Silicon wafer maker Wafer Works gears up for capacity expansions
Printer friendly
Hans Wu, Taipei; Jessie Shen, DIGITIMES [Friday 26 February 2010]

Taiwan-based Wafer Works, which makes small- to medium-sized semiconductor-grade wafers, plans to ramp its wafer-producing capacity in 2010 due to increasing orders. The company has revealed it will initially ramp up the output of its 6-inch production lines to 600,000 wafers a month in mid-June from 500,000 units at present, followed by additional ramps to be carried out in the first quarter of next year.

Wafer Works said it will implement a capacity expansion project at its Yangmei, Taoyuan plant, in the second half of 2010. The company is also set to undergo capacity expansion at its newly-established plant in the Longtan Science Park, Taoyuan in the first quarter of 2011. The latter expansion will focus on 8-inch wafers.

Wafer Works is positive about the market for 8-inch wafers, thanks to increased demand for power management ICs and the China market, the company noted.

Wafer Works has seen order visibility stretch to the second quarter of 2010, according to the company, adding that its manufacturing facilities in Taiwan and China are all running at full capacity. Orders have significantly exceeded capacity, the company said.

In other news, Wafer Works' epitaxy wafer subsidiary Wafer Works Epitaxial has cut into the supply chains of several foundries for automotive ICs in China, according to a Chinese-language Commercial Times report. The foundries include Taiwan Semiconductor Manufacturing Company's (TSMC's) Songjiang, China fab, Hejian Technology and Semiconductor Manufacturing International Corporation (SMIC).

Located in Shanghai, Wafer Works Epitaxial produces 6- and 8-inch epitaxial wafers and plays a part in its parent company's vertically-integrated business model.

華虹NEC五大特色工藝贏得多個熱點市場

出自：電子工程專輯 編輯：俞文杰
雖然已是各種智能卡芯片代工市場老大，但是上海華虹NEC電子有限公司（“華虹NEC”）絕不滿足于人們將之看作一家智能卡芯片的廠商。華虹NEC自2002年停止DRAM生產(chǎn)、轉(zhuǎn)向?qū)�業(yè)代工算起，就確立了通過特色代工工藝吸引戰(zhàn)略客戶的發(fā)展道路，并在激烈競爭的代工市場領(lǐng)域建立起了獨特的優(yōu)勢。“多年前，華虹NEC第一個采用8英寸生產(chǎn)線來制造MOSFET等分立器件時，被當(dāng)時絕大多數(shù)業(yè)內(nèi)人士指責(zé)殺雞用牛刀，然而，事實證明華虹NEC這一步棋走對了。目前華虹NEC分立器件累計出貨超過100萬片晶圓，這種新型代工業(yè)務(wù)模式也被競爭對手所效仿。”華虹NEC市場副總裁高峰對本刊表示。除了嵌入式非揮發(fā)性存儲器外，華虹NEC還在功率器件、模擬/電源管理IC、高壓CMOS以及基于SiGe（鍺硅）工藝的射頻等特色工藝領(lǐng)域不斷創(chuàng)新，并取得了不俗的業(yè)績。展望2010年，華虹NEC的五大特色工藝正好匹配了目前最熱門的一大部分產(chǎn)品與應(yīng)用，并將為廣大客戶、特別是本土客戶帶來高性價比的產(chǎn)品與服務(wù)。

從智能卡到MCU再到觸摸屏

從0.35微米工藝到0.13微米工藝，華虹NEC一直保持著嵌入式非揮發(fā)性存儲器技術(shù)的領(lǐng)先地位。目前華虹NEC正在開發(fā)的創(chuàng)新技術(shù)是1T（一個晶體管）快閃存儲單元�！�1T技術(shù)比現(xiàn)有技術(shù)可在更小的晶圓尺寸上實現(xiàn)更大容量的存儲單元，特別對于768KB或以上容量的存儲單元面積縮小更為顯著，這正是3G/4G手機終端的大容量SIM/JAVA卡所必需的�！备叻逭f道。

