APPENDIX C. SITE REPORTS Site: Fujitsu (Yamagata Plant) 1-6-1 Marunouchi, Chiyoda-Ku, Tokyo 100, Japan Date Visited: October 6, 1993 Report Author: M. Pecht ATTENDEES JTEC: P. Barela W. Boulton G. Harman G. Meieran M. Pecht HOSTS: C. Handa President, Yamagata Fujitsu T. Tsuchimoto Board Director, General Manager of Technology Group, Kawasaki Facility M. Nakazono Director and General Manager, Yamagata Fujitsu BACKGROUND Fujitsu manufactures and automates to the level needed to reduce costs and improve quality incrementally; nothing more, nothing less. Our JTEC group spent a half day at Yamagata Fujitsu (Fujitsu's Yamagata plant). The main technology that we came to see was Fujitsu's high-volume surface mount operations that are part of its manufacturing process for hard disk drives used in PCs and workstations. We were greeted by Chiaki Handa, President of the Yamagata plant, and by T. Tsuchimoto, Board Director and General Manager of the Technology Group at the Kawasaki Facility, who gave a general overview of the operations of Fujitsu. Then M. Nakazono, Director and General Manager of the Yamagata plant, gave the Yamagata plant overview. Altogether there were 12 managers from three different Fujitsu facilities (Yamagata, Kawasaki, and Atsugi). Fujitsu has some 71 overseas manufacturing and assembly-and- repair subsidiaries, in addition to numerous sales and other offices. It currently lists 161,974 employees worldwide. (Thisnumber does not include employees from Amdahl and other similar affiliates in Europe.) In addition to manufacturing computers, Fujitsu has entered the multimedia market and has been chosen by U.S. common carriers for broadband ISDN equipment, switching systems, optical transmission systems, and cellular mobile phones; it is offering similar equipment in parts of Asia, Australia, and Europe. It is also moving into software and other computer services around the world. YAMAGATA FUJITSU Yamagata Fujitsu is located in Japan's industrial district. Established in 1983, it had 100% investment by Fujitsu; 70% of sales go overseas. As of the JTEC visit, approximately 300 of 1,000 employees were staff, approximately 700 were operators; about 60% were male and 40% female; their average age was 24. The plant developed, produced, and managed sales of disk drives. The JTEC group The JTEC group toured the printed circuit board (PCB) assembly line (PCA) for 3-1/2 inch disk drives and the 2-1/2 inch hard disk drive assembly line (HDA). Printed Circuit Assembly Line for 3-1/2 Inch Disk Drives Yamagata Fujitsu's PCA line incorporates traditional semiautomatic component placement, soldering, and cleaning equipment with manual part and assembly feeding of machines. There was some manual part soldering and a lot of manual inspection, testing, and repair. The reasons given for all the manual work were (1) it was the most cost-effective considering the mix of PCA technologies they needed to handle, (2) there were many odd parts, such as special connectors and heat sinks that didn't easily lend themselves to automation, and (3) product demand was low, so that only two shifts were needed and further automation would not pay off. Over 600 pieces per minute could be manufactured. The components were predominantly plastic. Generally 60% of the boards were composed of SMT; 20% of these were single-sided and double-sided boards, 45% were four-layer, 30% were six-layer, and 5% were eight layers and up. In general the PCA process line was not by any means state-of-the-art and may not even be highly profitable. It was stated that the demand for these products was down. The 3-1/2 inch drive had a lot of manual handling: for insertion of some components, for rework, and for inspection. Burn-in equipment at the plant operates under the following conditions: 1. There are some fans covered with stainless steel mesh at the bottom of this equipment to circulate the inside air. 2. There are no heaters inside the equipment. As a number of hard disk drives are operating within a closed box, temperature inside the box increases significantly because of heat radiation from the disk drives. As a consequence, all hard drives kept in the box are burned-in under this high temperature even with no heaters. 3. In case of a perfectly sealed system, temperature becomes too high, so, a slit is put on the top side to maintain the hard disk drive's temperature at a range between 67 degrees centigrade and 73 degrees centigrade. The location and size of this slit are based on Fujitsu's know- how. The cleaning operation was OK, but there was so much handling after the cleaning with bare hands that it seemed as if the cleaning was somewhat superfluous. Of some special interest in the PCA manufacturing line was the removal of CFCs in October 1993 and the subsequent use of non-CFCs, replaced by a deionized water cleaning process. While there was no research being conducted in leadless solder, management was aware of efforts in Germany and did look at InSn, but considered it too expensive. The small FR-4 boards (to be mounted in hard drives) were demarked and partially cut out but remained in the larger mother boards for handling. Each mother board contained four or six smaller active boards. They were loaded with pick-and-place machines (said to have a placement accuracy of 50 micrometers - at other sites we saw machines that were more accurate, but 50 m was apparently adequate for the purpose). The boards were reflow soldered. In summary, the printed circuit line for 3-1/2 inch discs was "low tech": the machinery was old; the process was a combination of automated and manual assembly; there was lots of inspection; cleanliness was minimal; the burn-in was primitive. It was clear, however, that the important processes were well developed, and all the things we saw that seemed "low-tech" were indeed relatively unimportant (e.g., the burn-in caught what it needed to catch, and having a more sophisticated burn-in system was unnecessary, and hence no attention was paid to it). Thus, the operation appeared efficient for what it was doing, even though it was not elegant. It was obvious that Fujitsu has no intention of significantly improving this line. It was also obvious that Fujitsu was going to use what it learned on this line to make a step-wise improvement on its 2-1/2 inch drive assembly. It should be noted that the Yamagata plant is dedicated to manufacturing products for personal use, different from any other Fujitsu plant, such as its Numazu plant, which manufactures high- technology products. The main focus of the Yamagata plant is to utilize the existing equipment to a maximum extent to promptly manufacture high-quality products and at the same time to seek better production efficiency at a lower cost. As pointed out to the JTEC team, the Yamagata plant uses traditional manufacturing methods through semiautomated equipment and manual work on its printed circuit board assembly lines. While installed machines at the Yamagata plant such as screen- printer, parts-mounter, and air-reflow oven may be obsolete from the technical point of view and give the impression that the Yamagata plant continues to use old equipment, our hosts noted that the use of this equipment is optimal from the point of view of efficient investment. The plant does utilize an automatic appearance inspection tester for any minimum-size parts such as QFP (quad flat package) with 0.5 lead-pitch and 1005 chips, which are almost impossible to inspect manually. On the other hand, inspection and repair of any parts other than those minimum-size ones are carried out by manual tasks to maintain efficiency in capital investment. Hard Disk Assembly Line The hard disk drive assembly (HDA) line was highly automated and the equipment was modern and sophisticated. The environment was a 100-class clean room (not great by semiconductor standards, but certainly sufficient for the operation, especially since extra precautions were taken on the equipment where the magnetic heads were physically present), with automated testing of over 100 drives simultaneously. There was complete robotic handling of parts. The operators all wore appropriate clean-room clothing. This line was being prepared for high-volume production. The hard disk assembly operation was a model of efficiency. The entire operation was automated using Fujitsu-built robots and required attention to technology and detail. The striking difference between the PC board assembly area and the hard disk area was attributed (by us) to the fact that the disk was the guts, the family jewels, of the Fujitsu system; hence, no expense was spared to make this as efficient and low- cost as possible. New Technology We saw very little in terms of new technologies, but our hosts did mention various technical activities in other parts of Fujitsu, including 16-to-42-layer PC boards, 60-layer ceramic laminates used in its supercomputers, and the newest parallel supercomputer using GaAs chips. (These are mounted in single-chip packages that are then mounted on ceramic boards in order to get around the known-good-die problem.) Fujitsu was using opto- connections for external connections but using electrical connections inside. It developed its own router for these boards. Our hosts also mentioned a new (but published in MRS in 1992) photosensitive dielectric film using a blend of polymers that can be used for cheap MCM-D circuit boards. Line widths were 30 micrometers, and vias were claimed to be about 20 micrometers in diameter. They used Al metal lines and made up to four metal and four dielectric layers. They said this might be in production within two to three years. Our hosts said they were following the work in BGAs including the OMPAC by Motorola and some products by AMD. They were conducting some work in flexible PCB manufacturing and assembly, flip chip mounting and TAB mounting, with plans for bare chip assembly on a computer product (server) in 1994. Quality and Reliability Quality and reliability were assumed, but never initiated as a topic of discussion by our hosts. Their policy is summed up as "quality built-in," with cost and performance as prime considerations. Our hosts noted that today's components do not require burn-in, although at the board level there was some concern for solder bridges. In February 1993 Yamagata Fujitsu became ISO 9002-certified and at the time of our visit was expecting to be 9001-certified later in 1993 - the purpose is to be able to sell products in Europe. Management Philosophy As in other Japanese factories the Yamagata plant management paid attention to running equipment well, to continuous improvement, to cost reduction, and to size reduction. No magic technology was employed; just fruits of a detailed study and execution of automated processing and smooth-running, efficient operations. SUMMARY Fujitsu is motivated to improve circuit board density and reduce cost. The improvement in density had several implications; for example, to reduce package thickness, to improve mounting location, and to reduce weight. Reducing cost was the prime driver after functionality and customer needs were accounted for. To meet the objectives, Fujitsu was going to use flexible printed circuits, improved soldering technology, design of thinner packages, and specially plated through-holes. None of the technologies are revolutionary; rather, they are the straightforward application of continuous improvement to Fujitsu product lines. A Fujitsu representative aptly stated, "Indeed, we are still using an old-type equipment for the printed circuit board assembly line for 3.5 inch disk drives. We would like to stress, however, that the use of the fully automated equipment does not always represent 'high technology.' Through our experience over years, we have come to the conclusion that in certain cases manual work is sometimes more efficient rather than automated machines, as other vendors do in some cases." Site: Hitachi PERL and HIMEL 292 Yoshida-cho, Totsuka-ku Yokohama-shi Kanegawa-ken 244, Japan Date Visited: October 6, 1993 Report Author: L. Salmon ATTENDEES JTEC: P. Barela J. Kukowski L. Salmon R. Tummala HOSTS: Dr. Hiroyoshi Matsumura Mr. Ikuo Kawaguchi Mr. Toshiro Kinugasa Mr. Ryugi Nishimura Mr. Masahiko Yatsu Dr. Norikazu Tsumita Dr. Toshio Asano Mr. Takanori Ninomiya Dr. Hisashi Sugiyama Dr. Makoto Iida BACKGROUND Hitachi has nine corporate research laboratories with approximately 5,700 employees. The corporation spends approximately $3 billion per year on research and development. This amount corresponds to approximately 10.5% of total sales for the company. The first corporate R&D laboratory, the Central Research Laboratory, was established in 1952. We visited two of the corporate research laboratories, the Production Engineering Research Laboratory (PERL) and the Hitachi Image and Media Systems Laboratory (HIMEL), which are both situated at the Yokohama site together with the Systems Development Laboratory. PERL is a corporate research laboratory and has as its major goal research in the areas of advanced packaging, precision machining, factory automation, high-density magnetic media, and advanced semiconductor materials and processing. The laboratory has approximately 600 employees and was established in 1971. HIMEL is also a corporate research laboratory and is chartered to perform research in technology areas of importance to camcorders and other optical consumer products. The laboratory has approximately 400 employees and was established in 1972. The laboratory was originally an associated division of the Consumer Products Group, but was designated as a separate research laboratory in February of 1993. Its purpose is to work closely with the production divisions to provide technology required for new products. HIMEL is divided by products into four departments: projection displays, range finder video cameras, LSI design and fine mechanism development, and fax/word processors. Hitachi's philosophy regarding research is called Tokken and is based on tight coupling between production and R&D efforts. Connections between production and R&D