Misinformation: Intel, TSMC

[user_2013101]

The misinformation is not limited to Intel’s questionable claims of density superiority that I had written about. (Links at end)

[h=3]Immersion lithography[/h]
It is the most critical step in IC production. According to Wikipedia:

"Immersion lithography is a photolithography resolution enhancement technique for manufacturing integrated circuits (ICs) that replaces the usual air gap between the final lens and the wafer surface with a liquid medium"

The benefit:

"The resolution enhancement from immersion lithography is about 30-40%"

"The successful emergence of immersion lithography comes not just from its ability to extend resolution and depth of focus, but also from its timely introduction to the industry between 65 nm and 45 nm nodes.

The status

"Intel's 32 nm process uses second-generation high-k, metal gate technology, but this will be the first time Intel has deployed immersion lithography."

"On March 23, 2012, with the release of the Ivy Bridge chip, Intel's Senior Fellow Mark Bohr stated that the company will be able to extend its current immersion process to the 14-nm and even 10-nm chips before EUV would be necessary."

Immersion lithography - Wikipedia, the free encyclopedia

Most people will develop an impression, from the Wikipedia, that Intel is the pioneer and leader of this key technology.

[h=3]The misinformation[/h]
The data itself is basically true. But, the impression is false.

If the data is correct, then, where is the misinformation?

The misinformation is in the deliberate omission of the crucial point: immersion lithography was invented in 2002 by Dr. Burn Jeng Lin, VP of R&D at TSMC. TSMC is the pioneer; Intel, a follower. But, nowhere in the Wikipedia mentions Lin or TSMC. This is another example of the disingenuous pro-Intel misinformation.

In 2013, IEEE honors Dr. Lin with an award for this invention.

27th June 2013
IEEE honours inventor of immersion lithography

The inventor of immersion lithography – Dr. Burn J. Lin - has been recognised by the IEEE with the award of the IEEE’s 2013 Jun-ichi Nishizawa Medal.
...
“His perseverance in convincing the industry to make the change to immersion lithography has extended Moore’s law from 40nm to potentially as low as 10nm. That at least 82% of all transistors currently in the world have been made with immersion lithography is a testament to Lin’s impact,” IEEE said in its citation.

IEEE honours inventor of immersion lithography | Electronics Weekly

[h=3]More backgrounds[/h]
Dr. Lin invented the lithography itself in the early 1970’s at IBM Thomas J. Watson Research Center, Yorktown Heights, New York. The following is a paper by Dr. Lin published in 1975.

Deep uv lithography

He is also a member of the US National Academy of Engineering, an IEEE fellow and SPIE fellow.

In 2009, Lin received IEEE CLEDO BRUNETTI AWARD. IEEE wrote: "Dr. Lin has continued the cause for immersion lithography with groundbreaking papers that have mapped out scaling laws for super-high numerical aperture immersion optics, and he has led the development of defect-reduction methods to address concerns regarding the technology."

In Taiwan, Dr. Lin is considered "the father of lithography." In the US, Intel invented everything in chip-making.

<script src="//platform.linkedin.com/in.js" type="text/javascript">
lang: en_US
</script>
<script type="IN/Share" data-counter="right"></script>

 

[user_2013101]

Prior posts on Intel’s questionable density claims:

TSMC 20nm A8 is denser than Intel's 14nm Core-M, by a wide margin:
https://www.semiwiki.com/forum/f2/samsung-strikes-chip-deal-apple-4864-4.html#post17087

The pro-Intel virtual reality
https://www.semiwiki.com/forum/f2/10nm-intel-vs-tsmc-4919-2.html#post17173

Intel advantages falsely concluded
https://www.semiwiki.com/forum/f2/samsung-strikes-chip-deal-apple-4864-4.html#post17079

TSMC’s processes vs. the Intel model
https://www.semiwiki.com/forum/f2/samsung-strikes-chip-deal-apple-4864-4.html#post17097

 

[user_2013101]

Intel’s pervasive PR has been extremely successful among the general public.