盡管如此，智能卡業(yè)務(wù)僅占華虹NEC不到30%收入比例，華虹NEC一方面加大其余特色工藝平臺的技術(shù)創(chuàng)新與市場開拓，另一方面也大力將自己領(lǐng)先的嵌入式非揮發(fā)性存儲器技術(shù)向非智能卡產(chǎn)品領(lǐng)域拓展與延伸。

將嵌入式非揮發(fā)性存儲器技術(shù)拓展到MCU領(lǐng)域就是華虹NEC正在做的一個工作。在MCU市場，華虹NEC特別注意扶持中小型的潛力客戶�！盀榇�，我們采用了兩條腿走路的方式：一方面我們與國際IP廠商合作，以服務(wù)大的客戶；另一方面我們與臺灣及本土IP廠商合作，引入低價位的CPU核、同時配以良好的技術(shù)支持，這對中國IC設(shè)計公司非常重要�！蹦壳皫准冶就羶�(yōu)質(zhì)MCU廠商已是華虹NEC的重要培養(yǎng)客戶。

還有一個目前非�；鸬氖袌鲆彩侨A虹NEC的嵌入式非揮發(fā)性存儲器工藝可以延伸的領(lǐng)域，這就是觸摸屏市場�！坝|摸屏產(chǎn)品可以用到我們的多個工藝，包括集成嵌入式非揮發(fā)性存儲器和高壓CMOS的特色工藝，這一創(chuàng)新技術(shù)我們已開發(fā)出來，今年就開放給用戶使用�！备叻逋嘎�。在LCD驅(qū)動領(lǐng)域，華虹NEC是很多知名芯片廠商的代工伙伴，處于該領(lǐng)域的業(yè)界領(lǐng)先地位。現(xiàn)在，利用這一優(yōu)勢工藝，結(jié)合嵌入式非揮發(fā)性存儲器，華虹NEC希望能在觸摸屏市場大展拳腳。

BCD工藝迎合節(jié)能環(huán)保熱點

2010年最熱的LED照明/背光、高端電源管理和汽車電子中都缺不了BCD（雙極，CMOS和DMOS）工藝的器件，事實上，BCD基礎(chǔ)器件已成這些熱門應(yīng)用的核心器件。而華虹NEC已開發(fā)和正在開發(fā)的BCD工藝技術(shù)的節(jié)點涵蓋0.5μm、 0.35μm、0.25μm與 0.18μm，其中面向高端電源管理的“0.25/0.18um BCD”還是國家重點支持的重大科技專項項目之一。

“700V BCD是一種創(chuàng)新，基于該工藝制造的IC可直接工作在220V的市電下，而不用像以前那樣通過多個器件來實現(xiàn)。這樣就大大提高了效率，符合節(jié)能環(huán)保趨勢�！备叻褰忉尅Ｋ�透露目前已有國內(nèi)IC設(shè)計公司設(shè)計出700V的BCD器件，并正在與華虹NEC合作。而在LED照明和大尺寸LED液晶電視背光方面，華虹NEC已成功推出40V與60V的BCD工藝技術(shù)平臺，客戶包括國內(nèi)與海外廠商�！癇CD工藝技術(shù)的發(fā)展不像標(biāo)準(zhǔn)CMOS工藝那樣遵循Moore定律，而是向高壓、高功率、高密度三個方向發(fā)展�！彼�解釋。

在BCD和MOSFET的另一大應(yīng)用——汽車電子市場，華虹NEC已獲得了汽車電子符合性認證，正在與合作伙伴一起準(zhǔn)備產(chǎn)品認證。

鍺硅射頻技術(shù)讓無線產(chǎn)品更便宜

3G、WLAN/WIFI以及衛(wèi)星機頂盒等都是2010年無線市場的大熱，而這些終端產(chǎn)品中有一個占成本比例不小的器件——PA(功率放大器)。目前業(yè)界主流PA都采用砷化鎵工藝，但是砷化鎵工藝的代工廠數(shù)量少，代工價格也很貴，很多時候還會出現(xiàn)供應(yīng)緊缺情況。而鍺硅(SiGe)工藝比砷化鎵的成本要低很多，如果能采用單管的鍺硅代替后者，將為無線產(chǎn)業(yè)帶來一次革命。華虹NEC正在作此項研究，有望在今年內(nèi)取得突破。

