PHOTOMASK. Mask Industry Assessment: 2007 Gilbert Shelden, Shelden Consulting, Patricia Marmillion, SEMATECH/IBM Corp., and Greg Hughes, SEMATECH

Transcription

1 PHOTOMASK BACUS The international technical group of SPIE dedicated to the advancement of photomask technology. OCTOBER 2007 VOLUME 23, ISSUE 10 Mask Industry Assessment: 2007 Gilbert Shelden, Shelden Consulting, Patricia Marmillion, SEMATECH/IBM Corp., and Greg Hughes, SEMATECH ABSTRACT Microelectronics industry leaders routinely name the cost and cycle time of mask technology and mask supply as top critical issues. A survey was created with support from SEMATECH and administered by SEMI North America to gather information about the mask industry as an objective assessment of its overall condition. The survey is designed with the input of semiconductor company mask technologists, merchant mask suppliers, and industry equipment makers. This year s assessment is the sixth in the current series of annual reports. With ongoing industry support, the report can be used as a baseline to gain perspective on the technical and business status of the mask and microelectronics industries. The report will continue to serve as a valuable reference to identify the strengths and opportunities of the mask industry. The results will be used to guide future investments pertaining to critical path issues. This year s survey is basically the same as the 2005 and 2006 surveys. Questions are grouped into categories: General Business Profile Information, Data Processing, Yields and Yield Loss Mechanisms, Delivery Times, Returns and Services, Operating Cost Factors, and Equipment Utilization. Within each category is a multitude of questions that create a detailed profile of both the business and technical status of the critical mask industry. 1. Introduction In 2002, International SEMATECH reinitiated a survey of the mask industry, intending to enhance the level of understanding of the unique and critical issues associated with the photomask industry. 1 SEMATECH intends to conduct the survey annually to provide a substantial and valuable reference database. This database can be built over time to aid in identifying past trends and validating future projections. A very similar survey had been conducted annually for seven years but was suspended after 1999 due to the lack of a funding source to support it. The results from the past surveys were published as SPIE papers and presented at BACUS. They were found to provide valuable insights for both mask makers and mask users. 2, 3 The current survey has evolved over the last five years. The most significant changes were in 2005 when the survey was expanded, adding questions on operating cost factors and equipment utilization. Questions are grouped into eight categories: general business profile information; data processing; yields and yield loss mechanisms; delivery time; returns, service; operating cost factors; and equipment utilization. Within each category are multiple questions that create a detailed profile of both the business and technical status of the mask industry. 2. Assessment Approach For 2007, the survey was sent to thirteen photomask companies. Seven responses to the survey were received, which represents an acceptable cross-section of the industry. The survey was divided into seven major sections: 1. General Mask Profile Information This section seeks to establish a general description of the basic elements of the mask business: technology groundrule categories; application types (logic, memory, microprocessors, etc.); mask types by glass size (PSM, binary, OPC); magnification distribution; and distribution of fabrication methods used for patterning, etch, metrology, inspection and repair. The type of pellicle is also included. 2. Data Processing This section requests data on file size and data preparation time average, maximum, and 95 th percentile of each respondent s distribution. For the first time, write time data were gathered. The 95 th percentile is requested to help discern what file sizes and write times are truly representative of the norm without TAKE A LOOK INSIDE: INDUSTRY BRIEFS For new developments in technology see page 13 CALENDAR For a list of meetings see page 14 Continues on page 3.

