【MES】要450mm ，先給糖吃，其餘免談

2008.10.23

看到這篇文章有「他鄉遇故知」的感受。猶記得參加2007 Taiwan Semicon時，Applied Materials以及其他主要設備廠商都信誓旦旦的說，450mm的R&D經費太龐大，而建造一座450mm的晶圓廠需要天文的投資成本。在投資報酬率的考慮之下，並不可行，而極力說服以提高現階段300mm的設備的效能與生產力為後續研發的主軸。如今在此科技劇烈競爭激盪的時代，無論是設備廠商、IC設計廠商、或是晶圓的製造商無不焦灼等待何時「必需」加入，以期獲得企業永續經營的機會。歷史殷鑑不遠：8 吋晶圓廠興起，6吋廠相繼應聲倒地；之後，12吋廠旗鼓大張，許多8吋廠不得不面臨關閉的行列。因此，450mm戰局的投入只是時間點的引爆。

誠如作者預言：450mm廠的林立，將迫使晶圓廠「大者恆大」的局面，而將來只會有更多的Fabless IC設計公司出現。在現今消費者講究產品「輕、薄、短小」的風潮之下，產品的設計也必須滿足客戶此種需求，而450mm晶圓廠邁入16nm的技術所製造的產品相信會帶來更多不同的應用領域，將過去認為不可能的變成可能的生活必需品。然而，在目前經濟如此不景氣的情況之下，有誰能主導這場遊戲規則呢？這也許是半導體的龍頭大老莫屬，例如：Intel、Samsung 與TSMC。如果這三家公司在設備的支出佔所有設備廠的60%-70%，那被掐著命脈的設備廠諸如Applied Materials、ASML也不得不從。在RD經費緊縮的現在，設備廠商說「不」，也或許期望龍頭晶圓廠能挹入更多研發經費，加速450mm計畫的實現。目前已經450mm 的Prototype晶圓廠在位於Austin的ISMI與在位於日本的Murato Corp進行測試當中，也是個很好的証明。

450mm的來臨並不需要悲觀，相反地，它可帶來龐大的商機，但也同時讓半導體行業的競爭更加劇烈。除了機台設備必須到位之外，舉凡IT的Infrastructure與應用系統等，都必須有個新的思維，才能以更快的速度，提供即時正確的資訊。那時目前獨立的、號稱處理速度快的獨立伺服器會消逝嗎﹖那些無法與目前資訊科技相比擬的CIM技術會被新的取代嗎？還有資料庫如何應付更龐大的不同結構的資訊，以及時提供企業決策使用？還有整個管理模式如何去因應此種變化，而3rd Party廠商又如何提供專業化的諮詢，來滿足客戶的需求？這些都是很好的思考方向。

深入閱讀

 When Is No Really a No? 


Author : David Lammers
October 23, 2008

An executive at a major IC manufacturer likes to tell the story about a meeting in 1996 to discuss 300 mm wafers. One after another, CEOs of the major equipment companies said that they wanted nothing to do with 300 mm equipment, that they had no plans for 300 mm R&D. “I’m out of it,” they said, one after another. The next morning, the executive took a series of phone calls from the same set of CEOs, who told him, “What I said yesterday was just our public position. I want you to know that we want to work with you on 300 mm equipment.”

The executive believes the same kind of “melodrama” is going on now. All of the major equipment companies have 450 mm R&D programs going on, he said. The companies making automation equipment, wafer carriers, and the like, are either participating in the Interoperability Test Bed, now underway by the International Sematech Manufacturing Initiative (ISMI) in Austin, or are working at a similar automation test bed at a Murato Corp. facility in Japan.

At the ISMI Symposium on Manufacturing Effectiveness in Austin, others took a much different view of things. One source said he had canvassed his friends in the equipment industry who said they believed the willingness to do 450 mm tool development is declining. “It’s gotten less likely than at West. The resistance is getting higher.” The economy is much worse, and a quick scan of the headlines shows major equipment companies laying people off and struggling to make any gross profit.

Another source said he believes Applied Materials has a 450 mm development program going, initially aimed at a PVD tool. Another said a friend at a company he deals with has seen an ASML exposure tool rigged up with a prototype 450 mm stage. And an Intel materials scientist said the four major wafer manufacturers all are starting to make 450 mm single crystal wafers, though they don’t want to talk about it publicly.