“問題的關(guān)鍵是如何讓鍺硅工藝的性能進一步提升，而不增加太多成本。”高峰分析，“我們的策略是先在一些對性能要求不是非�？量痰牡胤教娲�砷化鎵。華虹NEC目前正在開發(fā)0.18/0.13um鍺硅工藝，將用于接收機的LNA(低噪聲放大器)中，同時還會用于對講機的PA中�！彼�同時透露，這個項目也是國家重點支持的項目之一，華虹NEC今后將繼續(xù)開發(fā)性價比更高的鍺硅射頻工藝技術(shù)平臺，用于更高端的產(chǎn)品中。

中國芯片廠商已開始在PA市場扮演重要角色，這個傳統(tǒng)上由國際廠商主導(dǎo)的市場，正在RDA（銳迪科）等中國廠商的帶領(lǐng)下向本土化供應(yīng)商滲透。華虹NEC在鍺硅項目上會不會再帶火一家本土的IC公司呢，我們將拭目以待。

TSMC takes on 40nm yields, high-k, litho issues
At the TSMC Japan Executive Forum in Yokohama, Shang-Yi Chiang, senior VP of R&D at Taiwan Semiconductor Manufacturing Co. Ltd (TSMC), addressed several issues about the silicon foundry giant.


Chiang discussed TSMC's 40nm capacity, yield issues, high-k and lithography. EE Times obtained a transcript of the presentation. Here are some of the issues discussed:

40nm capacity
As previously reported, Nvidia Corp. and others are struggling to get their 40nm wafers from TSMC. "There are four major messages I'd like to deliver in this presentation. Number one is 40nm technology happened to be a very high demand in the early stage, so we saw the customer demand ramp up so quick, more than what we had seen before for 40nm, and we are working very hard to make up the volume," Chiang said.

He added, "At this stage we only have fab 12 ready to tape production of 40nm and we are able to do about 80,000 wafers per quarter at the moment. These are 12-inch wafers. And this will be doubled by the end of this year, to 160,000 12-inch wafers for 40nm capacity by the end of this year, and partly from fab 12 and partly from fab 14."

40nm yield challenge
TSMC also addressed the issues on its struggle with 40nm. "You all heard about TSMC's challenge during the early part of last year. I report to you we are glad all this problems was behind us. We resolve this yield problem in the second half of last year. So we're glad the yield issue was over, and we are building the capacity very aggressively to fulfill the very high demand from our customers."

Chiang added, "Moving to 45- and 40nm is a lot more challenging. This is the first time we began to use 193 nanometer shrink immersion. That means the photo resist during exposure will be merged in water and is a very high potential defect. For this a very big challenge, we began to develop the third generation. We began to use the second generation low k material with a k value of 2.5 and at this k value the material become quite fragile so there is a lot of potential issues in the package side. So moving to 40nm that's why it's getting pretty challenging, pretty difficult to do."

28nm progress
Several months ago, TSMC rolled out its 28nm, which will have several options. According to Chiang, "The first node we're going to release for the 28nm will be we call the 28 LP. This is our poly gate and silicon oxide nitrate version. We will establish production at the end of June this year, about four months from now, and this is for the low-power application. Again, no high-k metal gate."

Going high-k and metal gate
At 28nm, TSMC is expected to have a high-k/metal-gate option. "The first high-k metal gate we call 28 HP for the high performance application will be introduce the end of September this year, and followed by three months later December will be the 28 HPL. This is the first high-k metal gate introduction for the low power application," Chiang said.

"At this moment the only way we know how to do that is the gate last approach. So I firmly believe everybody will migrate to using gate last in the future generation, and could be as early as 22, 20 nanometer mode. Even some of our competitors who claim gate first have unique advantage, I firmly believe they will move to gate last very quickly.

There are for the gate last process. It happen you can achieve the higher constraint from silicon germanium because by the time you remove the poly silicon from the gate the silicon gemainium can push the channel in a little bit to have more strength in the channel. But by far the biggest advantage, and that's why TSMC achieves the gate last is the first one you can have different work functions for different gate metal."

