability opened the door to participation in the rest of the project - an iterative series of design, implementation, and deployment pilots. Each pilot was intended to add new system features and increase the manufacturing process coverage, resulting in a system that can be consistently and flexibly applied to a single tool, a process module, or the entire factory. The APC application capability grew from simple data collection and single tool feedback control support to a full range of user-scriptable fault detection/classification, feedforward, and feedback control sequences. Over the same period of time, the process areas supported included RTA, CMP, poly/CD, deposition, and others.
A unique aspect of this project was the publication of updates to the APC Framework Specification at various stages in the project. This specification is a public document produced by SEMATECH to foster widespread industry adoption of APC technology and is consistent in content and format with the CIM Framework, an open architecture covering the entire manufacturing execution system domain. These phases of the project have raised and addressed many more of the concerns introduced earlier, especially risk management, incremental development, manufacturing customer acceptance, skill sets, and support.
The project is now in its final phases and has delivered genuine manufacturing and system technology benefits to AMD`s Fab25. Moreover, the project has already spawned a number of related development projects within the user, supplier, and manufacturing research communities.
The pace of evolution in systems technologies has never been higher, so it is difficult to predict what the manufacturing system industry will look like in five to 10 years.
Acknowledgment
Java is a trademark of Sun Microsystems. All other trademarks are the property of their respective companies.
Alan Weber is manager of the ObjectSpace Manufacturing Systems Business Unit, located in Austin, TX. ObjectSpace, 14850 Quorum Drive, Suite 500, Dallas, TX 75240; ph 972/726-4100, fax 972/715-9099, www.objectspace.com.
Mary Swedberg is plant manager at Eaton Corporation`s Semiconductor Equipment Division; ph 508/921-0750,fax 508/927-3652.