2 Page 2 Volume 23, Issue 10 Editorial Maskmakers under Hypnosis Wolfgang Staud, B 2 W Consulting 2 days after yet another very successful conference. As a first order our Thank You should go out to Bob Naber and Hiroichi Kawahira for putting together an excellent program: record submissions, a simple but very effective change to the length of each presentation, a very successful poster session, and probably the best attendance of any Friday Special Session on record. The local weather station had sent out warnings over the preceding weekend of a storm moving in, but the sun in Monterey could not have been any brighter and warmer. We enjoyed another scenic panoramic backdrop to our conference. Sessions in general were packed, with the 2 nd [Ferrante] room being at max capacity most of the time. Rick Wallace s keynote speech was very much like Jim Morgan s from a few years back: 30,000 ft helicopter view with little any of us could take home as a direct deep insight. The Steering Committee still needs to do a better job at coaching its speakers. And when looking at the Friday session, then Wallace s challenge - that the 7% that do think the next node is easier than the previous did not get the message was clearly contradicted. We saw clear advances on all fronts, nothing revolutionary, but good, solid, evolutionary progress. Writer specs, inspection, haze control, repair, etch, RET, DFM, Metrology and MDP all same incremental improvements. If we needed any indication where the industry is headed, then we got a clear message from the Friday Session on Double Patterning. The three opening talks by IBM, Intel and AMD showed us, what restricted design rules, rigorous cell layout optimization, and deep pockets for 2-mask-per-layer schemes will get you: a roadmap that pretty much has 32nm cleared, and is looking at 22 down the road all implemented with 193nm light. EUV for all three of these companies seems to have been pushed out [off?], and all three talk about the DFM efforts as all but completed. Bob Bigwood described DFM as not just hot spot fixes for manufacturing, and in his graphs had DP as more of an incremental rather than disruptive technology. Decomposition flows will see the largest impact in the synthesis part. And MDP takes the biggest hit in throughput. Luigi Capodieci went on to show the need for negative resists to become available, and potentially some pattern freeze or image stabilization technique. Several talks also mentioned the advantage of re-use of less expensive and existing equipment. Cymer s outlook on their laser roadmap fell fully inline with the empirical extension and improvements with higher repetition rates. The one DP technology that still has a lot of questions to answer is Spacer: can the layout tools handle the very specific challenges of self-aligned spacers and their very tricky trim masks. [Neither Cadence nor Mentor provided any answer here]. But enough of the technology challenges. We are on an optical forever roadmap, and no matter how hard MvdB last year wanted to make us believe that EUV is a COO equivalent solution we are heading down the DPL path. It almost seems like the entire industry is under this hypnotic spell right now, and even multiple exposures are being discussed. The entertainment provided an almost classical metaphor for the current state of the industry and I am waiting for Ricky Kalmon to go wake up. The true highlights of the symposium came from three very special individuals: Hiroichi Kawahira culminated his outlook into next year s program with a very inspirational one-man karaoke session of Sound-of-music. Hopefully this was a glimpse that next years entertainment as well. Andy Neureuther received the a lifetime achievement award for his program at the University of California, Berkeley and the rewards his students have garnered in best papers/posters and scholarship speak for themselves. But the true highlights of the conference were the two brief appearances by John Bossung. What an inspiration! After a short 15 minute talk on Thursday morning, where he was looking back all the way into 1957, he received an extraordinary ovation during the evening s banquet. It became clear that neither he nor his wife had any idea of what an iconic statue he had become. Mark Mason sitting at the head table, IM-ed his friends in Dallas that he was sitting next to John Bossung and the reply was: What are you guys plotting?. To top even all of these honors, John Bossung approached us early Friday, and asked to be allowed to re-donate his award for a special scholarship next year! In a class of his own. N E W S BACUS News is published monthly by SPIE for BACUS, the international technical group of SPIE dedicated to the advancement of photomask technology. Circulation Managing Editor/Graphics Linda DeLano Advertising Sue Siegfried BACUS Technical Group Manager Pat Wight 2007 BACUS Steering Committee President Brian J. Grenon, Grenon Consulting Secretary John Whittey, Vistec Semiconductor Systems, Inc. Quarterly Meeting Chair Robert (Bob) Naber, Cadence Design Systems, Inc Annual Photomask Chairs Hiroichi Kawahira, Sony Atsugi Technology Ctr. (Japan) Larry S. Zurbrick, Agilent Technologies, Inc. International Chair Wilhelm Maurer, Infineon Technologies AG (Germany) Education Chair Wolfgang Staud, B 2 W Consulting Newsletter Editors Artur Balasinski, Cypress Semiconductor Corp. Warren Montgomery, Albany Nanotech Sponsorships Susan Siegfried, SPIE Sponsorship Consultant Members at Large Frank E. Abboud, Intel Corp. Uwe Behringer, UBC Microelectronics (Germany) Ute Buttgereit, Carl Zeiss SMS GmbH (Germany) Chris Constantine, Oerlikon USA Inc. Benjamin G. Eynon, Jr., Sematech and SAMSUNG Gregory K. Hearn, SCIOPT Enterprises Kurt Kimmel, IBM Microelectronics Div. Paul Leuhrmann, ASML Mark Mason, Texas Instrument Inc. John A. Nykaza, Toppan Photomask, Inc. J. Tracy Weed, Synopsys, Inc. Larry S. Zurbrick, KLA-Tencor Corp Society of Photo-Optical Instrumentation Engineers. All rights reserved. P.O. Box 10, Bellingham, WA USA Tel: Fax: SPIE.org customerservice@spie.org

