CORDIS Archive
This page has been archived. It will no longer be updated.
September 1995
CONTENTS
Contents
1. Introduction and Background
2. Flexible Manufacturing Topics
2.9 Factory information systems
2.10 Flexibility and optimisation
2.13 Moving to 300 mm wafer size
2.14 European equipment strengths
Mr. Bomchil GRESSI Mr. Browne GEC-Plessey Mr. Bruchez Independent Consultant Mr. De Prost Matra MHS - TEMIC Mr. Doering EM Microelectronic Marin Mr. Honold Jenoptik - Microfab Mr. Laes Mietec - Alcatel Mr. Madore ATMEL ES2 Mr. Montier SGS - Thomson Mr. Noll AMS Mr. Pesneau SGS - Thomson Mr. Pfitzner FhG - IIS Mr. Schuster - Woldan Siemens Mr. Wilkens Philips Semiconductors Mr. Broster DGIII F/3 Mr. Beernaert DGIII F/3 Mr. Kelm DGIII F/3 Mr. Guedj (Rapporteur) DGIII F/3
This workshop held in Brussels on 20 September 1995 was attended primarily by managers from IC manufacturers. They presented and discussed issues and priorities for the future from a variety of points of view. This workshop was one of a series held during September 1995 in order to gather input from industry in readiness for revising the RTD Information Technologies workprogramme in the domain Technologies for Components and Subsystems.
Over the past five years the semiconductor industry has experienced a broadening of scope in terms of its key applications. Furthermore, the industry has become increasingly "customer-driven". However, the principal challenge, as in previous decades, is to sustain a fast pace of technological change whilst maintaining customer satisfaction in terms of delivering competitive products, on time, at the lowest cost and highest quality level. European IC producers have now successfully narrowed the technological gap, and achieved world-class manufacturing productivity levels.
Flexibility in manufacturing is a key issue for IC companies irrespective of whether their prime products are high-volume standard memories or modest volume application specific devices. All of the companies are involved in a continuous process of change in terms of customers, product mix, process technology, manufacturing equipment and materials. To achieve flexibility and control in manufacturing enables maximum use of working capital and responsiveness to customer's needs. Moreover, maintaining competitivity requires continual investment in manufacturing facilities, process technology and product design methods as the goal posts are forever moving. Even though the participating companies compete with their products on world markets, many of the issues in terms of facility design, equipment selection, automation and work flow control are topics on which there is considerable added-value to be gained by sharing practical experiences and ideas. In this respect the ESPRIT funded MST projects (1990-95) were a particular success() in that besides achieving production process economisation and improvements in manufacturing efficiency and productivity, they also demonstrated the indirect value to all participants of cooperative working and information sharing in the manufacturing domain. This in turn led to significant cost savings for the participating IC manufacturing industries.
Consequently, this workshop focused on flexibility in the manufacture of electronic circuits. Within this theme participants were invited to make short presentations on their current experiences and future needs, guided by the topics proposed for discussion, which included:
The discussions split into two threads. On the one hand it focused on manufacturing efficiency in a multi-product and multi-process facility, and on the other it considered high volume products in a dedicated facilities. In both cases the common goals are to achieve improved equipment reliability and availability (direct cost and performance impacts), to gain tighter control of process parameters by improving in-situ measurement and feedback loops, and to maintain overall manufacturing flow control and scheduling.
The semiconductor market is currently forecast to grow to ~$300 Billion by the year 2000. The stakes in the industry have never been higher. When investment decisions are taken it is important to complete the construction of the new fab in 18 months or less, and to have a very efficient ramp-up to volume production. One of the participants has undertaken 5 shrinks of its 4 Mbit DRAM, bringing it down from 90 to 50 mm² in five years. This form of process and product re-engineering is becoming typical as the industry forces down costs to retain competitiveness.
The level of investments in semiconductors required to meet the capacity demand has never been as high as at present. According to Dataquest, investments in Europe substantially exceed current production levels relative to the rest of the world. Manufacturers are expected to spend more on capital investment for semiconductors in the next three years than the cumulative spending over the last 37 years. In these contexts, optimal use of capital is vital and the goal of maintaining a high degree of flexibility in both the design and operation of the facilities is seen as both wise and essential.
A key element to maintaining profitability in the quickly evolving IC industry is to ensure the fastest possible production ramp-up for new processes and products. It is essential to gain maximum and timely benefits from expensive R&D and capital investment through quick traversal of learning-curves to achieve high-yields, cost reductions and reduced time-to-market. One way of achieving this is to develop new processes closer to, or in, the production environment. This approach, coupled with implementation of real-time, in-situ process control was rated highly on all the participants' priority lists.
Moreover, the industry is increasingly practising what is known as "concurrent engineering", where development and characterisation of processes and products are tackled in parallel. This can provide substantial gains in terms of time to achieve production status compared to the more serial development practices of the past. It does however, imply larger numbers of engineers engaged in tasks of a "developmental" nature.
Many of the equipment discussions mirrored those held in the Equipment and Materials workshop already held in the same context(). It is clear that improving the overall level of equipment efficiency, availability, and productivity represents the major challenge for semiconductor manufacturing. These factors directly impact overall factory productivity, causing cycle-time inflation, yields reduction and late customer product delivery. Moreover, in excess of 50% of an IC fab costs are equipment depreciation related.
One of the workshop participants specified an overall equipment utilisation target at 70% within the next 5 to 7 years. This figure might appear to be low but it takes into account that there is a considerable discrepancy between productive utilisation figures for non-vacuum equipment and the typically lower figures presently experienced for vacuum equipment. This latter factor raises concerns in view of the rapidly increasing complexity of vacuum processing equipment (e.g. cluster tools) which are being used more and more in production lines.