groups include personnel exchange, common planning, and frequent meetings to assure coordination. The charter of each of the organizations in Hitachi is summarized in Table PERL.1. This chart is idealized and is not always rigidly followed. Research staff will sometimes work at the development facility for extended periods during technology transfer. In other cases, the development staff may come work at the research facility for an extended period. Table PERL.1 Summary of Organizational Responsibilities Mr. Kinugasa from HIMEL first gave a presentation on Hitachi's design and manufacture of camcorders. He began with presentation of a roadmap for reduction in camcorder size through integration of camcorder functions into fewer LSI and fewer discrete components. The path he outlined shows an evolutionary approach to reducing the number of components and does not hope for revolutionary changes in IC technology for at least the next four years. He projected continued use of hybrid and PCB technology to package the components and exploitation of shrinking feature sizes in IC technology. TAB and bare chip technology are not part of the current packaging plan because these technologies are more expensive to produce. Current PCBs are designed with a 0.5 mm pitch. Efforts are under way to decrease pitch size to 0.4 mm and 0.3 mm, and the technologies will be included in products if the cost is not too high. Hitachi is developing optical inspection techniques for 0.3 mm technology because x-ray inspection is blocked by the large number of metal layers on the PCB. Inspection for 0.3 mm is currently performed manually. The most advanced SMT process Hitachi has developed is 0.3 mm pitch on a six-layer PCB that is 1.6 mm thick. The solder paste screen is made using an Ni additive process. Super solder is not used because solder deposition was found to be too uneven. Mr. Kinugasa indicated that packaging accounts for 20% of IC production cost, and he anticipates that the percentage cost will continue for new products. The major impediment to dramatic reduction in camcorder size is the size of the recording tape and associated mechanical parts required to read and write to the media. During a tour of HIMEL facilities, our hosts indicated that HIMEL is seeking to reduce discrete component size below the 1005 size part. They also indicated that the plan for HIMEL includes development of the technology required to build a pocket-size, all-solid-state camcorder. Our hosts suggested that some effort is being expended to investigate bare chip technology using conductive adhesives in order to address environmental concerns regarding lead-based solders. The JTEC team was given a tour of the clean-room facility used to pursue research in the areas of advanced disk drive media and thin film MCM development. The facility is approximately 10,000 ft(superscript 2) and is class 100/1,000. PERL is also investigating high-density magnetic media and advanced thin diamond-like coatings to protect hard disk media. Research on magnetoresistive magnetic storage is carried out at a separate Hitachi laboratory. The thin-film-on-ceramic MCM process is being pursued to support high-performance supercomputer applications. The technology provides excellent thermal properties and is designed to dissipate heat from high-power ICs. Dr. Inoue indicated that the technology will be used for the next generation of high-end supercomputer, but that it is unclear what technology will be used for packaging for the generation after that. He stated that thin-film MCM technology may be important for workstations, but that the cost of the process will have to be decreased significantly before the technology can be used in that area. Current research efforts are concentrating on cost reduction in the process, but there are no immediate plans to use the technology in Hitachi workstations. We were told that this technology would not be used for consumer applications due to its high cost. We were given a tour of the facilities for several projects at PERL. There is a strong effort to develop the technology and equipment required to automatically inspect and evaluate the quality of products and processes. Successful equipment and techniques developed in the laboratory are immediately implemented in the manufacturing line. PERL makes extensive use of image processing to decrease highly accurate inspection of high-definition CRT screens. PERL is also active in developing systems to automatically inspect green sheets for multilayer ceramic packaging and solder quality in SMT reflow processes. Scientists at PERL have developed an advanced PCB that they call TAF-II. The process uses an additive metalization process that results in improved resolution: line widths as small as 140 microns in 35-micron-thick copper. The process flow for the TAF- II process is shown in Figure PERL.1. The lamination, drilling, and catalyzation steps are similar to those in standard PCB processes, but the following steps differ from standard processes: Solder resist is formed through an etched stencil; the plating process is an electroless plating step in a bath with a pH of 12. The PCB typically contains 4-6 layers, but the PERL researcher claimed that the process would support as many as 12 layers. He also claimed that the TAF-II process is 20% less expensive than the standard PCB process because less plating solution is used in the selective plating process. The TAF-II process was licensed to several companies in 1989, and a new process, TAF-III is in development. TAF-III will use lithography to define the solder resist layer and will have line widths as small as 80 microns. Figure PERL.1. Brief process flow. Dr. Makoto Iida presented work on thin molded plastics for laptop computer cases. He showed research on methods to reduce the thickness and mass of the plastic required for a domestic Japanese portable PC product, describing a roadmap for thinning body plastic thickness from 3 mm in 1985 to 1.5 mm in 1993. PERL researchers are currently working to reduce the size of the plastic case by improving structural design of the case, the polymer material used in the case, and mold design. Hitachi has developed its own polymeric material, together with an associated company. One of the greatest challenges is to control flow of the resin during filling of the mold. PERL researchers have analyzed resin flow using a simulation program to predict the best method to inject resin into thin molds. Dr. Iida stated that although composite materials can be made thinner, they are only used in higher-value products such as computer pen pads, due to the high cost of the material. Composite materials can be molded as thin as 1 mm and weigh less than conventional plastics. Overall thickness of the portable product was also reduced by using the LCD display as a structural member. A shock-absorbing material was used to assure that LCD reliability remained high. Dr. Iida indicated that PERL is investigating molded circuit boards, but present cost of materials is very high. He indicated that the thermal expansion characteristics of the molded circuit board material are very important. Dr. Iida then described a rework tool that has been developed to add or remove solder from SMT parts with a pitch as small as 0.3 mm. The tool removes excess solder using a braided wick, and adds solder using a novel technique. A droplet of solder is held by surface tension to the end of a heated point and the solder is placed on the unsoldered joint. The solder and the lead are heated in a single step. There were several general comments made during the JTEC visit to Hitachi that are worthy of note. First, although a high level of automation is evident in the production philosophy of Hitachi, individuals at PERL emphasized the importance of an "appropriate" level of automation. They stressed that overdependence on automation is as detrimental as underdependence on automation. They further stressed the importance of the skilled individual to improve production yield and to retain production flexibility. The managers at PERL indicated that process floor workers should have a college education. When asked, they agreed that it is difficult to attract high-caliber people to production tasks, but indicated that showing employees the importance of their contribution to production is the best method to attract them. A second theme was the importance of cost in all decisions regarding technology. Cost was clearly a critical consideration in every discussion regarding proposed technology. There was always a clear product path, and the cost of a technology was compared to a projected, required production cost. One statement made by Dr. Matsumura summarized this attitude. He said, "We used to determine the price of a product by adding an acceptable profit margin to the projected cost to produce a product. Now the price is a predetermined quantity, and we add an acceptable profit margin to arrive at the required product cost." Site: Ibiden Co., Ltd. 3-200, Gama-Cho, Ogaki City Gifu Prefecture 503, Japan Date Visited: October 5, 1993 Report Author: J. Peeples ATTENDEES JTEC: M. Kelly G. Lim N. Naclerio J. Peeples HOSTS: Mr. Hidetoshi Yamauchi Corporate Overseas Planning Office Mr. Kazuhisa Ohno Manager of Sales and Marketing for the Overseas Affairs Division Mr. Osamu Fujikawa Director of Technology and Development Mr. Koji Hosada General Manager of Overseas Marketing and Sales of Electronics Mr. Keiji Adachi Sales Engineer, Overseas Marketing and Sales of Electronics BACKGROUND Ibiden began life as a power company and is now a manufacturer as well. It offers products in the following three areas: IBI Electronics Branch -double-sided/multilayer/special boards (small) -IC packaging based on the special board technology -continuous processing -up to 12 layers -width traces in the lab -small vias and blind/buried vias -flip chip on board (COB) for one customer -aluminum core substrates for power dissipation IBI Inorganic Chemical Branch -calcium carbide/carbon/acetylene for the steel industry -fine-grain graphites -electrodes/fixtures -ceramic fibers -IBI, wool, insulator IBI Building Materials Branch -IBI board - laminate material for home construction and decor nonflammable construction materials Ibiden shares its product branches between manufacturing plants. Any given plant may be involved in the production of electronics, building materials, inorganic chemicals, and/or new products. Generally, however, one or two plants specialize in production for a particular branch. R & D ACTIVITIES IBI-Techno is the calling card of Ibiden's R&D activities. Advanced product development is conducted in the New Products Development Branch. When ready, the technology is fused into one of the major product branches. The development of plastic pin grid arrays (PGAs) and application-specific integrated modules (ASIMs) were discussed as examples of this advanced product development process. Ibiden discussed the ASIM concept with me during my April 1990 visit. I could not detect that the concept was any closer to product at the time of the JTEC visit in late 1993. Fundamental technology development is the charter of the Technology and Development Department. Activity in this area includes investigation of biotechnology, superconduction, ceramats, and optical materials. Ibiden Electronics is working mainly in two product areas: IC packages and printed wiring boards (PWBs). IC Package Products Ibiden has leveraged its competency in laminate substrates technology into an IC package business. It makes small interposer substrates with facilities for chip attach (plated wirebond pads, semi-cavities, etc.) and add pins. Its major product line, by far, is the plastic pin grid array (PGA). It is very intense in this product development area, which is driving performance as well as costs. Our hosts showed examples of plastic PGA substrates in the 2 nH inductance range. Ibiden has chosen to get out of the TAB package business because it feels that it was unable to compete with the major suppliers of TAB packages. It has just begun prototype production of BGA; it has one package of over 500 pins. Our hosts characterize this program as being very U.S.