At Semicon West in July, there were plenty of executives who said they were dead set against 450 mm, that it doesn’t make economic sense, now or ever, that the industry should focus on reducing cycle times for 300 mm tools, and so on. Executives at Applied Materials, ASML, Novellus, all have been adamant in their opposition to 450 mm. SEMI has reflected their members’ views in an aggressive series of white papers and public meetings, arguing for focused attention on 300 mm productivity improvements.

Is this series of “No’s” a means of wresting R&D dollars out of the major device makers? For a couple of years now, discussions have been going on about creation of an R&D fund that would take money from the IC manufacturers that back 450 mm and divvy it out to the equipment makers. While that would seem to make perfect sense in a time of squeezed R&D budgets, nothing has been decided, ISMI managers said at the symposium.

So who to believe? Are the major IC manufacturers which back the 450 mm transition -- which include Intel, Samsung, and TSMC – big enough to get what they want? If a handful of companies control 60-70% of the equipment spending, at some point are they going to get what they want?

It takes some listening skills to figure out which way the wind is blowing in 450 mm development. It is like a kid who asks his mother for a new bike. She says ‘No, but if you turn off the TV and get out there and rake the leaves, Santa Claus might bring you something good.’

Meanwhile, the technical investigation continues. A source at a lithography maker said one question is whether the 450 mm wafers can be made flat enough. By the time 450 mm factories are built, leading edge device makers will be using 16 nm generation technology, he said. That could put the onus on wafer flatness like never before. Exposing ultra-thin photoresists on the much larger wafers could prove to be an engineering challenge. In response, the Intel materials scientist said that if EUV lithography is in use then, the improved depth of field provided by EUV could help in that regard. “There are so many challenges, it is hard to say that flatness is the biggest. But we will figure out how to solve these problems,” he said.

That seems to be one school of thought. That this time around is no different than the transitions to 200 mm or 300 mm wafers. The same technical and business challenges, the same posturing, with the end result being adoption of the larger wafer size.

That may be how things will go this time around. The Big Three will get even bigger, the smaller chip companies will go fabless, and the push to do 450 mm will be aided by a pool of R&D dollars for the equipment industry. One by one, challenges will be overcome, both technical and economic.

Somehow, however, the No’s coming from the equipment industry this time around seem different than a dozen years ago. It feels different. If equipment CEOs are saying No in public and Yes in private to big customers like Intel and Samsung, that would qualify some of them for Acadamy Awards in acting. More likely, they are sitting down with long faces and saying, ‘Give us enough money to get the ball rolling.’ In today’s business climate, that seems a reasonable survival strategy. But the leaves sure are piling high.

From: http://email.semiconductor.net/cgi-bin2/DM/y/hBAgI0N3Sdn0Xjk0Dkqp0EL&rid=828757675


Response:

October 23, 2008
In response to: When Is No Really a No?
Stephen Frenkel commented:

David, You address an important point. It's important for players to realize the confusion and misunderstandings that arise when they contradict themselves. Developers are left wondering whether they should invest in the research and technology, sometimes going back and forth and, at the very least, not moving forward at full force. Should the equipment industry indicate their true intentions, they'd probably have the resources and technology much faster because the developers would know that moving forward with technology development is a wise investment. This brings to mind two points about asking questions - the first is, when someone indicates they don't like your proposal, it's important to follow up. Don't just shrug and move on, ask questions - Why don't they like the proposal? How doesn't it meet their needs? With answers to those questions, you find out more about what's important to them and more about what you should be focusing on. The second point is to state the obvious - something like "I'm confused, with my limited understanding, I think the benefits of moving to 450mm would be beneficial for these reasons (and list them). Is that not something you're interested in? How should I know if "no" means "no" given what happened with the 300 mm development?" Asking open ended questions about their needs and intentions, probing for more information that gives you a full picuture and then earnestly listening for the answers is the best and only way to discern where the industry is going. This will give you a better sense of whether to invest in 450 mm technology or to work on 300 mm productivity improvements. The funny part of all this is that the players probably know what they want (and experience with 300 mm technology proves that). What they don't realize is that they have a much worse chance of getting it (especially quickly) if they don't ask or indicate their interest. Thanks again for raising a valid point. It's important to recognize the impact to the industry when there's miscommunication that leads to lost opportunities for development. Stephen Frenkel Director of Negotiation Programs MWI sfrenkel@mwi.org www.mwi.org