Stable process tools, minimisation of test-wafers and high equipment up-times, coupled with lower mean-times-to-repair have become key factors in equipment selection. Moreover, to ensure that process-flows are not impaired, "smart" predictive maintenance schemes are highly desirable. Questions raised on this topic included achievement of integrated factory automation, provision of appropriate software tools, standardisation of equipment data interfaces, improved staff training schemes, ergonomic management / factory information system interfaces and accumulating overall impact through refined cost of ownership models.
These discussions were further reinforced by a presentation of a "Total Predictive Maintenance" programme, currently implemented at one of the participating companies. Key elements are:
Equipment interface standards must both improve and be implemented in order to achieve more capable and efficient data communication with cell-control, shop-floor-control and management / analysis systems. There is a growing tendency towards introducing SMIF interfaces to equipment being installed in the new fabs and in this respect Europe is strong and well represented.
Further considering the question of factory automation, in particular because the IC industry can generate "million of information items/lot", it was clear that although manufacturing execution systems have been in use for at least 15 years, there is no single package which integrates all functions needed in the shop floor into a single logical model. There is now a clear move away from the monolithic legacy systems of the past towards open, flexible architecture systems based on a framework established by SEMATECH. Integration of old and new systems with locally engineered data analysis tools across "live" multinational manufacturing facilities requires considerable effort by the industry, a large support staff and thus an inherent high cost-of-ownership factor.
Increasingly detailed data coupled with complex data analysis causes sluggish query response times. Acceptable response times would be on average 2 seconds and <4 seconds for 95% of queries. In the future, standard format, event-driven message busses will assure message transport. Standardised application programming interfaces will bring flexibility, real on-line event processing, and easier maintenance procedures.
Participants reported that during the past five years work in Europe had progressed, notably in terms of implementation of statistical process control tools, cell controllers, scheduling techniques, and improved interfacing to management systems. However, the full replacement of existing systems with homogeneous, integrated open systems will take several years to complete.
The workshop included presentations on "multi-product wafer service" and the "multi-product wafer train" from two ASIC companies. These presentations provided good examples of customer-oriented manufacturing, whereby customers could trace the progress of their own product along the fab-line. It was claimed that such an approach can drive improvements in cycle-time and inventory control with positive impacts on yields and costs.
The discussion which followed served to highlight the fact that different businesses require different forms of optimisation. The ASIC manufacturer has to achieve process stability in a manufacturing operation tuned to offer flexibility in terms of widely differing product volumes and process mixes alongside a fast turn-around prototyping service. However, flexibility is also a key issue for a volume manufacturer who needs to be able to manage fast ramp-up (and ramp-down) of high-volume products.
It was reported that for small-to-medium volume ASIC business the use of single-wafer processing and integrated, standardised cluster tools can help to achieve flexibility, reduced cycle-times and potential cost reductions. However, there is an additional requirement to install good software support for batch sorts/merges and splits. ASIC businesses also place strong emphasis on cycle-time reduction which brings thermal and cleaning processes under scrutiny as well as improved metal and inter-metal processing to maintain good yields.
Wafer cleaning was also identified as an area with potentially high gains in processing time. It was suggested that maybe wafer cleaning was done too frequently. Improved understanding of the cleaning requirements and mechanisms could result in a reduction in cleaning steps with an inherent impact on yield and cost reduction. This area needs further study to identify yield vs. costs trade-offs.
No workshop discussing manufacturing and equipment would be complete without a discussion on 300 mm wafer size issues. Timing of the realisation of leading-edge fabs is still clouded with a degree of uncertainty, but sometime during 1998 seems most likely. The operational cost ratio in moving from 200 to 300 mm is currently estimated at 1:1.8, with an estimated revenue per wafer ratio of 1:2.5. None of the workshop participants indicated specific plans for their own companies but all are studying the matter closely. Points which were discussed included single wafer processing versus batch processing systems, cost of ownership models, general equipment and interface specifications, overall cost increases versus die/wafer expectation, etc. There seemed to be good agreement concerning the need to establish a working group / forum to keep European semiconductor manufacturers and equipment makers in close dialogue and aware of the developing international situation. Whilst equipment makers will have to immediately devote development effort to the 300 mm tools, the semiconductor companies who are not involved in the first wave of 300 mm fab constructions will have a short breathing space.
Recognised European strengths in equipment are, notably, lithography, sputtering, epitaxy, rapid thermal processing, SMIF/mini-environments, metrology & analytical, and assembly and packaging. The familiarisation and selection of European sourced equipment is increasing. The launch of the SEA programme by the EC is recognised as a step in the right direction. SEA is also drawing interest from equipment users in other parts of the world, some of them, notably from the US and Korea, are already participating. SEMATECH has also expressed interest in participating in assessments.
Finally, the rapidly growing concern for environmental issues poses some serious questions on the equipment side, notably how to reduce consumables and treat waste. Taking an active interest in these issues, ahead of stricter environmental regulations is considered as paving the way for cost-savings in the long run.
A great deal of material was presented and discussed during the workshop. Clearly, manufacturing issues remain as important to the IC industry as ever. It became clear that achieving flexibility is common interest of semiconductor manufacturers irrespective of the size or nature of their business. In short flexibility should be considered as an integral part of manufacturing.
Companies appear ready, willing and able to continue to exchange information on manufacturing issues, as they have over the past five years, for they can easily identify the added value of such exchanges.
It remains to define an appropriate forum to catalyse these exchanges. In this respect, semiconductor industry working groups are seen as a possible means of implementing an open information exchange between companies on a range of selected topics.
The themes which attracted most common interests and where it was indicated that cooperation would be most beneficial are listed below :
The URL for this page is /esprit/src/tcsfmws.htm
It was last updated on 25 October 1996, and is maintained by John Magan - john.magan@dg3.cec.be