-centric but realize that some devices may have to be provided to U.S. end users by Japanese semiconductor vendors, and it therefore must do the package development. Substrate Products Ibiden is a major supplier of PWBs. It produces subtractive laminate substrates at a rate of 25,000 square meters per month of double-sided substrates and 15,000 square meters per month of multilayer substrates. It is also active in several more advanced substrate areas, alternate materials for CTE and power dissipation, and additive processes for higher routing densities. Topan, which began life as a printing company, is a very important Ibiden competitor in laminate substrates. Yamamato and Compaq are also viewed as important competitors, especially for U.S. business. Samsung is a licensee of IBI-Techno. Ceracom is a ceramic-cored PWB and is Ibiden's "low-cost," low- expansion substrate for flip chip. Ibiden currently single- sources Ceracom substrates but is interested in supplying the raw material and technology to U.S. substrate suppliers. It is not interested in supplying end products (substrates) of Ceracom to the United States. Ceracom currently has two major Japanese customers and few applications. Ceracom costs are between thin film metalized ceramic and FR-4 at the finished circuit board. Material cost is higher than ceramic substrate and FR-4. Ibiden's additive process is still in the prototype stage of development. Development seems focused in the areas of adhesives, dry film plating resists, and plating technology. Plating technology is critical to the additive strategy. Ibiden is doing the work to ensure thickness consistency and peel strength. Thickness consistency is not a given, even in the electroless process. Peel strength is the major reliability concern. Ibiden's adhesive resin for additive process consists of two different epoxy resin systems (one is solid and the other is liquid). These resins show different strength against some chemicals by having different hardeners (e.g., acid and amine). This process additionally mitigates cracking in the high-stress regions of small vias. Peel strength is also a bit of a "specsmanship" issue in that strength levels have been characterized as appropriate to the older, much more macroscopic technologies, which may not extrapolate reasonably to the very-fine-pitch additive products. Ibiden is also investigating what can be done with solder resist height to make assembly of additive boards easier. Its researchers think that a product with a resist 20 mils higher than the copper trace can eliminate solder bridging. Ibiden's build-up multilayer PWB product is still in development. Build-up is intended to be similar to IBM's surface laminar circuit (SLC). It is anticipated to be a double-sided subtractive core with two build-up fine-line layers on each side of the core (a total of six layers). Build-up will have a very high routing density due to the 50 m conductor widths and the 100 m vias. It is currently about five times the cost of a normal laminate substrate but about eight times less than an MCM-D substrate. Laminate substrates are the cost target for this technology. The major changes in Ibiden's MCM-D substrate technology are the move away from polyimide to an epoxy dielectric for cost reduction and the deployment of a "hot press" aluminum nitride core for enhanced thermal conductivity and dimensional stability. This extra stability results in a 1 mil line and 1 mil space capability for wiring. Ibiden feels that the $6-8 subtractive board of today would cost $40 in the MCM-L build-up technology, but sees a path to drive that to $15. The same circuitry would cost $300 today in a copper/polyimide MCM-D substrate or $120 for a copper/epoxy implementation. GAMA PLANT TOUR Mr. Fujikawa gave us a tour of the package production at the Gama plant. He feels that the current staffing level may be too high at about 300 persons, of which 20% are considered indirect. The plant was somewhat eclectic in layout and was literally covered in paper: SPC control charts with goal lines extended into 1995 were everywhere. There were safety and "one-point" displays in every section, as well. The "one-point" concept is an interesting one. A cache of visual aids is kept in each section from which the section supervisor or engineer can present a "one-point" discussion daily to the workers. The subject will be a very specific commentary on some facet of their job, like the correct way to use a tool or the details of why a particular process step is performed. There is no effort to tie the daily events together; the intent is rather to expose the workers daily to some level of very specific training. The IC plastic package substrates and the Ceracom substrates are processed on very similar, if not the same, production lines. Drilling and routing of the package substrates takes place in a single room. Production control appears to be totally paper- based. There were paper travelers with every lot of material, and no bar code readers were seen to be integrated into any of the production steps. The design area was a large, well-lighted open office space that housed about 50 uniformed Ibiden and contract designers. Mr. Fujikawa again mentioned his concern that overhead support was duplicated within the various departments of the facility. A major distinction between the package business at Ibiden and its substrate business is that the package business can focus on a relatively small customer set: the Gama plant deals with tens of customers, whereas the substrate plants must deal with five hundred to a thousand customers. AOYANAGI PLANT TOUR Ibiden manufactures laminate printed wiring boards at its Aoyanagi plant. It provides layout services for about 60% of its printed wiring board orders. This plant employs about 1,500 persons and exports about 30% of its production to Europe, the United States, and Singapore. This is the most active of all the IBI plants. It seemed busier, more disorganized, and a bit dirtier during this visit than it did during my visit in 1990. There were definitely more people in areas that were essentially fully automated during my last visit. Work Environment While safety awareness was apparent, there was no evidence of safety requirements. The smocking that did exist seemed much more focused on protection of the product (e.g., hats and masks to keep the product clean) than on human safety (e.g., lack of eyewear). The factory is by no means a "smoke-free environment." As in all Japanese factories, no street shoes are allowed. GENERAL OBSERVATIONS AND CLOSING REMARKS Below is a list of general observations and impressions: Primary electronic products are PWBs and IC packages. Package products include COB, PGA, and plastic leadless chip carriers. Product focus is on higher performance through higher density. Ibiden is keeping R&D investment flat in spite of the Japanese economic slowdown. R&D focus is on alternative materials and processes. Ibiden provides product design service or can receive design data from its customers. Ibiden provides low-volume or high-volume products and services. Cost per product is volume-dependent. Significant growth is expected in fine ceramics. Ibiden designs and builds much of its own production equipment. Environmental impact is a stated concern. Some waste is managed on-site; some is transferred to specialty companies. Economic slowdown resulted in excess capacity and underutilization. While Ibiden does not expect to be fully utilized for some time, it is optimistic that it has a bright future due to its investment in technology. My summary impressions of Ibiden are very aligned with my overall impression of the visit to Japan. Having visited some of the same facilities and having met with some of the same people three years prior to this trip, I was very impressed with the company's consistency of technology and product strategy. Ibiden seems to understand its core competency very well, although our hosts never spoke in those terms. Ibiden Electronics does laminate PWB well. It is willing to drive that technology to new product opportunities (plastic PGAs), align it with other Ibiden technologies for new product opportunities (with ceramic for Ceracom), augment it for new product opportunities (with novel surface science for additive processes), and repeat the process (new ceramic cores and additive processes for the build-up MCM-D products). I did sense a lack of visible progress. The laminate facility appeared very much as I had seen it three years earlier. Process control and worker communications technology seemed stagnant as well; however, neither of these seemed to limit the capability of the plants. Site: Matsushita Electric Industrial Co., Ltd. 3-1-1 Yagumo-Nakamachi, Moriguchi Osaka 570, Japan Date Visited: October 6, 1993 Report Author: J. Peeples ATTENDEES JTEC: M. Kelly G. Lim N. Naclerio J. Peeples HOSTS: Keith Nishitani GM-Business Coordination, Overseas Dept. of the Corporate Technology Mgmt. Office Dr. Yamazaki Director, Circuit Manufacturing Technology Lab., Corporate Production Engineering Division Shunichi Murakami GM, Process and Materials Development Dept., Corporate Production Engineering Division Mark Kawai GM, Control and SW Development Dept., Corporate Production Engineering Division Hiroaki Fujimoto Senior Engineer, VLSI Technology Lab, Semiconductor Research Center Yoshihiro Bessho Engineer for Materials and Components Research Lab, Components and Devices Research Center Hiroshi Asai Manager, Marketing Department BACKGROUND Our visit was divided into three distinct sessions: a tour of the Circuit Manufacturing Technology Lab of the Corporate Production Engineering Division, a tour of a chip size inductor factory, and a discussion on chip mounting technology. The JTEC visit began right where my visit in 1990 had ended, in the Circuit Manufacturing Technology Lab, one of five subdivisions of the Corporate Production Engineering Division. Mr. Murakami discussed the process development plans. Some of his handouts were identical to those I had been shown in 1990, only updated to show progress. Panasonic, like most of the Japanese companies I visited, seemed to have a remarkable degree of strategic technological consistency. My opinion is that similar U.S. companies are less likely to invest long-term in specific technologies, but would rather jump to new "promising" alternatives. Mr. Hiroshi Asai, Manager of the Marketing Department, gave us a rough sketch of Matsushita: -$60 billion company composed of 252,000 people -R&D budget of about $3 billion (5.8%) flat for five years -17 Japanese locations housing 30 organizations -12 Divisions (including Panasonic FA) -22 overseas manufacturing plants -component group revenue is $300 million/month from 80,000 product types We were each presented a copy of My Management Philosophy by Matsushita's founder, Konosuke Matsushita. It is 66 pages full of gems of wisdom. Following are some of the quotables: -Good times are good but bad times are still better. -It is my hope that managers would use their knowledge as expert economists [to] work for measures that would be in the best interests of the people and the nation. -Foreign expertise may still have a role to play but more important now is that Japan develop its own technologies. -My proverb about management says that if we fight a hundred wars, we should win a hundred victories. -You pray for the survival of your rival because you want another chance to demonstrate your superiority. R&D ACTIVITIES The Circuit Manufacturing Lab is the process and equipment development arm of Panasonic Factory Automation. It develops whatever is necessary to service new products. Much of the work in process is for palmcorder or for flat panel applications. We were told that about 20 engineers may be involved for three months to transfer a new FA line into production. The lab we visited is set up for touring. A graphic arts display emphasized the importance of the Panasonic solder paste development to its fine-pitch (0.3 mm in production) stencil, place, and reflow process. The key point was that Panasonic has achieved a more spherical solder particle shape than normal, allowing the paste to have more predictable flow characteristics. Our hosts were excited about this and were similarly excited about their in-house-developed conductive epoxy; neither of these formulations are available for sale in the United States. This lab exhibited the latest equipment technology for FA. New since my 1990 visit is the emphasis on lower-cost methods of direct chip attach and an increased involvement with glass substrates. Panasonic is using normal wirebonders with a special shear cycle to attach a gold ball bond to an IC pad and immediately shear the wire at the top of the ball. This leaves a small tail (like a Hershey's Chocolate Kiss), which is then flip- mounted into conductive adhesive sites on a glass substrate. The process requires no tooling or wafer-level processing. Another area that has developed dramatically since my last visit is the film area. Panasonic has put TAB into full use as a glass and PC board interconnection media. It has special equipment for what our hosts refer to as FOG (film on glass) and FOB (film on board). Both technologies are demonstrated as applied to flat panel display production. LCD drivers are mounted to a sliver of PC board material. A TAB interconnection is first hot bar bonded to the glass flat panel using anisotropically conductive adhesive tape. The other end of the TAB is then pulse laser bonded to the driver PC boards. With FOG and FOB, Panasonic has in place the first fully automated LCD assembly line in the world. The current line does a single edge of the LCD per cycle. The new generation will assemble all three edges simultaneously. The Circuit Manufacturing Technology Lab, as a corporate staff function, provides assistance to all system divisions across the Matsushita group on SMT fine-pitch process and equipment development. It is primarily intended to enhance Matsushita Electric's manufacturing capability in order to increase its competitiveness. Membership is 600,000 yen annually, open only to Matsushita's system divisions, its subsidiary companies, and its affiliated companies, not to any party outside of Matsushita, regardless of whether it is a domestic or overseas entity. Almost as an aside, we were given a tour of a factory of SMT mountable inductor coils. This facility was extremely noisy. Wireless headsets must be worn in order to hear the tour guide. Fine wire is automatically wound on a core that is terminated to a continuous leadframe and encapsulated. These are moved around on reels of 40,000 parts each. Parts are marked and visually inspected by computer, then excised from the lead frame and tested and binned. Velocity and quality are very high. Relatively large numbers of people are involved in this production. Process improvements seem targeted at further integration of processes. The current process demands loading the test station from a parts feeder after excising from the continuous leadframe. After testing, parts are once again binned into parts feeders for final packaging. The new process will trim and test the parts and put them directly into rails to be packaged without the parts feeders. This new method reduces the space required for trimming, testing, and packaging by 10 square feet; floor space reduction was mentioned as a key reason for the improvement. As this station is replicated many times, this results in a significant utilization opportunity. One-fourth or more of the floor space appeared to be dedicated to test stations. Mr. Hiroaki Fujimoto, Senior Engineer of the VLSI Technology Research Lab, discussed transfer bump, micro bump, stud bump, and new bonding methods for chip assembly to substrates. Transfer bump is a method developed by Matsushita to displace the need for wafer scale processing to form the gold bumps required for TAB inner lead