Plans for 22nm
"Going forward, we plan to introduce 22nm node about two years after we introduce 28nm, so the first introduction we'd like to be in the Q3 12, and this if for the high performance version. And followed by the low-power version about the end of Q1 13," said Chiang.

"Going to 22nm and beyond, as we like for a new module to be introduced, in 20nm, 22nm, the first will be we go to the second generation high-k metal gate."

Litho direction
TSMC has inserted 193nm immersion and is looking at EUV and maskless. "We will continue using the 193 immersion with double-patterning in the early stage, and we will migrate to EUV or multiple e-beam direct write if one of these technologies can be more mature and more cost effective," Chiang said.

"And lithography side the industry seems like the mainstream trend will be to go to EUV and you may or may not have heard about the cost of EUV. If you buy one EUV tool with a matched track, it will cost like 80 million dollars for just one tool. I was shocked to sign a P.O. only couple weeks ago shortly before my vacation. I signed a 1.9 million euro($2.57 million) purchase order for a clamp.

This clamp is a custom made clamp only for EUV. We have to mount a special clamp on the ceiling and this clamp will be used to lift the EUV tool when we install the machine, and when we do the maintenance. This tool is so heavy, no other tool can lift it up, and this custom-made clamp costs us 1.9 million euros ($2.57 million), just to buy a clamp. It's really shocking."

- Mark LaPedus
EE Times

Analysis: UMC needs new strategy

Silicon foundry vendor United Microelectronics Corp. (UMC) celebrated its accomplishments during its 30th anniversary at a low-key event in Taiwan.


But will 2010 be UMC's last party? Despite the current upturn and renewed growth at UMC, many wonder if the foundry vendor will survive in the long run, as the company has fallen behind the technology curve, seen a key customer in Xilinx Inc. defect to its rival and been the subject of takeover rumors.

Now, there is speculation that UMC may join IBM Corp.'s fab club or form a new R&D alliance with Texas Instruments Inc., in an effort to play catch-up—or even survive—in the foundry market. Others believe that GlobalFoundries Inc. or its big investor—Abu Dhabi's Advanced Technology Investment Co. (ATIC)—may take a stake in UMC in return for fab capacity.

In fact, there was once a rumor that ATIC would buy UMC. Rumors aside, this could be a make-or-break year for UMC. At one time, the company sat comfortably as the world's second largest foundry vendor, behind neighbor Taiwan Semiconductor Manufacturing Co. Ltd.

But besides TSMC, UMC had little competition to speak of, that is, until recently. Now, it faces competition from a trio of strong players, including TSMC, GlobalFoundries and Samsung Electronics Co. Ltd.

Faced with a new set of rivals, UMC's current strategy may not be viable in today's climate. The company is a member of Sematech, but besides that, the foundry vendor is not part of a major R&D alliance and prefers to go it alone in technology.

And at one time, UMC attempted to keep up in the process-technology race with TSMC. Now, UMC is quietly shifting towards more of a "fast follower" strategy. It continues to develop leading-edge processes, but it is a step or two behind the leaders.

The fast-follower strategy could keep UMC relevant for the near term, but the question is will its customers continue to believe in the company? And can it keep up in the process-technology race without partners? After several requests, UMC refused to discuss its strategy for EE Times.

Clearly, though, UMC is on a slippery slope. "I have talked to some people, who say: 'UMC has totally lost it,'" said G. Dan Hutcheson, CEO of VLSI Research. "UMC fell behind technically in terms of process technology."

On the other hand, UMC is still a viable player, perhaps by default. "The fabless guys don't have a lot of choices (at the high-end). If you don't have a high-volume (product), it's sometimes hard to deal with TSMC," he said. "In some cases, UMC's yields are better than TSMC."

"UMC is still the No. 2 foundry, but they will possibly drop to number three next year," added Dean Freeman, an analyst with Gartner Inc. In market share, UMC could fall behind GlobalFoundries, he said.

A more pressing question is whether UMC can keep up in the technology race. "It's going to be tough, because they don't have a technology R&D partner," Freeman said.