【MES】Yeh! 450mm 的時代即將來臨了

2008.7.14


Yeh! 450mm 的時代即將來臨了。

猶記得去年Applied Materials Brooks Automation Intel 等大廠來台舉辦有關300mm Prime的研討會，會中有人提問：設備廠商如Applied Materials 是否會朝向450mm的趨勢邁進，Applied Materials 的回答語帶保留地說，450mm 的設備研發龐大，目前在市場的規模尚未明朗之前，基本上Applied Materials 是不會冒進，反而鼓勵Fab廠商，以加強機台設備的Efficiency，提高生產力的300mm Prime為目標。然而，Chip Maker的龍頭Intel 已然建立首座的450mm Fab廠，而韓國的Samsung 與台灣的TSMC 也積極地透入，要求設備廠商提供解決方案。暫且不估算450mm的設備投資有多大，有了買家之後，且這趨勢將無法避免，各設備廠商當然會暗中較勁，而台面 上卻嚷嚷450mm還不到時候。 但由Sematech近來對450mm 的Test Bed測試來看，450mm 已不再是搖不可及的目標，只是在2012年以前所有的設備廠商能否能到位(Ready)，這目標會比較有爭議。

當初主要的設備廠商極力推薦300mm Prime的主要目的在於致力改善OEE (Overall Effecitiveness & Efficiency)。他們宣稱300mm的Fab廠的稼動率並沒有發揮極致，有很大的改善空間。事實也的確如此。但，問題在於OEE的改善需要Fab 廠與設備廠商兩者互相配合，才能有改善的成效。然而，現實環境裡設備廠商不是晶圓的製造者，所以無法真切感受到Fab廠的痛處。例如：Fab廠的 Stocker一直為人詬病，因為其無法確實掌握Carrier的動態。再者，RTD已被300mm列入必備控管Carrier移動管理的利器，然而 Rule Based的規則建立，需長時間的觀察學習，致Carrier的運載管理的效能打了折扣。 依我的看法，與其將力氣花在提升機台稼動率的改善，來提升良率，到不如直接跳入450mm的行列，運用產能的擴張，來改善良率的不足，之後再去Tune機 台的稼動率來的實際。之所以會有這種建議，是來自於: (1) 450mm 是個無可避免的趨勢，Fab廠除非自願提早下車，否則還是趕緊加入。(2) 將良率提升由70%至80%，與80%提升至90%兩者所投入的人力資源是無法相比的，後者更可能受限於環境技術等因素，所以用傳統的思維去考量，用高昂 的機台設備去量產，縱然良率相同，但是產出卻是驚人的。在Time to Market的時代，時間就是金錢，切莫等到機台良率都達到完備了，再想到市場的需求，這時為時已晚了。

參考

http://www.eetasia.com/ARTP_8800534242_480200.HTM

Sematech: 450mm program is on track

Posted: 14 Jul 2008

International Sematech is moving full speed ahead with its 450mm programs, but the question is whether the industry can meet its lofty goals in building 450mm fabs by 2012.

On July 9, chipmaking consortium Sematech provided an update on its next-generation 300- and 450mm programs, saying that they are on track and making steady progress.

The consortium is up and running with its "factory integration test bed" facility for the development of 450mm fab tools. Sematech is also testing silicon wafers based on 450mm technology. And the group claims it has made progress on its so-called "Next Generation Factory" (NGF) program, geared to bring lower costs and reduced cycle times in 300mm wafer manufacturing.

Recently, Sematech unveiled two next-generation fab programs: 300mmPrime and the International Sematech Manufacturing Initiative's ISMI 450mm effort.

Need for 450mm?

There is widespread support among the fab-tool community for 300mmPrime, which looks to boost the efficiency of existing 300mm fabs, thereby pushing out the need for 450mm plants.

The newer, more controversial ISMI 450mm program, announced last year at Semicon West, calls for some chipmakers to make a more direct transition from 300mm to the larger 450mm wafer size.