bonds. Gold bumps are formed on a glass substrate and transferred to the inner lead of the TAB frame via thermo- compression prior to inner lead assembly. Matsushita has marketed transfer bump worldwide for several years now. Micro bump technology is a flip chip method that uses adhesive shrinkage to electrically and mechanically connect the interconnection bumps to a typically glass substrate. Micro bump uses glass and ceramic substrates. The adhesive is photo-curable; thus a more transparent substrate is naturally preferred, but edge-wise illumination will work with opaque substrates. Micro bump is currently in use in thermal print head assembly and is capable of 10 m lead pitch. Mr. Yoshihiro Bessho, Engineer for the Materials and Components Research Laboratory spoke on the stud bump bond (SBB) method of attachment. SBB employs the sheared gold ball bonds mentioned earlier to adhesively mount the chip to a variety of substrates. This technology is limited to pitches greater than 100 m but is extremely flexible and simple in application. The ball bond studs are dipped into conductive adhesive and then glued to substrate. The wire bond "tails" are allowed to set the conductive paste penetration depth and apparently can do so accurately enough that bridging between bonds is not a problem. An undercoat adhesive more firmly secures the overall system. Similar to what we saw and heard at Sony, Matsushita views process development on a par with product development. The Corporate Production Engineering Division (Circuit Manufacturing Technology Lab), reports at the highest level of the company and appears to be marketing what is considered a corporate core competency. As mentioned earlier, the consistent pursuit of a technological strategy is striking. The laboratory's dedication to fine pitch or to transfer bump or other low-cost flip chip technologies is long-term and not likely to be redirected or curtailed short of the set objectives. The Matsushita strategy seems to be a familiar one: make it small and cheap! Site: Matsushita-Kotobuki Electronics Saijo Division 247 Fukutake, Saijo Ehime, Japan Date Visited: October 5, 1993 Report Author: R. Tummala ATTENDEES JTEC: P. Barela W. Boulton G. Meieran G. Harman M. Pecht R. Tummala HOSTS: Norio Meki Director, Research & Development Laboratory Hideo Sakai Assistant General Manager, Sales Department Hiroshi Yamauchi Manager, HIC Manufacturing Section Yasuyuki Baba Chief, Development Dept #5, R&D Laboratory Yasuhisa (John) Kobayashi Coordinator, Semiconductor and Appliance Section BACKGROUND Matsushita-Kotobuki Electronics Industries (MKE), Saijo Division, located in Saijo, Ehime Prefecture, on Shikoku Island, is one of the affiliated companies of the worldwide company Matsushita Electric Industries ($43.75 billion in FY 92), famous for the brand name "Panasonic." MKE, established in 1969, has several divisions in each prefecture on Shikoku and also subsidiaries in the United States, Singapore, Indonesia, and Ireland. Total 1992 sales were 308.9 billion yen (about $3 billion); annual investments were about 14 billion yen ($130 million). MANAGEMENT, ORGANIZATION, AND PRODUCT CYCLES The organization chart, Figure Pan.1, shows corporate functions including planning, personnel, general affairs, accounting, finance, administration, components purchasing and legal/patent section, and divisional arms that are independently managed. Each division is responsible for its own manufacturing, development, design, factory automation, and quality assurance with corporate support from R&D Laboratory, Design Center and Quality Assurance Center. Figure Pan.1. Organizational chart. MKE has about 5,200 employees; the total number of engineers and personnel in manufacturing are 600 and 3,100, respectively. The divisions work specifically to improve current technology for the products to be released within a couple of months while continuing to manufacture existing products. The corporate R&D Laboratory works to develop next-generation products together with the design section of each division. MKE creates new technology every two or three years that results in new products being introduced. At the Saijo Division, 300,000-500,000 VCRs are manufactured per month. This high productivity is achieved by a high degree of factory automation. TECHNOLOGY There are two packaging technologies in use at MKE, one using conventional PWB and the other using a ceramic substrate. MKE is perhaps one of very few companies using ceramic technologies in consumer products and pushing the leading-edge aspects of these technologies in achieving (1) miniaturization, (2) design standardization, (3) low cost, and (4) high reliability. MKE is using QFP with about 100 to 168 I/O at 0.5 mm lead pitch currently and is expected to migrate to 0.4 mm with forced air (N(subscript 2)) convection reflow. The sophistication of MKE at Saijo is in ceramic technology based on hybrid IC (HIC) introduction in the late '70s and low- temperature co-fired ceramic (LTCC) technology used since 1990. MKE's ceramic strategy during the last decade has been as follows: Hybrid IC with Ag/Pd 1981 Hybrid IC with Cu 1988 LTCC with Cu 1990 The MKE plant had shipped about 60 million HICs and 170,000 LTCC modules as of the time of the JTEC visit. It has the capacity to manufacture about 1 million/month HICs and 10K/month LTCCs. Each handling size is roughly 4 inches square. MKE's LTCC technologies with Ag/Pd and copper are schematically illustrated in Figures 4.8 (p. 74) and 4.10 (p.75), respectively, showing the substrate characteristics that include the capacitors, resistors, and conductors. The current design ground rules being practiced are indicated in Figure Pan.2. Electrode pin allocation technology, such as lead frame, lead array, and BGA, is used for connecting LTCC and PWB. Pb(Mg(subscript 1/3)Nb(subscript 2/3))O(subscript 3)-PbTiO(subscript 3)-PbO is the principal ingredient for embedded capacitors. The LTCC is being used currently in peripheral application - tape memory systems - and is expected to be applied in cellular, automotive, camcorder, and computer applications. The LTCC has proven to be more advantageous than equivalent PWB process/materials relative to the application at system level. MKE's LTCC process involves manufacturing the green sheet and producing approximately 0.10 to 0.15 mm diameter via holes on the green sheet by numerically- and mechanically-controlled punching equipment. After filling up and drying copper oxide paste into the via hole, conductor pattern is also applied on the green sheet by the copper oxide paste. These printed green sheets are laminated together under a heat pressure and fired at 550 degrees centigrade in air to remove organic components sufficiently. Then copper oxide is reduced to copper at 350 degrees centifrade in nitrogen atmosphere containing 10% hydrogen, and the copper and ceramic substrate are sintered at 900 degrees centigrade in nitrogen. This unique process provides highly reliable LTCC with copper electrode. Figure Pan.2. Design rule (typical). FUTURE TECHNOLOGY DIRECTIONS Table Pan.1 shows the directions MKE is expected to take. The future directions include (1) upgrading LTCC by buried CR, (2) fine line by photo process, (3) BGA technology, (4) high- precision LTCC by nonshrinking process, and (5) flip chip. Table Pan.1 Future MKE Directions Site: Meisei University 2-590 Nagafuchi, Ome-shi Tokyo 198, Japan Date Visited: October 8, 1993 Report Author: W. R. Boulton ATTENDEES JTEC: P. Barela W.R. Boulton HOSTS: Dr. Otsuka BACKGROUND Dr. Otsuka recently retired from Hitachi to assume a position at Meisei University, where he teaches a course on advanced information technology. He is affiliated with IEEE and chaired the VLSI workshop held in Kyoto December 1994. He teaches computer and computer packaging technology. Through workshops, he teaches engineers from Japanese companies. He has research contracts on electrical characterization of packages. PACKAGING TECHNOLOGY IN JAPAN Dr. Otsuka limited his formal remarks to the topic of "Main Packaging Technology in Japan (Single Chip Packaging)." He provided a number of important insights into the future developments of electronic packaging technology in Japan. 1. Korea or America is expected to take over long-term leadership in LCD packaging technology. Dr. Otsuka made it clear that LSI packaging technologies have a different origin and core competence than LCD packaging technologies. The development of LCD technologies has come from the merging of TV cathode tube and silicon wafer process technologies, which gives Koreans and Americans the potential to take leadership in this area. Korea has made a strong commitment to LCD development, and recently, Samsung and Gold Star have announced massive investments of $400 million and $300 million, respectively, in TFT technology development programs through 1995. Dr. Otsuka was not optimistic about Japanese companies maintaining leadership in CD technology. He didn't feel that large Japanese companies could work with small companies to develop such a new infrastructure. At the time, large companies were worried about small companies competing with them. Dr. Otsuka believed that Japanese companies' departmental structures were too competitive with each other to cooperate on merging new technologies. He felt that presidents cannot control their department heads. That means that they cannot develop large-scale systems like Apple did with Macintosh. Small companies also provide a threat to large companies by their dedication to a single technology. While big companies are investing in CRT technology, they cannot give the same effort to new LCD technology. The large companies are worried about the potential of small companies. Active matrix LCDs by Japan Hosiden are a threat to Sharp because they are selling to Apple and other U.S. companies like Boeing. Companies like Sony are integrating backwards into their own components to increase profitability. That is causing a breakdown in supplier relationships and causing less coordination on new development. In the future, Dr. Otsuka sees the United States maintaining strength in high-frequency devices like computers and microprocessors. He expects Korea will be strong in memories and LCDs in the future. Low-cost micro assembly applications will remain the strength of Japan in the future. If Japan loses low- cost QFP packaging technology, then he expects that Japan will be in trouble. Since plastic molded lead frame packages represent 85% of electronic product applications, Japan's electronic industry must continue to lead in this area. 2. Japan will remain the leader in LSI packaging technologies. Dr. Otsuka said that over 85% of electronic packaging uses plastic molded lead frame technologies. Because these packaging technologies are so pervasive in Japanese industry and because so many companies are committed to them, he argues that Japan will continue to dominate in low-cost packaging technology. LSI packaging technologies include materials, parts, subassemblies, and assembly process. Most companies have specialized in specific areas of plastic packaging technology, which provides Japan with the strongest low-cost packaging infrastructure. This infrastructure is distributed across both large and small companies in Japan and, because it is pervasive, is unlikely to change its technological direction. He pointed out that it would be virtually impossible for any one company or group to change the technological focus of this infrastructure. Dr. Otsuka argued that, as a result, the low-cost electronic packaging industry in Japan will continue to develop current plastic packaging technology, because it is the easiest direction for all industry participants to continue in. 3. Low cost is top priority in the Japanese industry, which requires mass production technologies. Because of Japan's cost requirements, Dr. Otsuka argued that LSI plastic packaging technology has to utilize mass production technology. This mass production orientation will continue to utilize existing packaging technologies. New technologies will require new mass production technologies, which would be expensive to develop and lead to higher costs. To keep from incurring such added costs, Dr. Otsuka argued that companies will continue to push existing packaging technologies to their limits rather than add new high-cost technologies. This strategy means that Japanese companies seek to solve performance problems related to current plastic packaging technologies. 