3 Volume 23, Issue 10 Page 3 Continued from cover. having extraordinarily large, anomalous files over weighted in the analysis. 3. Yields and Yield Loss Mechanisms Overall yield by glass size and mask type (binary, phase-shifting method) is given. Fifteen yield loss mechanisms are offered to create an industry Pareto chart separating binary and phase shift mask losses. 4. Delivery Time Data on average delivery time and time for the first three layers of a new mask set are collected by four different mask type categories: binary, binary with moderate to aggressive OPC, EPSM, and APSM. 5. Mask Returns This section collects data on mask returns according to nine categories. 6. Mask Service Mask maintenance services divided into seven service categories are described. 7. Operating Cost Factors Cleanroom space, number of exempt and non-exempt employees, and the allocation of engineering resources are included in this section. 8. Equipment Utilization Scheduled downtime, unscheduled downtime, and engineering activity by tool type are collected. Data include the number of tools to allow for normalization among respondents. The complete sets of questions 1-6 are shown in Appendix A. David Powell Consulting, the same company that had been contracted for the surveys over the last five years, was again the focal point for data collection. The company removed tarticipant identification markings to make the raw data anonymous and then forwarded the information to the authors of this paper. Data were requested from survey participants for the 12-month period from July 1, 2006, through June 30, The data were loaded into a spreadsheet and summarized for lowest, highest, and average values. Respondents received an Excel file with complete summary data (beyond the scope of this paper) for further analysis as an incentive tp participate. SEMATECH s intention is to continue sponsoring the survey as an annual event. Based on continuing experience, the survey is expected to evolve in scope and clarity. The authors invite critique and suggestions from the industry, including mask users, to maximize the utility and value of the survey. These comments will be forwarded to the author of the next survey summary. 3. Data Appendix A shows the survey questions 1-6 and the average, highest, and lowest response for most questions; it should provide readers with an opportunity to data mine the survey at their leisure. Appendix B follows with charts of the key observations that can be drawn from the data and some year-to-year trends. Although the survey had seven participants, not every participant answered every question. Additionally, responses were not received from every mask maker; therefore, this survey should not be considered complete or totally comprehensive. However, the responses from the majority of the major mask manufacturers allow the survey results to be interpreted as representative of the industry as a whole. Comparisons with the 2002, 2003, 2004, 2005, and 2006 surveys may not be totally accurate since each survey represented a different cross-section of the overall industry. Additionally, where possible and appropriate, the results from the 2003, 2004, 2005, 2006, and 2007 surveys have been volumeweighted. The following defines the average mask maker who participated in the survey, using this survey s operating cost factors and equipment utilization data: square feet of cleanroom square feet is Class square feet is Class square feet is Class employees 175 manufacturing non-exempt 45 manufacturing exempt 55 engineers 12.0 process control 4.0 yield improvement 12.9 equipment maintenance 13.9 R&D 4.0 data processing 6.1 sustaining engineering 0.4 NGL 1.8 other 16.6 mask writers 9.7 CD measurement 25.4 defect inspection tools 7.0 defect repair tools masks are shipped per year 1340 engineering plates written 1183 engineering plates measured 312 engineering plates inspected 32 engineering plates repaired Some summary observations drawn from the data are as follows: Masks fabricated for sub-130 nm node technologies have increased from 15.6% in 2006 to 16.9%. 59.8% of all masks are still built to a 250 nm or larger design node. This is slightly increased from 57.2% in The sub-90 nm node technologies are at 10.0% compared to 8.3% in 2006, and the sub-65 nm node technologies are at 2.8% compared to 1.6% in It is possible that these differences would have been greater if all companies participating in the past surveys had participated this year. Logic designs are the largest sector of mask volume but are decreasing. They currently account for 43.0% of the masks made compared to 52.6%, in From a mask volume standpoint, memory and microprocessor volumes are low at 10.0% and 3.3%, respectively. 70% percent of the photomasks are made on 6025 format (6-inch square) substrates. Five-inch substrates are used for 24% of masks, and all other glass sizes represent 6% of the total. Of the 6-inch masks, 7.3% are attenuated phase shift masks (PSMs) and 0.8% are alternating PSMs. The percent of 5X magnification mask continues to exceed 4X 53.5% versus 32.7%, and 1X is 10.0% of the total volume. All other sizes account for the remaining 3.9%. Design data file sizes are averaging 3.7GB, up from 2.7 GB in The maximum data file size was 439 GB also up from 333GB in was the first survey that asked about mask write times. The average write time is 2.8 hours. The 95 th percentile write time is more than double that at 5.6 hours; the highest reported write time is at 35.2 hours (1.5 days). Continues on page 4.