Many fab-tool vendors are reluctant to endorse the next-generation wafer size or devise 450mm tools, saying that it is simply too expensive. Many vendors claim that 300mm fabs are suitable for most applications and the real goal for the industry is to improve the productivity of current plants.

"There is still a lot of concern and debate" about 450mm fabs among the equipment makers, said Scott Kramer, VP of manufacturing at International Sematech, but "the tide has shifted over the last 12 months."

A few fab-tool and materials vendors have develop 450mm technologies, but many suppliers have publically slammed Sematech's 450mm program, saying the economics simply don't add up.

However, the mood is somewhat beginning to change, especially when Intel Corp., Samsung Electronics and Taiwan Semiconductor Manufacturing Co. Ltd in May reached an agreement on the need for industry collaboration for 450mm wafers starting in 2012. Intel, Samsung and TSMC indicate that the semiconductor industry can improve its return on investment and reduce 450mm research and development costs by applying aligned standards, rationalizing changes from 300mm infrastructure and automation, and working toward a common timeline.

Intel, Samsung and TSMC represent a major chuck of the world's capital equipment buyers. Because those companies are pushing for 450mm fabs, it could jumpstart the development of the next-generation wafer size.

Many believe that 450mm tools will not be ready in the 2012 timeframe. Even Kramer acknowledged that the 2012 timetable for 450mm fabs is "very aggressive."
"Those are risky goals," he said.

300mm vs. 450mm

To jumpstart the 450mm era, Sematech last year announced a plan to devise a "factory integration test bed" facility for the development of 450mm fab tools. The proposed facility would help enable chip-equipment makers to develop the initial fab-automation gear, such as carriers, load ports, modules and other items.

Providing an update on the "test bed," Tom Abell, 450mm program manager at Sematech, said the facility is operational. At present, Sematech has put the "test bed" at the Advanced Technology Development Facility, the consortium's former R&D foundry. Based in Austin, Texas, that facility was recently sold to SVTC Technologies Inc.

The facility is using the first 450mm handlers from Brooks Automation Inc. and carriers from Entegris Inc. The pitch specification for these tools is 10mm. With the fab-automation gear, Sematech has demonstrated a 450mm wafer running at 100,000 cycles, Abell said.

Sematech is also in the process of developing a standard for 450mm silicon wafers. At present, there are five wafer-thickness standards vying for dominance in the arena, each with their own set of "tradeoffs," Kramer said.

Initially, Sematech is exploring 450mm wafers with an overall thickness of 925-micron. Last year, Japan's Nippon Mining & Metals Co. Ltd claimed to have developed the first 450mm polycrystalline silicon wafers. Sematech is testing wafers from Nippon Mining, but the consortium is also talking to other silicon wafer suppliers, Kramer said.

NFG Program

The consortium also claims it has made progress on its 300mm NGF Program, which focuses on global infrastructure for 300mm hardware and software. It includes 300mmPrime and is supported by ISMI's four other programs in continuous improvement, 450mm manufacturing, metrology, and environment, safety and health.

"The 300mm NGF Program offers a wider look at 300mm productivity with a broader set of initiatives—and it works for companies whose business plans don't necessarily include a larger wafer size," said Kramer in a statement last year. "Our priority is to extend productivity improvements to existing 300mm fabs in addition to supporting 'green field' facilities."
The overall goal of the program is to reach a 30 percent reduction in cost per wafer, and a 50 percent reduction in cycle time. Like last year, Sematech said it has not been able to reach those targets.

In new simulation data, the consortium claims it is coming closer to its goals. It has simulated a 30- to 40 percent boost in cycle times and 10 to-15 percent improvements in cost. In other data, it has demonstrated a 60 percent boost in cycle times and a 10 percent improvement in cost.

The bottleneck remains in moving the wafer lots from one tool to another. The goal is to process wafers without any delays, according to Sematech.