4. Current plastic molded package technologies will meet future miniaturization requirements. Dr. Otsuka feels that QFP packages can be designed with 0.15 mm pin pitch. Such fine pitch will allow 800 pins on a 32 mm package, or 1000 pins on a 40 mm package, or 384 pins on a 16 mm package. To support his argument, Dr. Otsuka gave examples of developments that are currently under production or development in Japan. He provided an example of 344 pins on a 28 mm package using 0.3 mm pin pitch that is currently in production at Hitachi. Dr. Otsuka argued that Japan's strength is due to the extensive infrastructure of companies involved in plastic electronic packaging technology. Japanese companies have developed QFP packages with 500 pins, available from Nippon Printing Company by 1996, using 0.3 mm pin pitch. Some firms are currently developing this process using 0.2 mm pin pitch chips. This will allow plastic QFP packages to surpass ceramic PGA packages in pin density. Such strong development power comes from Japan's massive investment in all areas of plastic packaging technology. Nikkei Electronics described these developments in its August 2, 1993 issue (p. 94). Dr. Otsuka also described Kyushu Matsushita's 0.15 mm pitch soldering process that he expects to be commercialized within several years. This process uses a solder precoat (super solder) with the body being pushed down as infrared heat is applied. The company uses solder bumps to ensure contact, but staggers the bumps to minimize shorting. This process would not require new machine technology for production utilization. Kyushu Matsushita is planning to use this approach in its next-generation camcorder production. While it is currently using TAB technology, Dr. Otsuka believes that the approach will be used with SMT in the future. Hitachi is currently producing circuits with 90 micron wire bonding on chips with spaces between the wire of 120 microns. Oki and Nippon Steel have demonstrated wire bonding with 40 micron pitch. Dr. Otsuka believes that we will see this in production in the future. Fine pitch requires materials with higher resistivity. Fine pitch creates additional soldering problems by causing high current on the printed circuit board. Dr. Otsuka suggested that new materials will be used on fine-pitch boards. For example, conventional FR-4 through-hole resistivity is too low at 130 degrees centigrade to be used on fine-pitch boards. High Tg epoxy provides better resistivity, but for fine-pitch boards, the high resistivity requirements can be met with BT resin. Mitsubishi Gas Chemical has used BT resist (BTM 450) on contacts with 1.27 mm separation. Japan will overcome low yield problems caused by fine pitch packages. Dr. Otsuka agreed that the move to fine pitch production will cause higher reject rates in the beginning. However, he argued that Japan has a long history of taking lower yield technology and improving it. In the United States, he believes that companies change technologies if they have low yields, but not in Japan. In addition, QFP is more inspectable than BGA even in low yield situations. Inspectability is very critical to ensuring quality products. 5. Finer pitch packages will automatically improve operating characteristics. Fine pitch will increase product design flexibility. Since height is very important, single-layer boards are preferred in package design. Also, single boards are simpler to produce and are therefore cheaper to produce. Fine-pitch soldering would mean that we do not need two-layer bonding systems. Two-layer bonding has lower reliability because it depends on the solder connections. Simple packages are the best in every characteristic. For 0.3 mm pin pitch, a 40 mm (superscript 2) QFP package can have over 500 pins. With 0.2 mm pin pitch, 34 mm(superscript 2) packages can have 600 pins. At 0.15 mm pin pitch, we can have 600 pins on a 28 mm(superscript 2) package, comparable to a 1.0 mm pitch bump BGA chip. Dr. Otsuka used the previous examples to argue that these dense packages will be available within this decade. Fine-pitch packages will provide additional advantages. Dr. Otsuka pointed out that finer pitch allows the use of three-set wiring design instead of two-set wiring between ground and power leads. Three-set wiring can increase output from 80-90 ohm to 95- 105 ohm at 200 MHz and above. Finer pitch allows use of three-set packages for high-power packages. At the same time, impedance can be reduced once pin pitch falls below 0.3 mm. Japanese researchers have simulated 0.05 mm pin pitch designs and found that impedance drops 50%. Coupling capacitance can be further reduced in fine-pitch applications by using shorter wires. Shorter wires also means that multilayer wiring in single-chip packages is not essential. As long as power consumption stays below three watts per chip, chip scale packaging with short wires and fine-pitch provide the best design solutions. This has already been utilized in memory DRAM developments and should be viable in LSI packaging. As long as power consumption stays under 3 watts, natural cooling systems can be used and costs kept lower. Dr. Otsuka also pointed out that reduced voltage requirements and reduced load capacitance can save power consumption (P=1/2(superscript *)CV(superscript 2)). Lower voltage reduces coupling noise. He then argued that as we reduce package size, everything improves: "If we can communicate MHz at 3.3 volts, then the same package using 1.5 volts can operate at 400 MHz." He believes that QFP chip scale packages will be able to pass 200 MHz in the future. QFP will be competitive with BGA on price-performance measures. Dr. Otsuka believes that QFP packages are better than BGA packages because they are inspectable and don't require new assembly technology. While the U.S. is interested in BGA because of its high pin count, Dr. Otsuka believes that 0.15 mm pin pitch on QFP packages will make them competitive with BGA alternatives. Since most Japanese companies do not have BGA capabilities, he expects continued QFP design and technology improvements to keep QFP packages cheaper than BGA packages. Figure Meisei.1 shows the competitiveness of different packaging technologies, and additional insights offered by Dr. Otsuka are shown in Figure 4.5 . Dr. Otsuka pointed out that ceramic substrates are cheaper in high- performance applications, but he believes that improvements described in the above statement will make QFP boards cost- competitive in future high-performance applications. The critical improvements are in the soldering process for wire bonding systems and in reducing popcorn problems of chips. If these problems can't be overcome, Japan will not be competitive in the future. But if Japanese firms can meet future miniaturization requirements with QFP technology, they will continue to be the low-cost producers of high-volume electronic packaging. Figure Meisei.1. Competitiveness of different packaging technologies. . Other areas considered important by Murata include module circuit application-specific components and vertical technology integration within the company. As of the JTEC visit, recent capital investments had been made for the production of ceramic materials, monolithic ceramic capacitors, components for imaging equipment, chip components, microwave components, and electronic modules. Aggressive capital investments were also made for thin-film processing technology and the development of equipment for gallium-arsenide semiconductor devices. At the time of the JTEC visit, Murata planned to have a gallium- arsenide integrated circuit pilot line in production by the end of 1994. The motivation for establishing the line, as expressed by our hosts, is to protect intellectual property and to improve Murata's competitive posture in the microwave and RF module business by focusing on future personal communication products. Murata is doing extensive work in sensor development. Present applications for sensor products include engine knock detectors and fire and burglar alarms. An advanced sensor application that was discussed is a piezoelectric gyro for sensing angular accelerations. The current product measures 21.5 x 8 x 8.5 mm and is used in the Sony TR2 and TR3 palm-sized camcorder for vibration stabilization. A larger, more accurate version (24 x 24 x 58 mm) is used in Pioneer's automobile GPS navigation systems. Future markets for Murata sensors include accelerometers and chemical/gas detection. A new silicon micromachining program has been initiated at the Yokohama R&D Center to develop microminiature accelerometers, sensors, and oscillators. The manufacturing lines JTEC panelists toured in Kyoto are most impressive. The total process is automated. The equipment appears to be very robust and capable of handling very high-volume throughput. Over 80% of the equipment being used was designed and built by Murata. Much of the equipment is customized to meet specific product requirements, which justifies the in-house development; however in more general terms, Murata recognizes the importance of internal equipment development in order to protect its competitive advantages. There was evidence of continuous improvement in the manufacturing processes, with particular emphasis on equipment improvements. Murata has evidently mastered the manufacturing process for high- volume, miniaturized discrete components. Perhaps this explains its dominance in the world market. There was nothing magic about what we saw: outstanding equipment designed for automation, meticulous attention to detail, continuous improvement, qualified personnel, and a committed management team. It became very evident during our discussions that Murata places a very high priority on customer interactions. It was pointed out that in some cases Murata people visit customers once or twice a day. Procurement planning is regularly updated, and Murata can expect at least a three-year lead time from customers for new product requirements. It was likewise evident that the needs of suppliers were taken into account when negotiating reduced costs. Suppliers, it was stated, are no good to us if they are driven into bankruptcy. Suppliers are considered part of the team and expected to contribute to the competitiveness of the product. The visit to Murata substantiated much of what has been identified as Japanese strengths: Focus on the fundamentals Pursue continuous improvements Drive continuously to miniaturize components Work closely with suppliers and customers Design and build your own equipment Eliminate waste Make cost reduction a primary objective in design and manufacturing Automate whenever it is cost-justified Build on core competencies Follow an evolutionary product strategy Move increasingly to integrate components to produce functional modules Murata, like other Japanese companies visited, was concerned about the effects of the recessionary period that industry was experiencing at the time of the JTEC visit. The attitude of our hosts, however, was very positive about the future of Murata. If anything, I expect that the present economic climate will strengthen the company and better prepare it for the competitive challenges of the future. Site: Nippondenso Kota Plant Nippondenso Co., Ltd. 1-1 Shawa-cho, Kariya-shi Aichi-ken 448, Japan Date Visited: October 7, 1993 Report Author: N. Naclerio ATTENDEES JTEC: M. Kelly N. Naclerio J. Peeples HOSTS: Mizutani Shuji Dir., Automotive Electronics Div. Shigehiko Ito Gen. Mgr., Electronics Manufacturing. Dept. Mark R. Hunter Public Affairs Department BACKGROUND Nippondenso is a leading manufacturer of automotive components and electronic systems. In 1992, Nippondenso had sales revenues of about $12 billion. Approximately 9% of sales was invested in capital and another 7.3% in R&D. About 56,718 people were employed by Nippondenso in 1992. (The above figures are based on consolidated 1992 data.) Some of the themes that drive Nippondenso's business strategy are high performance, safety, and fuel efficiency. In-house research ranges from semiconductors to micromachines to materials, and from AI to telecommunications to control theory. Fuzzy logic systems have been developed and are in use for anti-lock braking systems (ABS), cruise control, and suspension systems. Nippondenso uses its micromachine technology to manufacture accelerometers and other sensors for automotive applications. As a demonstration of its micromachine prowess, it built a miniature car, barely 5 mm long, complete with electric motor. Nippondenso plans to fuse automobile technology with state-of-the-art electronics to generate new products in many fields. Examples of nonautomotive products we saw included portable cellular phones, hand-held bar code scanners, and factory automation systems. Nippondenso manufactures components and systems for nearly every automobile manufacturer in the world. While Toyota is its largest shareholder and customer, it also sells components to most of the other Japanese auto makers, the Big Three U.S. companies, and most European companies. Nippondenso claims to be a leader in 14 product areas including electronics, fuel injection systems, braking control, navigation, air conditioners, fuel pumps, and radiators. Nippondenso product groups do their own product development and manufacturing but do not have their own sales forces since most of their products are sold to a few automobile companies. THE KOTA PLANT Activities at the Kota plant include manufacture of automotive ICs, electronic control units (ECUs), and communications equipment. This plant is located in Aichi Prefecture on 274,000 m(superscript 2) and was dedicated in April 1987 after one year of construction. The site is about 50% developed, with three main buildings. A new wafer fabrication facility occupies about 18,000 m(superscript 2) of floor space, and a four-floor electronics factory occupies 173,000 m(superscript 2). The total value of the products manufactured in the Kota plant in 1992 was about $160 million per month. There are approximately 5,000 pieces of equipment and 3,700 employees. IC Manufacturing Nippondenso's IC production is being transferred from a 5" line at the head office in Kariya to a new 6" line at the Kota plant. Nippondenso has been in the IC business for 25 years. While its people recognize that they cannot compete with large merchant IC companies in products like general purpose microprocessors or memories, they still feel that in-house IC manufacturing gives them a competitive advantage. Two advantages of building devices in-house cited by Nippondenso are (1) the ability to produce specialized products not available on the market, and (2) the ability to avoid sharing proprietary designs with potential competitors. Nippondenso manufactures complex, more expensive VLSI chips that are specially designed for automotive applications. Engine Control Unit Assembly Line The main four-story building at the Kota site houses chip assembly, hybrid manufacture, SMT board, and final assembly. Five different engine controller product families including over 120 different engine controllers are made on a single assembly line. The line begins with traditional surface mount assembly including adhesive application, component pick-and-place, add form component assembly, and test. After testing, a conformal environmental coating is applied to the boards. The boards are then assembled into metal enclosures that are sealed and marked. The completed modules go through high-temperature burn-in before final outgoing test. After testing, the modules go into a stocker on the factory floor to await a daily shipment to nearby Toyota automobile assembly