4 Page 4 Volume 23, Issue 10 Continued from page 3. On a volume weighted basis, GDSII accounted for 66.5%, Mebes 30.8%, OASIS 0.8%, and Other 1.8% of data formats used. 21.8% are received in SEMI P10 format. Data preparation CPU processing times averaged 6.0 hours compared to 9.1 hours in The maximum observed was 160 hours compared to 247 hours in 2006 and 150 hours in The average yield for binary masks is 95.6%, which is the highest reported yield to date, 2006 was 94.8%. Sixinch binary masks yielded at 95.3%, up from last year s 94.6% Six-inch PSM yields are 68.5% for attenuated and 69.9% for alternating. In 2006, the yields were 74.1% and 59.0%, respectively. The major process-related yield loss mechanism continues to be defects, which account cumulatively (hard and soft, all types) for 57.3% of the yield loss (binary masks 54.1% and PSM 65.7%). Administrative and manufacturing errors combined account for 20.6% of the mask yield losses, while CD control accounts for 15.3% of the yield losses. Mask manufacturing continues to improve the critical specification of delivery time. Delivery of the first three binary masks is averaging 2.3 days with the overall average increasing to 4.0 days. (In comparison, they were 2.6 and 4.8 days in 2006.) Delivery time for attenuated PSM averaged 2.7 days for the first three layers and increased to 5.1 days for all layers. (In comparison, they were 5.5 and 8.9 days in 2006.) Delivery time for alternating PSMs was 10.5 days on a very small sample. Mask return volumes have too small a sample to report. The largest return reasons are other at 22.3%, soft defects at 20.9%, and data prep at 20.3% of the returns. Hard defects bad repair were 13.0% of the returns with an additional 9.4% reported as hard defects. Administrative errors accounted for 9.4% of returns. Maintenance and service was dominated by damaged pellicles 31.0%, other at 27.6%, and non-removable particles at 17.8% 4. Conclusions The mask industry assessment survey continues to provide a reasonable picture of the state of the mask industry despite the decrease in the number of participants (i.e., seven). The data, although not a complete portrayal of the industry, are still valid for extracting information. Some interesting observations are summarized in Section 3 and graphically presented in Appendix B. These observations will only be enhanced in value as the history and trends are created over time. Mask data volumes continue to grow at a rate greater than 2.6 times per 3-year node. This is beyond the International Technology Roadmap for Semiconductors (ITRS) prediction of 2 times per node. Since the shaped beam mask writer time is related to the data volume written, write times could be expected to grow at a similar rate and thus might exceed the 15-hour write time for the high end masks 95 th percentile within 3 years without significant learning on new mask writers. PSMs continue to have significantly lower yields than binary masks, perhaps because the most difficult bleedingedge mask layers are the PSM layers and mask makers have always encountered yield issues with the newest, most difficult masks. Hard defects continue to be the largest source of yield loss and, therefore, a major influence on cost and delivery time. Hard defects account for 57.3% of yield loss, making defect-free manufacturing and an improved ability to repair defects the most opportune areas to improve mask yield. Administrative and manufacturing errors are the second largest yield detractor at 20.6%, followed by CD at 15.3%. 5. Acknowledgments The authors heartily thank the participating companies-compugraphics International Ltd., Dai Nippon Printing, Hoya, Intel Mask Operation, Taiwan Mask Shop, Taiwan Semiconductor Manufacturing Company, Ltd., and Toppan Printing-for their spirit of cooperation in providing this valuable data to the industry. The authors thank Lara Chamness and Dan Tracy at SEMI North America for co-sponsoring the survey as in the past and administering the actual mailing and collection of data. The authors also thank Dede Adams at David Powell Consulting, Inc. for her role in data collection. And most importantly we would like to thank SEMATECH and its member companies for their funding and the technical assistance and guidance provided in the generation of this paper. 6. References 1. K. R. Kimmel, A Mask Industry Assessment: 2002, 22 nd Annual Symposium on Photomask Technology and Management BACUS, Proceedings of SPIE Vol Part 1, pp. 1-14, B. J. Grenon, 1999 Mask Industry Quality Assessment, 15 th Annual Symposium on Photomask Technology and Management BACUS, Proceedings of SPIE Vol. 3873, pp , E. Gonzalez-La O, 1998 Mask Industry Quality Assessment, Proc. SPIE 14 th Annual Symposium on Photomask Technology and Management BACUS, Proceedings of SPIE Vol. 3546, pp , B. J. Grenon, Mask Makers Data Book, 2002 Edition, published by Grenon Consulting, Inc., K. R. Kimmel, A Mask Industry Assessment: 2003, 23 rd Annual Symposium on Photomask Technology and Management BACUS, Proceedings of SPIE Vol. 5256, pp , G. Shelden and S. Hector, Mask industry Assessment, 2004, Photomask Technology Conference, Proceedings of SPIE Vol. 5567, pp. 1-12, G. Shelden and S. Hector, Mask industry Assessment, 2005, Photomask Technology Conference, Proceedings of SPIE Vol. 5992, G. Shelden and P. Marmillion, Mask industry Assessment, 2006, Photomask Technology Conference, Proceedings of SPIE , 2006