- Mark LaPedus
EE Times

【MES】MES 客戶端 UI 的演進

2013.9.4

MES系統的開發源自製造業的需求，而早期 (甚至現在) 系統開發人員以程式設計技術背景為主，因此系統開發著重在功能面，而甚少注意到客戶 Client 端 UI 的要求。

早期在恐龍時代，以Digital (後被 Compaq 併購） 之 Vax 系統，堪稱喧騰一時的 MES 系統 Promis，在 Client 端的處理是 Terminal Based，即客戶透過終端機，連線到此 Mini Computer，而客戶在螢幕前所見僅是行列之文字，要求客戶手動敲入資訊，再送達給遠端伺服器處理。Promise 後來補強UI的介面，包裝為獨立的套件。當年半導體龍頭台積電TSMC就是使用Promis的領頭羊，而台灣興起的各半導體廠也隨後跟進，捧紅了Promis。然而，Promis 若非台積電有IT大軍Support，同時有當年Promis 的Vendor 為其跨刀 (當然使用者付費)，這軟體還真難用，光是那報表Report 的開發，如今有能力開發的人也汲汲可數。當年也享譽一時的MES系統Work Stream也是Terminal Based。這軟體公司被Applied Materials 購買之後就欲振乏力，從此消失匿跡。 在業界以賣半導體機台設備的 Applied Materials 玩死很多很棒的 MES 系統，這也已經是這行業眾所皆知的事實。當年據其內部業務透漏，這些MES系統的營收比例幾乎是Applied 的雞肋，而促銷 MES 的業務在Sales Reviews 會議中幾乎無法昂首高論，所以那些MES會落得那樣的結局也難怪了。

後來資訊系統架構拜Windows / Intel 聯手躍進，頓時業界一面倒，傾向擁抱 Client / Server 架構，而捨棄恐龍時代那集中式的處理。PC 的普及，Microsoft 當道，MES 系統在 UI 端的開發難免不向Microsoft 試好，而採用所謂易學、易開發的工具 Visual Basic。當然，Visual 系列的開發工具幾乎成了台灣軟體業的王道，而Microsoft也成了眾多程式設計師欲擁抱的大腿。如今回顧，我們應該跟Oracle 的CEO Larry Ellison 懺悔說：「我們錯了！」當年Oracle 採用 C 語言，建構出如今已是獨霸全球的 Oracle 資料庫系統，而當年選用 Visual Basic 的廠商，在Microsoft 式微之後，宣布 Visual Basic 5.0 被淘汰之後，這決定鐵定讓當年以此辛勤耕耘的廠商只能苦吞。想想一套運行穩定的應用程式耗費廠家多少人力物力，豈僅是當年那購入成本所能比擬的？這就是為何台積電 TSMC 寧可汰換掉 Promis那周邊應用，保留Promise 核心 Kernel，而竭盡戮力尤其周邊開始發展個個應用系統，也決不隨著時代潮流，替換掉這老古董 Promis。不過說實在話，類似Visual Basic 這類UI開發工具，當年的確省卻客戶端敲打鍵盤的辛苦，而很多輸入可由滑鼠代勞。但，以Visual Basic 這工具開發系統仍有其缺點。畢竟以此工具開發，系統 UI 是滑鼠「硬」擠出來的，所以對程式設計開發人員而言，UI 最好一致，變動愈少愈好，有很多時候在開發中不得不犧牲客戶端的和諧性 User Friendly。再者，這類工具以現今 e 世代來看，這些 UI 實在是不美觀啊！

引發我寫此文的主要原因在於思索如何將 IT 的新工具納入 MES 系統的開發。 這想法也適用於其他應用系統的開發。在 UI 端，客戶的工作是多工，即 UI 端只是客戶與系統的介面，因此不同的應用系統開發商應「犧牲小我」，儘可能以瀏覽器 Browser作為客戶端的介面，而不是另行開發特定的UI介面。這時用戶IT的職責在於如何將不同的應用系統，無縫地整合在此單一的UI介面上。當介面以瀏覽器為軸開發時，UI 介面不再類似以 Visual Basic 那般，只以功能面為中心，而能回歸到使用者，由使用者的角度去設計，納入更多的人性、美感，而提昇每個操作頁面 Web Page的質感。過去我們在系統以能滿足客戶的功能性為主，如今應強調應用服務的整合性與 UI 的介面美學。目前市面上有許多製作 UI 的工具，例如：Adobe 的 Dreamweaver、Photoshop，Autodesk的 3D Max 等等，後者可視為網頁開發的輔佐工具。所以下次新一代的MES系統要開發時，除了招募優秀的技術人員，更別忘了美工人才的重要性。我相信一套「賞心悅目」的應用軟體更能提高工作效率呢 !