plants. The entire 1,170-meter manufacturing line, including assembly, burn-in, and test, has only one direct labor worker who folds multiple rigid boards connected by flex cables into a metal enclosure. We were told that this operation had not been automated because the necessary machine would be too expensive. There is zero changeover time between products on the line. In fact, we were told that complex and relatively simple engine controllers were intentionally mixed to balance the throughput of the line. As a practical matter, controllers are usually made in batches of 16 since that is the size of the cassettes used to transport boards on AGVs. In order to simplify material handling, all of the products have a common width, but appeared to vary in length by as much as a factor of three. Every board type is identified by a bar code, and every piece of equipment on the line has its own bar code reader. As a result, the number of different components that must be simultaneously loaded on the pick-and-place machines is probably higher than on most lines producing only one product at a time. Production Systems Almost all of the manufacturing equipment we saw at the Kota plant was produced internally by Nippondenso. Our hosts claimed that because of the company's unique needs, it is actually cheaper to develop equipment in-house rather than purchase more general-purpose tools. One tool for visually inspecting solder joints reportedly took over two years to develop internally. Over 200 people at the Kota site are involved in the development of production equipment and processes. The average piece of equipment on the manufacturing line is either replaced or upgraded every three to four years. Some of the larger pieces of factory equipment are developed at another Nippondenso site dedicated entirely to that purpose, but many of the smaller pieces of equipment are developed and manufactured on-site. In all cases, customization of the equipment, fixture development, and programming is the responsibility of the manufacturing site. Nippondenso is just beginning to market production equipment such as robots to external customers. Factory equipment and production processes are designed by teams that include both hardware and software engineers. The Kota plant was the first to implement a factory-wide CIM system, now in place at all other plants. This system unites data from the head office and all the factories to meet quality standards and delivery times. (In addition to factory-level data, information from design and sales is also included.) Multichip Modules We saw some ceramic MCM modules or hybrid assemblies built at this plant using flip-chip assembly. The flip chip used is similar to that employed by some U.S. automotive manufacturers and involves plated copper bumps with solder. The main Japanese concern with the wider use of MCMs is reportedly cost. Nippondenso says that MCM-L is very close to introduction in some of its automotive applications. Its automotive customers are not too anxious to have risky "new" technologies used in their cars; however, Nippondenso thinks that MCM-L will be the lowest-cost solution for many ECU products. Another driver for the use of MCM technology will be miniaturization of engine control functions so that they can be mounted directly on sensors and/or actuators. Nippondenso representatives feel that by 1998 multiple functions will be combined into modules containing sensors, processors, and actuators. These distributed processors will share data across a vehicle-wide network. Impact of Automation on Productivity The highly automated production line at Kota results in a high degree of flexibility, very high manufacturing quality, and the elimination of most direct labor. During the time the JTEC team spent in the ECU assembly plant, we did not see any modules in reject bins. According to one host, only a couple of modules per month fail final test. In five years Nippondenso expects to have added a second wafer fab and 50% more assembly area while also hoping to reduce by half the total number of people on the site to less than 2,000 from the current 3,700. Customer Relationships Nippondenso has close ties to its automotive customers. Large customers provide regularly updated five-year production requirements. Engine control units are essentially built to order and delivered to customers every day. In order to provide similar service to North American auto makers, Nippondenso operates a similar assembly line in the United States. Engineer Training Nippondenso runs its own two-year college for training engineers. Managers tend to hold four-year degrees from university engineering programs. Practical training in areas such as equipment design takes place almost entirely within the company. During the first five years of employment, engineers each receive about 100 hours per year of formal technical training resulting in the equivalent of a master's degree. In the sixth year, about 10% of the engineers are selected for the management track and receive another 200 hours of technical training. After ten years about 2-3% are selected to become assistant managers and receive additional training. By this point, the assistant managers have earned the equivalent of a Ph.D. within the company. Management and business training is also provided for those technical managers. The fraction in nonengineering fields who become managers is perhaps 10%. Display Technology Nippondenso manufactures passive liquid crystal displays (LCDs) for black-and-white instrument displays. It also manufactures a holographic display produced for the Lexus automobile that projects the vehicle's speed two meters in front of the windshield. Nippondenso's R&D labs have very recently developed a color anti-ferroelectric LCD display technology that offers a wide viewing angle and the potential to be very low-cost. A ten- inch diagonal display suitable for personal computer applications has already been demonstrated. Future Directions Nippondenso believes that future automobiles will contain highly integrated, but distributed, electronic systems composed of sensor/processor/actuator modules that share data over a vehicle network. When asked about the role of fiber optics in those automobiles, our hosts responded that the company's experience putting optical fiber links in the Toyota Sentry door lock system taught them that optical data links are too expensive for automobile applications; Nippondenso was no longer pursuing them at the time of the JTEC visit. Nippondenso is already shipping highly integrated communication/navigation/entertainment systems for luxury automobiles. A demonstration at the plant showed a new system being manufactured for the Lexus automobile. A color flat-panel display built into the automobile's console can display television when the car is parked. While driving, it can be used to display map information or travel guides from CD ROM. The system can also display position information obtained from an automotive navigation system that includes both GPS and dead- reckoning. The dead-reckoning system uses wheel rotation information from the ABS sensors in the front wheels of the car. When the car is put in reverse, a CCD camera shows what is behind the car. If the car gets too close to another vehicle or obstacle, a warning is flashed on the screen. An FM-broadcast traffic information system was scheduled to begin pilot operation in Japan in 1994, allowing the car navigation system to also display traffic congestion information. Aftermarket GPS/navigation/entertainment systems are also being produced by a number of other Japanese electronics companies including Sony, Matsushita, and Pioneer, and they are reportedly very popular with young Japanese. SUMMARY Nippondenso is a best-of-breed automotive electronics manufacturer. It manufactures vehicle electronic systems for sensing, control, communications, and navigation. Perhaps as a result of its close relationship to Toyota, it has embraced a flexible manufacturing approach that enables it to manufacture engine control units with a lot size of one and zero changeover time. This flexibility does not appear to cost the company in- line throughput or manufacturing quality, both of which appear to be world class. Like other large Japanese electronics companies, Nippondenso is vertically integrated, and it designs and manufactures most of its own production equipment. Because of its unique approach to flexible manufacturing, it is possible that in-house development of production equipment such as SMT assembly is actually cheaper than purchasing full-featured machines on the open market. Nippondenso manufactures key components such as semiconductors and micromachined sensors because of the view that doing so gives it a proprietary advantage. The use of MCM technology at Nippondenso will be driven by cost reduction and the desire to integrate sensors, processors, and actuators into compact modules that can be located directly on engine and drive train systems. Most formal training of engineers takes place in-house, with all engineers receiving 100-200 hours of formal technical training per year. The company's skills as a low-cost producer of very high-quality electronic systems is enabling it to branch out into other markets such as personal communications and mobile point- of-sales terminals. Nippondenso believes that total manufacturing efficiency can only be achieved by closely coordinating product development with production engineering. To achieve this, Nippondenso has long been developing specialized manufacturing equipment and methods in-house. Site: Nitto Denko Corporation 61-7, Aza-Sasadani, Yamadera-cho Kusatsu Shiga 525, Japan Date Visited: October 6, 1993 Report Author: G. Harman ATTENDEES JTEC: W. Boulton G. Harman G. Meieran M. Pecht D. Shelton HOSTS: M. Takemoto General Manager Y. Takashima Manager M. Sano Manager, Design Section S. Omori Manager, Quality Assurance H. Tabada Chief Researcher, Reliability Evaluation Center T. Mitarai Assistant Manager, Overseas Department M. Kaneto Researcher BACKGROUND Nitto Denko is a diversified materials company with headquarters in Osaka. Nitto Denko's philosophy is to listen to the customers who are on the leading edge of technology and to develop or improve products for their use. The company was founded in 1918 to produce electrical insulating materials in Japan. Each product center has its own development group and profit center and issues its own financial report. Nitto Denko is an independent company, not part of any keiretsu. About 30% of its business is semiconductor- and electronics- related. Other areas include industrial products, packaging products, engineering plastics, medical products and membrane products. Nitto Denko is a major plastic molding-compound supplier for semiconductor chip encapsulation. Our hosts also indicated that Nitto is the major producer of polarizing films (used in flat panel displays) and low friction rings (used in floppy disks). The company's newest venture is to produce cultured (fermented) ginseng soft drinks to help improve its profit margin. Nitto Denko has manufacturing plants for various products located all over the world, including three in the United States that will be expanding soon as the firm moves more production off shore. Nitto Denko's philosophy emphasizes that it listens to its customers - a lesson that many American companies could learn. It has a standing committee to help solve applications problems of companies using its products. Planning for future growth is exceptional. Company personnel have outlined the entire semiconductor assembly and packaging process and would like to develop products that will contribute to or simplify each step. A diagram showing this approach is given in Figure Nitto 1. A further example of Nitto's exceptional planning is its development and fitting of future products into the U.S. SIA roadmap for semiconductor development by the year 2000. Our hosts at Nitto Denko stated that they have a department of "Research on New Basic Technology," but nothing we saw was identified as coming from that. In fact, when asked how many Ph.D.s they have in this particular plant, their response was "only one." Before a recent reorganization they had four. Figure Nitto.1. Nitto's technical development plan for semiconductor related materials. Nitto's Kameyama plant is highly goal-oriented. It has a product line that primarily consists of adhesives and packaging and encapsulation materials. The company then looks for ways to apply these products and expertise to the field of electronics materials. The JTEC team discerned no fancy gear or advanced research projects; just a strict "improve the product and look for new areas to apply products," no-nonsense approach. The company is quite creative in looking for new applications, but in the electronics areas, it is not trying to expand its product line into ceramic packaging. It is the company's understanding of the characteristics of plastic materials that separates it from its competitors. Nitto appears to be a world leader in this area. Finances Nitto had net sales of $2.187 billion with a net income of $34.3 million in 1993. This compared with a net income of $55.7 million in 1992, and $71.3 million in 1988. The sharp decline in 1993 resulted from the increased value of the yen, coupled with downward cost pressure from its Japanese customers. It is planning to expand production overseas to help overcome the effects of the increasing value of the yen. The biggest problem facing Nitto is the flat sales volume for plastic molding compounds. The volume of molded plastic packages continues to increase, but the thinner package dimensions, e.g., small outline devices, may be only 0.5 mm thick, and improved mold design leading to reduced loss of plastic during the molding operation has led to flat sales. It seems that both of these trends will continue in the foreseeable future. However, it is not obvious what can be done to improve this situation. Nitto has been a leader in developing low-stress molding compounds