5 Volume 23, Issue 10 Page 5 Continues on page 6.

6 Page 6 Volume 23, Issue 10 Continued from page 5.

7 Volume 23, Issue 10 Page 7 Continues on page 8.

8 Page 8 Volume 23, Issue 10 Continued from page 7.

9 Volume 23, Issue 10 Page 9 Continues on page 10.

10 Page 10 Volume 23, Issue 10 Continued from page 9.

11 Volume 23, Issue 10 Page 11 Continues on page 12.

12 Page 12 Volume 23, Issue 10 Continued from page 11.

13 Volume 23, Issue 10 Page 13 Industry Briefs Sponsorship Opportunities Sign up now for the best Photomask 2007 sponsorship opportunities. Contact: Sue Siegfried Tel: Advertise in the BACUS News! The BACUS Newsletter is the premier publication serving the photomask industry. For information on how to advertise, contact: Sue Siegfried Tel: BACUS Technical Meetings BACUS holds technical meetings in the Bay Area approximately every quarter, from 8:30 to 11:30 am. If you are interested in presenting a paper at this meeting, contact Robert (Bob) Naber, Cadence Design Systems, Inc., Tel: ; naber@cadence.com BACUS Corporate Members Aprio Technologies, Inc. ASML US, Inc. Brion Technologies, Inc. Coherent, Inc. Corning Inc. Gudeng Precision Industrial Co., Ltd. Hamatech USA Inc. Inko Industrial Corp. JEOL USA Inc. KLA-Tencor Corp. Lasertec USA Inc. Micronic Laser Systems AB RSoft Design Group, Inc. Synopsys, Inc. Toppan Photomasks, Inc. Inverse Lithography Makes Market Progress Aaron Hand, Executive Editor, Semiconductor International, interviewed Luminescent CEO Moris Kori. Luminescent announced the sale of a Luminizer inverse lithography system to a leading semiconductor manufacturer in the US. It is one of their two products: the Lithography Explorer (LE). The other one is the Luminizer Mask Synthesis (MS), an OPC and assist feature replacement production-level system. The LE system allows optimizing illumination, mask, scanner, and design simultaneously. However, both systems are based on the same core IP, i.e., an inverse solver which is an algorithm developed to solve inverse problems for imaging. The semiconductor manufacturer plans to use this tool for 32 nm development. The MS tool is being qualified for 45 nm logic and 50 nm DRAM technologies, by two other customers. Luminescent claims Inverse Lithography Technology is an OPC replacement, which integrates the assist features with the main pattern correction, rather than developing OPC fetaures separately, as an add-on to that pattern. They also claim there is a need to replace the current OPC and rule-based assist feature scheme, due to the improved correction capabilities and optimized mask data process flow. The company objective is to grow into a well-established supplier for the OPC industry, which, they believe, is going to be an even bigger one than what it is today. Double Patterning Makes Headways, too Toppan Photomask Inc. (Round Rock, Texas) and CEA-Leti (Grenoble, France) have signed a joint development agreement to explore double patterning (DPT) for extending 193 nm lithography to 32 nm technology. DPT takes k 1 factors back to workable levels and bridges the gap between immersion and extreme ultraviolet (EUV), not expected to be available for volume production until at least But double patterning comes with its own set of challenges, including stringent overlay requirements, CD control and the decomposition of complex patterns. According to Craig West, director of applications for Toppan, much like immersion, some leading semiconductor manufacturers may adopt DP sooner if the pace of progress continues. The current inflection point is at 36 nm HP where single exposure, even with NA of 