UMC recently rolled out its 45-/40nm process and has even talked about its 28- and 20nm developments. In reality, though, UMC "is not as aggressive in technology" as TSMC and others, Freeman said. UMC remains "six to nine months behind TSMC."

Simply put, UMC is down but not out. "Bottom line: Don't count them out just yet," said Steven Pelayo, an analyst with HSBC

Backtrack
UMC's roots can be traced to the 1970s, when Taiwan started its IC industry. A government-sponsored R&D organization in Taiwan, the Industrial Technology Research Institute (ITRI), was formed in 1973. By acquiring technology from ITRI, UMC became the first chip maker in Taiwan. At the time, it owned a 4-inch wafer fab.

In the early years, UMC developed and made chips for low-tech toys and games. For years, the company was a minor IDM, but it finally gained some notoriety in the 1990s. In the early 1990s, UMC developed its own x86-based microprocessor.

Intel Corp. put UMC on the map by suing the Taiwan chip maker for patent infringement. Shortly after those events, UMC exited from the x86-based processor market and decided to switch from an IDM to a foundry strategy.

In doing so, it spun off its chip units into independent companies. The most notable spin off was Mediatek Inc., which has become a powerhouse in the consumer and cell-phone chip business today.

UMC's early days in the foundry business were colorful and tumultuous. In the 1990s, two of the company fabs mysteriously caught on fire and were destroyed. And at the early part of this decade, UMC's top executives were involved in an alleged scandal over investing in China.

At the time, the Taiwan government forbad the island's companies from investing in chip fabs in China. UMC was allegedly involved in investing in a foundry startup in China called He Jian Technology Co. Ltd.

But following a move by the Taiwan government to relax its rules in China, UMC last year paid $285 million to acquire the 85 percent of He Jian that UMC did not already own. However, that deal is still pending. A recent report by the U.S.-Taiwan Business Council on the Taiwan semiconductor market stated that Taipei is standing in the way of the deal because investment regulations stipulate that there can be no more than three Taiwan chip fabs in China, and that all three are spoken for.

Despite the problems (and itself), UMC turned into a viable foundry vendor. In the early part of the decade, UMC and TSMC were neck-and-neck in terms of sales. And in 2003, UMC beat TSMC to the punch in the 90nm market.

More recently, however, TSMC skyrocketed and its sales soared. In contrast, UMC plodded along and struggled to ship its 65nm technology. Amid a slump in its business, Jackson Hu in 2008 suddenly resigned as chairman and CEO of UMC. UMC announced that Stan Hung was elected to the position of chairman. Shih-Wei Sun was named CEO. Hung had served as UMC's chief financial officer. Sun was UMC's chief operating officer

Shift in strategy?
For 2009, UMC reported sales of NT$88.6 billion ($2.8 billion), down 4 percent compared with 2008, the company said. UMC posted a net income for the year of NT$3.9 billion ($121.9 billion), compared with a net loss in 2008.

UMC recently announced it would increase its capital expenditures for 2010 to be between $1.2 billion and $1.5 billion to accommodate strong demand from customers at advanced technology nodes. The company said it spent about $551 million on capital expenditures in 2009.

In total, TSMC was in first place in the overall foundry rankings in 2009, as the company's sales hit $8.997 billion in 2009, down 15.2 percent over 2008, according to Gartner. TSMC's share fell from 47 percent in 2008 to 44.8 percent in 2009.

UMC was in second place again. The company's sales hit $2.730 billion in 2009, down 7.7 percent over 2008, according to Gartner. UMC's share jumped from 13.1 percent in 2008 to 13.6 percent in 2009.

GlobalFoundries emerged and became the No. 4 player in the foundry business in 2009, according to Gartner. GlobalFoundries, the manufacturing spinoff of Advanced Micro Devices Inc., last year also acquired Singapore's Chartered Semiconductor Manufacturing Pte Ltd. Chartered was in third place in 2009.

2010 started off bad for UMC, when Xilinx recently said it will use leading foundry TSMC as one of two foundry suppliers for its 28nm FPGAs, a major strategy shift that has been the subject of industry rumors and analyst speculation for weeks.