as well as in low alpha particle filler materials, and it is well positioned to maintain leadership in these various technical areas, as well as in the new products mentioned below. Company officials hope that increased overseas production will overcome the flat sales and profits in the older product areas. Specific Products Nitto is developing a number of new products based on its core technologies. One such product is a new die attach material for application to the whole wafer called "Elep Mount." This is a conductive epoxy film applied to the backside of the wafer before sawing. This application can save several assembly steps and thus time and money. (Note: This is an old idea originating in the '70s at Fairchild. Early in the '80s, such films were unsuccessfully marketed by Stauffer Chemical, a U.S. company, which sold that division to Ablestick. The film is apparently not marketed now. Amicon also tried to market a conductive epoxy screened-on-wafer variation of the system in about 1985. Both had technical problems relating to water pickup in the uncured resin during sawing, even if the wafer is not sawed through, which assembly people prefer to do. A water-repellent uncured epoxy film with new specialized equipment, with IC manufacturing cost pressures, may make it successful for Nitto.) Other semiconductor-related products were discussed. We requested information on a product called ASMAT. This is being developed by Nitto for an ARPA contractor (U.S.) who is planning to use it as area-array contacts for whole-wafer test and burn-in, to obtain known-good-die. The contacts are hemispherical bumps, precisely electroplated (and/or etched) 48 micrometers in diameter on large two-metal layers of polyimide films. Nitto has a tough job ahead to cover all of the obvious problems inherent in placing a contact-covered polymer thin film on hundreds/thousands of bond pads in a wafer, e.g., no metallurgical interaction with the bond pad is permitted at burn-in times and temperatures, and the film contacts must be position-stable at all times and temperatures. Also, the contact resistance varies with applied force, so each one of thousands may require its own, or at least regional separate spring loading. Nitto is considering plating the bumps with nickel, gold, rhodium, osmium, etc. It expected to publish a full paper on ASMAT at the ISHM symposium held in November 1993. Other products in development included improved molding compound adhesion to palladium plated leadframes, high-thermal- conductivity molding compounds for power devices, and others that need lower thermal resistances. Ideally, these materials could be developed to replace ceramic packages. Nitto is also working on recyclable molding compounds (presumably thermoplastics) and low- viscosity compounds intended to go under flip chips (a process developed by IBM). Nitto is also developing mold compound pellets that are dust-free for use in the future (class 100) clean-room assembly and packaging areas. Nitto is also studying trapping agents that can slow the migration of bromine ions (fire retardants) in molding compounds and thus increase gold wire bond reliability. It is developing compounds that are stronger, have greater leadframe adhesion, and have lower moisture absorption, for use in TSOPs (thin small-outlined packages). In addition, Nitto is developing materials that require no post mold cure (currently such cure takes five to six hours at 175¡C). It is using the Cornell University software for modeling its thermosetting plastics. Also being developed are lines of adhesive contacts and adhesive tapes for die attach, TAB assembly, flexible mounting for chip carriers, etc. It seems to have a well thought out strategy for applying its specific technologies to the entire range of assembly and packaging needs of the industry. PLANT TOUR The Nitto Denko plant and much of the equipment was not new, but certainly appeared to be adequate for the job. The failure analysis and other materials analysis lab facilities appeared adequate, but by no means grandiose. There were all the important analytical and environment test facilities needed for evaluation of plastic packaged devices: IR microscope, SEM, radiography, EDX and WDX, TMA and TAG, pressure cookers, thermal cycle and thermal shock, etc. There were no super advanced capabilities such as advanced surface analysis (Auger, STEM, ESCA, etc.), but the company seems to have access to those capabilities if needed. Our main impression of the analytical facilities is that they are extensively used. Work was going on in most of these facilities (which are located in one large room.). Each piece of equipment was clearly labeled in English, apparently for the benefit of the many visitors who come to the facility. A large area is devoted to lead frame plastic molding equipment for running the many experiments necessary to develop new molding compounds. These experiments are designed by finite-element- modeling of the mold compound during molding to minimize wire- sweep and air bubbles. (Nitto may use some of the Cornell software imbedded in its own). There was considerable QA equipment available for testing the molding compounds (gelling time, rheology tests, spiral flow testing, etc.). We saw several examples of QA processes during our tour. The plant was typical of a chemical manufacturer rather than an electronic (clean-room environment) manufacturer. This was especially evident in the areas where the mold compound was formulated and pelletized. However, it should be emphasized that there was sufficient cleanliness for the products, and if more is needed, it will be supplied. The level of automation in the preparation of molding compound is not particularly high, but adequate for the job. Site: Oki Electric Industry Co., Ltd. Honjo Plant 1-1, 4-Chome, Ojimaminiami Honjo-shi Saitama 367, Japan Date Visited: October 7, 1993 Report Author: J. Kukowski ATTENDEES JTEC: P. Barela J. Kukowski L. Salmon R. Tummala HOSTS: Tohru Handa Executive Mgr, New Business Development Div., Telecommunications Group Yoshinobu Tateishi Gen. Mgr, Honjo Plant, Tele. Group Yoichi Kohara Gen. Mgr, LSI Packaging Eng. Dept., LSI Assembly Eng. Div., Electronic Devices Group Saburo Iida Gen. Mgr, Tele. Devices Devt. Div., Tele. Network Gp. Yasuhide Ohnuki Mgr, Functional Devices Devt. Dept., Tele. Devices Devt. Div., Tele. Network Group Jiro Utsunomiya Manager, Functional Devices Devt. Dept. Tele. Devices Devt. Div., Tele. Network Group Mitsuhide Yamada Manager, Functional Devices Devt. Dept. Tele. Devices Devt. Div., Tele. Network Group Yasuo Iguchi Research Manager, Microelectronics Dept. Research Lab. R&D Group Toshiyasu Takei Manager, P.C.B. Eng. Dev. Sect. Takasaki Plant Computer and Network Systems Group Akihisa Yamada Manager, Eng. Relations Dept., Eng. Admin. Div. Hitoshi Shibuya Assist. Mgr, Functional Devices Devt. Dept. Tele. Devices Devt. Div., Tele. Network Group BACKGROUND Oki Electric, Japan's first telecommunications manufacturer, was founded in January 1881. It is developing its business globally, pursuing all aspects of research and development, production, marketing, and service from an international perspective. The company's business sectors are telecommunications systems, information processing systems, and electronic devices. The company produces a wide range of products that include switching systems, transmission systems, telecommunications terminals, data communications systems and peripherals, underwater acoustic systems, telemetry and telecontrol systems, automotive electronics systems, and electronic devices. The company's vision is to capitalize on the technical strength of the three business sectors to be successful in the coming multimedia age. HONJO PLANT The Honjo manufacturing plant produces products for the telecommunications systems business sector. The products being produced at this site include asynchronous transfer mode (ATM) switching systems, modems, facsimiles, telephones, and other peripherals. Technical Presentations Detailed technical presentations were given on the following subjects: -multichip modules -polyimide-copper thin film multilayer substrate -development of post-CFC cleaning technology defluxing -development of single- PPM soldering systems -LSI packaging technology -packaging technology for fiber-optic devices Multichip Modules (MCM). The subject of MCM was presented by Mr. Iida. The topics covered were benefits, Oki development direction, MCM structures, design specifications for LTCC and copper polyimide thin film multilayer substrate, Oki's implementation schedule, and identified applications. Benefits The benefits for creating a MCM are high speed, miniaturization, high reliability, and low cost. The driving forces are low cost and high speed. The key issues in achieving low cost and high speed have been identified as 1. miniaturization, as well as monolysic LSI (large-scale integration) - very fine pattern of substrate - flip chip bonding 2. assurance of known good die (KGD) 3. concurrent design (board/module/LSI) Directions of MCM development at Oki 1993 MCM-C (multichip module-ceramic) - Substrate: Low-temperature co-fired ceramic (LTCC) - LSI: Wirebond / flip chip - I/O: Quad flat pack type (QFP). 1994 MCM-D/C (multichip module-deposited/ceramic) - Substrate: Modified Cu/Pl on LTCC - LSI: Flip chip - I/O: BGA/QFP type (ball grid array) - High power LSI and high reliability 1995 MCM-D/L (multichip module-deposited/laminate) - Substrate: Cu/BT resin on BT base - LSI: Flip chip - I/O: BGA/QFP type - Low power LSI Applications The ATM SW module, optical interface module, digital LSI tester module, clock recovery module, and super multi pins LSI package have been identified for potential applications to incorporate MCM technology. During the technical presentation on MCMs the key features of the LTCC substrate and the copper polyimide substrate were reviewed. Figure Oki.1. MCM structures. Table Oki.1 Design Specifications of MCM Substrates (Low Temperature Co-Fired Multilayer Ceramic Substrate) Table Oki.2 Design Specifications of MCM Substrates (Copper Polyimide Thin Film Multilayer Substrate) Table Oki.3 Implementation Schedule Polyimide-Copper Thin-Film Multilayer Substrate Oki's developments and research in polyimide-copper thin-film multilayer substrate were presented by Mr. Iguchi, who gave us technical details on the following topics: 1. properties of the polyimide resin (thermal, physical and electrical) 2. manufacturing process flow of polyimide-copper thin film multilayer substrate (TFML) 3. comparison of ground configurations (solid, gridiron, biased gridiron, shifted+bias-gridiron) 4. test results for characteristic impedance, propagation delay, transmission loss, cross-talk, and temperature cycling for connection resistance 5. planarization process Oki Development of a Single-PPM Soldering System. Mr. Takei presented Oki's challenges and accomplishments with respect to the goals of CFC elimination and single-PPM soldering. CFC Elimination. Our hosts reviewed the utilization of CFC materials at the Honjo plant and the conventional reasons for CFC cleaning of PCBs. The presentation detailed the development process undertaken to meet the goal of CFC elimination. The process consisted of defining the development target/specifications, defining alternative approaches to eliminate CFCs, the development of experimental fluxes, and testing of the fluxes. Details of the reliability test specification, solderability test specification, contact resistance test specification, and final test results were reviewed. Oki has completed a successful development of a flux with RMA reliability, RA solderability, and ease of test pin contact. The Oki Honjo plant manufactured PCBs for 10 months using the new flux without any major problems reported, and it eliminated the use of CFCs in March 1992. Nitrogen Flow Soldering System. The nitrogen flow soldering system required development of 3 major subsystems: swing spray fluxer for application of low solid flux; low-consumption nitrogen and flux fume chamber for management of nitrogen gas and flux fume flow; and cover plate over the solder bath to eliminate micro solder balls. Technical details and test results were reviewed. Results indicated a failure rate of .02% on a sample size of 21,002 PCBs assembled. The smallest component lead pitch on the PCBs tested was 0.5 mm. Semiaqueous cleaning. The presentation on the development of semi-aqueous cleaning technology was given in two parts. Part one discussed the development of alternative solvents, and part two discussed the development of a cleaning machine. Solvent development. Mr. Takei defined the basic criteria for selecting an alternative solvent: -rosin-based solder paste and fluxes can be used -cleanliness (ionic contamination) is equivalent to CFCs -chemical residue does not damage PCB reliability -no attack upon the parts and PCBs -safety (no poison and no flash point) -low running cost -no damage such as bubbles to cleaning machine A detailed matrix for three alternative semiaqueous solvents was reviewed. The review covered the topics of solvent properties, damage to other materials, bubble height and elimination time, and drying ability. Cleaning Machine Development. There are several basic development criteria for a cleaning machine: -350 mm x 400 mm size PCB to be cleaned -in-line type and max loading of PCB is 3/min -no problem for operational reliability -pure water to be used for rinse -lower ionic contamination than CFCs -immersed water of PCB to be dried up -waste water process to be closed system and low process cost -lower machine cost than other commercial cleaning machines Detailed technical matrices were reviewed, covering cleaning methods, waste water process, influence of liquid quantity taken out by passing PCB concentration, cleanliness of PCBs, machine stability, and reliability and quality of post-CFC cleaning methods. LSI Packing Technology Mr. Kohara covered the following topics on the subject of LSI packaging technology: Logic LSI Packaging - logic LSI package roadmaps - high