1.35, is not capable to ramp for highvolume manufacturing (HVM). On the other hand, a two-year gestation period is needed to align integration needs across device design, EDA and RET tools, maskmaker tools, and wafer litho toolmakers and processes. but EUV is still alive There seem to be a quickly growing number of industry experts who speak out against EUV viability. Indeed, if EUV lithography doesn t see real action until the 22 nm mark, there s a distinct possibility that it will become a onenode technology something this industry has always shied away from. Not surprisingly, nanoimprint folks do not speak highly of EUV. However, EUV technology certainly seems to be making some headway lately. Cymer has made great advances in source power and debris mitigation, reaching 50 W with its LPP source and collector lifetimes of about a year. Europe recently completed its More Moore program, listing several achievements it made in EUV developments. The interest remains high and uncountable millions of dollars continue to be poured into the technology, while other solutions may lie outside the industry s reach because there are simply no dollars left to be thrown their way. To receive announcements for these meetings, send an message to patw@ spie.org; in the body of the message include the words subscribe info-bacus.

14 Join the premier professional organization Page 14 Volume 23, Issue 10 for mask makers and mask users! About the BACUS Group Founded in 1980 by a group of chrome blank users wanting a single voice to interact with suppliers, BACUS has grown to become the largest and most widely known forum for the exchange of technical information of interest to photomask and reticle makers. BACUS joined SPIE in January of 1991 to expand the exchange of information with mask makers around the world. The group sponsors an informative monthly meeting and newsletter, BACUS News. The BACUS annual Photomask Technology Symposium covers photomask technology, photomask processes, lithography, materials and resists, phase shift masks, inspection and repair, metrology, and quality and manufacturing management. Individual Membership Benefits include: Subscription to BACUS News (monthly) Subscription to Microlithography World (quarterly) Quarterly technical meetings in the Bay Area Reduced registration rates at BACUS Photomask Technology annual meeting Eligibility to hold office on BACUS Steering Committee spie.org/bacushome Corporate Membership Benefits include: One Voting Member in the SPIE General Membership Subscription to BACUS News (monthly) One online SPIE Journal Subscription Exhibit Space discount of 8% at either the Photomask or Advanced Lithography Symposium Listed as a Corporate Member in the BACUS Monthly Newsletter spie.org/bacushome C a l e n d a r 2008 The 24th European European Mask and Lithography Conference (EMLC 2008) January Hilton Hotel Dresden, Germany SPIE Advanced Lithography February San Jose McEnery Convention Center San Jose, California USA spie.org/al Photomask Japan April Hotel Pacifico Yokohama Yokohama, Japan Photomask Technology 6-10 October Monterey, California USA spie.org/photomask SPIE is an international society advancing an interdiciplinary approach to the science and application of light. International Headquarters P.O. Box 10, Bellingham, WA USA Tel: or Fax: customerservice@spie.org SPIE.org Shipping Address th St., Bellingham, WA USA 2 Alexandra Gate, Ffordd Pengam, Cardiff, CF24 2SA, UK Tel: Fax: spieeurope@spieeurope.org You are invited to submit events of interest for this calendar. Please send to lindad@spie.org; alternatively, or fax to SPIE.