Xilinx said it is using TSMC and Samsung's foundry division to make 28nm parts, which are expected to begin sampling by the end of this year. Xilinx has long used a two foundry strategy at each process node. Samsung first joined Xilinx' foundry supplier roster at the 40nm node, supplanting Toshiba.

Xilinx' shift to TSMC is a bitter pill for rival foundry UMC, which has been a foundry supplier to Xilinx for more than a decade. Some analysts blamed 65nm yield issues at UMC for a supply glitch that last summer materially impacted Xilinx sales, speculation which UMC later denied.

UMC will continue to manufacture Xilinx parts at 65-, 40nm and other nodes. According to a Xilinx executive, TSMC and Samsung offered Xilinx the best process technology options for high-performance, low-power process technology at 28nm.

All is not lost for UMC. "They have lost Xilinx' initial 28nm development work, but they still have majority share of 45- and 65nm (at Xilinx)," said HSBC's Pelayo. "And by the way, the vast majority of Xilinx wafers are still done at 65nm. The 28nm, high-volume production crossover is at least a couple years away—and who knows what will happen until then."

UMC still has a plethora of customers despite the loss. "As a reminder UMC's strategy is to not compete on the leading-edge with TSMC and GlobalFoundries. Xilinx was the only customer where they were the primary leading-edge supplier," he said. "We have confirmed (UMC) has in fact gained some share at Mediatek, Qualcomm and Broadcom. Texas Instruments remains a robust customer too."

In fact, UMC is seeing huge demand. Like most foundries, UMC's biggest problem right now is that its capacity is "very tight," he added. But the loss of Xilinx, coupled with capacity issues, may have prompted UMC to re-think its strategy.

It is planning a private placement of no more than 10 percent of its total shares or about $400 million. UMC did not elaborate. Analysts believe that UMC is gearing up for a new R&D partnership in an effort to play catch-up in the foundry market.

"In general, it sounds like they are positioning for some kind of technology alliance/licensing. The Street may speculate a new relationship with IBM (similar to the IBM tech licenses that SMIC, Chartered/GlobalFoundries and Samsung have)," said Pelayo in a recent report. "Alternatively, this could be related to a potential stronger relationship with a key customer like Texas Instruments."

At one time, there were rumors that UMC would join IBM's "fab club," which includes IBM, GlobalFoundries and Samsung. Meanwhile, ATIC was rumored to have approached UMC in January with a view to taking a stake in UMC and securing additional production capacity. The link was denied at the time

Fab ramp
Meanwhile, UMC is speeding up its fab ramp. The company kicked off official operations of the phase-three and phase-four production facilities at its 300mm Fab 12A at the Southern Taiwan Science Park (STSP).

UMC is also pushing previously-announced 40nm process. "Currently, UMC has a total of 10 fabs continuously operating around the world, in Hsinchu, Tainan, Japan, and Singapore," said UMC Chairman Hung at the company's anniversary party in a statement.

"Now, we offer the most advanced 40nm volume production technologies," he said. "Independent technology development has reached the leading-edge 28nm using gate-last high-k/metal- gate technology, with current plans to achieve silicon IP pilot capabilities by end of 2010. Early this year, UMC also began working with customers on planning and initial development of advanced 20nm technology."

The company could also be cooking up a deal with GlobalFoundries. That means UMC would tweak or redo its process efforts and follow the same path as IBM's fab club. Several years ago, UMC had a process technology effort with IBM and Infineon.

That deal was scrapped because UMC and IBM clashed over the directions of the technology. At the time, UMC didn't need Big Blue's technology and it could remain independent. Now, the circumstances are much different and UMC's future could be hanging on the balance.

Now, three vendors—GlobalFoundries, Samsung and TSMC—appear to be engaged in a new capital spending race in an effort to gain share in the foundry race. The trends could be a problem for the other digital foundries, such as SMIC, UMC and others, many of which cannot keep up with the spending race and will likely fall behind the curve.

The question is clear: Who will be left standing when it's all said and done? TSMC, GlobalFoundries and Samsung will survive, observers said. UMC and SMIC could hit the wall and/or may get acquired, they added.

- Mark LaPedus