pin count future direction - wire bond technology improvement - comparison of high-power package cross-sections - thermal resistance - high-speed packages Memory LSI Packaging - technology roadmap for memory-SOJ - technology roadmap for memory-TSOP - heat resistance in reflow soldering - concerns of thin plastic package TSOP - package outlines TAB Package (TCP - Tape Carrier Package) - technology roadmap for TAB Packaging Technology for Fiber-Optic Devices Mr. Ishii presented a technical overview on packaging for fiber- optic devices and reviewed the specifications and block diagram for the optical parallel transmission module. Also included were the topics of coupling efficiency for photo diode and laser diode array modules and bit error rates. FACTORY TOUR The first area we toured was a pilot production area to produce LTCC. The manufacturing process is enclosed in a class-10,000 clean room. This process was in the technology development stage at the time of our visit. The equipment being evaluated for this process consisted of screen printers, inspection stations, NC hole punch, lamination, ovens, laser trim, laser solder for outer lead bond, and thermode bonding for inner and outer lead bonding. The process included screen printing of resistors and capacitors. Table Oki.4 Memory Package (TSOP) Technology Roadmap The second area the JTEC team toured was a production facility for the assembly of printed circuit boards and end products. Our tour was focused in the area of the printed circuit board assembly. The manufacturing facility was organized into islands of automation; their manufacturing operations are controlled by a central host computer. The operation begins with an automated storage and retrieval system. The management information system (MIS) defines whether or not incoming material, received from a vendor, requires inspection before being placed into the storage area. Components are automatically identified and retrieved by the use of computer control in combination with a bar code system. An automated guided vehicle is used to deliver material to the process lines; this includes magazines of printed circuit boards and prekited component feeders for the placement machines. The standard screen, place, and reflow manufacturing process is being used for surface mount assembly. Bar code labels are applied to bare printed circuit boards, and automated in-line inspection is performed immediately after screen printing of solder paste, component placement, and solder reflow. Automated inspection and test of individual components is performed within the placement machines. Solder reflow profiles are downloaded from the host system. Presently Oki is using vision for solder reflow inspection but is developing a 3-D laser inspection station. A warp prevention robot was developed for processing of large printed circuit boards through the reflow oven. Design rules have been established to obtain optimum yields within the manufacturing process. The rules define the placement location of components onto the printed circuit boards. SUMMARY Oki is organized into three business divisions that are highly integrated. The company possesses the technological capabilities required for success in the coming multimedia age. The organization is involved in development and production of products, systems, materials, automated equipment, and manufacturing processes. Oki gives credit to automation as the key in obtaining product yield improvement with high quality and reliability. The driving forces for its development of new products are cost, weight, and miniaturization. During the JTEC visit our hosts were open in technical discussions and extremely hospitable. Site: Sony Corporation 6-7-35, Kitashinagawa Shinagawa-ku Tokyo 141, Japan Date Visited: October 4 and 5, 1993 Report Author: N. Naclerio ATTENDEES JTEC: P. Barela W. Boulton N. Naclerio HOSTS: Yoshiyuki Kaneda Senior Managing Director Yoshiyuki Yamada Member, Board of Directors, Dir. of Research Center Takehisa Okada Senior General Manager, FA and Precision Products Junzo Wachi Deputy Senior General Manager, Marketing, FA and Precision Products Yunosuke Hayakawa General Manager, Production Technology Coordination Div. Production Technology Development Systems Engineering Div., FA Yoshiko Numata Government and External Relations Div., Trade and External Relations Group Koichi Motegi Manager, Government & External Relations Div., Trade & External Relations Group BACKGROUND Sony Corporation has $34.4 billion in worldwide sales, of which $10.5 billion is in the U.S., $8.9 billion is in Japan, $9 billion is in Europe, and $6 billion is in Asia and other areas. Sony has 126,000 employees with 35 manufacturing plants in Japan, 13 in North America, 14 in Asia, and 14 in Europe. The company is organized around business sectors including audio products (audio components, general audio, automotive, and personal communications); video products (personal video, home video); television; personal information systems; and business and professional products. Our host, Mr. Kaneda, was responsible for component technologies (including semiconductors, electronic devices, batteries, and recording media) and factory automation. Mr. Kaneda described the borderless, global marketplace in which Sony competes for new and existing markets. He stated that in addition to building creative products, it is necessary to develop production innovations and equipment for advanced assembly and ultraprecision in order to maximize the company's value added and profits. SONY'S COMPETITIVE STRATEGY Precision Manufacturing And Concurrent Engineering Mr. Hayakawa stressed the importance of precision manufacturing and concurrent engineering in order to produce compact, lightweight, and highly reliable products. As an example, for the TR5 camcorder (introduced in 1989), product planning began with a vision of a video recorder that would fit in the palm of one hand. This fixed the external dimensions of the recorder and drove the size of the circuit board and recording head. Development work began concurrently at the design divisions for issues of high-density component packages, assembly, PCB, IC, tape transport mechanism, recording cylinder and head, lens, and high-density recording tape. Simultaneously, development began in the respective production technology organizations. In the area of printed circuit board technology, the product requirements meant the use of the very latest surface mount technology. Subsequent versions of the TR5 reduced the PCB size by 75%. In order to achieve 100 µm solder lands with 100 micrometers spaces, Sony developed a low melting point solder with more uniformly shaped and sized solder particles and a new metal screen technology for more uniform printing. The TR1 camcorder introduced in 1992 achieves 20 components/cm(superscript 2), or about twice the component density of the TR5. Sony feels that future products will require 30 components/cm(superscript 2). Discrete surface mount components as small as 1.0 mm by 0.5 mm are currently being used. Surface mount technology is also used in Sony's larger products because it is less costly, higher performance, and can be manufactured with higher quality and reliability than through- hole technology. Another example of the need for precision manufacturing in the camcorder is in the recording head. In the manufacture of the very narrow head with 0.3 micrometer head gap, it was necessary to cut ferrite material to a thickness of 2 micrometers and a height of 110 µm. This required the development of a spindle to precisely control a diamond blade. The head is wound 55 times with 30 micrometer wire that must pass through small slots in the head. This required the development of a precision winding machine. The development of these machine tools proceeded concurrently with the design of the product which could not have been built with production equipment available in the marketplace. Therefore, according to Mr. Hayakawa, "manufacturing of the required production equipment must be done in-house. This has become a truly creative activity because it makes the manufacturing of creative products possible." Sony has turned its expertise in factory automation into a new product area. Sony sells most of the equipment it develops to outside customers. We toured the FA demonstration center and surface mount (SMT) assembly training center where customers come to learn how to operate an SMT line. Sony's FA group has grown to about $500 million a year in revenue, with 80% of that still coming from internal Sony customers. Materials And Process Technology Also cited as important were materials and process expertise. As an example, the design of the compact disk player was cited. The heart of a compact disk player is the optical pickup head. In designing the head, Sony considered alternative methods of precision assembly with regards to suitability for automation, reliability, and contribution to size, weight, and cost reduction. Adhesive bonding was the best method for size, weight, and cost reduction, but was poor in the areas of reliability and automation. Because it offered significant product advantages, Sony decided to focus on improving the productivity of the bonding technique. Key parameters to optimize were surface preparation, bonding agent, and process control technique. Sony developed a light ray cleaning method to improve surface wetability and selected nine different bonding agents for joining various components in the pickup head. Sony now manufactures 60% of all the world's optical pickup assemblies. Added Value and the Importance of Manufacturing Key Components Mr. Hayakawa stressed the importance of manufacturing key components of a product from the viewpoint of value added. In the case of a compact disk player manufactured by Sony, only 10% of the value added comes from assembly. In contrast, the key devices (optical pickup, semiconductors, lenses, motors, transformers, and PCBs) account for 55% of the total product cost. Similarly in the case of the 8 mm camcorder, assembly is about 12% of the added value, and key devices (CCD, ICs, drum, PCB, magnetic head, motor, lenses, viewfinder, and sensors) make up about 60%. Sony manufactures about 60% of the key devices for the compact disk player and about 45% of the key devices for the 8 mm camcorder, and it intends to increase both percentages in order to increase profitability. In the case of the Video Walkman, Sony's palm- sized TV/VCR, Sony procures the display from an "associate" company. According to Mr. Kaneda, the miniaturization technology for the electronics and tape transport mechanism was more important than the display technology in realizing the products. The only flat panel displays manufactured by Sony are for use in camcorder viewfinders. Sony feels that LCD technology is only applicable to moderate-sized displays and that its Trinitron CRT technology is superior for wide-screen, high-definition applications. Effectiveness of Automation In addition to processes that require automation in order to achieve precision that cannot be achieved with human hands, Sony cites several other benefits from its use of automation. These include reductions in factory start-up time, in manufacturing defects, and in manpower requirements. In the case of the optical pickup head for the compact disk player, early manufacturing involved manual assembly and adjustment by highly skilled technicians. Worker training limited Sony's ability to ramp production and expand overseas. Automation of the line took a lot of preparation, but was accomplished in four months. Replicated lines in Singapore and France took only two to three weeks to bring up to speed. The operators got one month of training in the Japanese factory prior to startup of the overseas plants. In the case of color television production, robots were introduced to replace 99% of the manual parts insertion. In addition to the reduction in labor costs, the defect ratio was reduced by 90%, the PCB rework rate was reduced by 75%, and the increased product uniformity allowed Sony to greatly simplify final product adjustments. In the case of the Walkman a completely automated assembly and adjustment system was installed. Defect levels during initial manual assembly were 0.2% in the first week and 0.1% in three months. However, during the first week of automated assembly, the defect rate dropped to a steady 20 PPM. The introduction of subsequent Walkman models into the same factory required additional investments of 9.1%, 3.5%, and 1.5%, respectively, for the product changeovers. In addition, the time to bring up the new systems dropped from six weeks to five weeks, then to three weeks, and finally to one week. As a result of Sony's skills in automation, coupled with its focus on design for manufacturability, it was able to increase sales by 121% between 1987 and 1990 with only a 35% increase in the number of direct operating employees. Manufacturing processes are first established and streamlined in terms of manpower, equipment, and materials, then the final system is automated. Because of the heavy use of automation, an increasing fraction of the operators at Sony are involved in maintenance and indirect operations. Sony employs over 1000 people in its factory automation and precision products group and 300 in production technology within Mr. Kaneda's organization. Quality and Reliability According to Mr. Kaneda, any product introduced into the marketplace must have high quality and reliability. Sony tries to reuse well characterized components from previous products in order to achieve high quality in newly introduced products. As much as possible, basic failure mechanisms are identified in the prototyping stages. Once failure mechanisms are understood, stress tests can be designed using heat, environment, or time. Product Lifecycles Product lifecycles are determined by market conditions. On average, minor product changes occur about once a year. However, in the case of the Sony Walkman, new models are introduced every six months or less. Major model change