Intel is an American hi-tech icon, but most Americans don’t know where Taiwan is located. When Intel declares that it commands widening leads over TSMC, none has doubts. People take it for granted. Media repeats the perception millions of times. 99.99% of the population are not IEEE members. The public has no way to understand the true picture. But, I hope the semi professionals know better.

 

[Fred_Tan]

It's not just Intel PR working hard these days.
The Korean wannabes are equally, if not, harder at work

 

[lefty]

Dear user 2013101, did you know that Wikipedia can be edited !
Anyway, I fixed the the entry for you (if someone deletes my edit, then we have evidence of a Intel PR conspiracy :-D)

 

[user_2013101]

The info above, re Immersion and its inventor Dr. Lin, is a presentation of facts.

The following, however, are my personal speculations.

TSMC has reiterated multiple times its 10nm schedule, which is insanely aggressive: to enter 10nm volume production in or before 4Q16, less than 18 months after the 16FF+, while Intel admitted that 10nm is hard to implement and declines to commit itself to a fixed date.

Either TSMC has set itself up for a humiliating disappointment.

Or, purely out of my own speculation, perhaps TSMC has developed some sort of unannounced breakthroughs, as innovative and pivotal as immersion it invented before, so that it is confident in the 10nm delivery.

 

[hist78]

About Dr. Burn J. Lin

Dr. Burn J. Lin joined TSMC in 2000. Like many other TSMC’s executives (probably not including Dr. Morris Chang), he likes to keep low profile and not too often to grant interview opportunities to media. But for Dr. Lin, there are two exceptions. First exception is because he is a devout Christian (since he was 14 years old), several Taiwan Christian media organizations have interviewed him throughout the years. Second exception is that many interviews were conducted around this summer after he was elected as Academicians of the prestige Academia Sinica in Taiwan.

Here are some interesting information about Dr. Lin I gathered and translated from some reports originally written in Chinese.

1. Dr. Lin was born in Vietnam (1942).
2. Dr. Lin is a member of US National Academy of Engineering. He holds 66 patents and published 121 papers.
3. He moved to Taiwan to start his high school senior year and went on to study Electrical engineering in National Taiwan University. Later he got his Ph.D. in Electrical Engineering from Ohio State University (1970)
4. Dr. Lin plays tennis and loves photography. When he was young, his mother bought him a camera that started his interest in photography, optical, and light. And he claimed that photography helped him to secure the heart of his wife when they first met.
5. He spent 22 years at IBM T.J. Watson Research Center and held various research and management positions until his retirement in 1992.
6. During his early years at IBM, X-Ray was the main direction of the IBM. Often in a meeting he was the only proponent of using light to improve manufacturing technology. Once his department head, also a X-Ray researcher, made a joke saying Dr. Lin is the lonely person of the optic/light technology in the whole department. He replied that “there are many empty seats available in the optic/light research and a lot more opportunities there “
7. As a devout Christian, Dr. Lin often uses words from the bible to describe the situation. In terms of Immersion lithography, he said: “God used water to save the fish and God made another miracle to allow water to save the semiconductor industry.”
8. Dr. Lin was recruited by TSMC senior vice president (R&D) Dr. Shang-Yi Chiang (retired) to join TSMC in 2000. Dr. Lin recalled that he didn’t know Dr. Chiang actually took a lot heat to support his research.
9. Talking about Dr. Shang-Yi Chiang. He was the person Dr. Morris Chang depended upon to make the TSMC into a technology leader instead of a follower. When Dr. Chang retook TSMC CEO position in 2009, he also persuaded Dr. Chiang to come out from retirement to lead TSMC’s R&D again.
10. In one of the report/interview, it stated that TSMC is targeting to use Immersion lithography into 7nm manufacturing process (not sure if that’s direct quote from Dr. Lin).
11. When a church in the county that TSMC is headquartered kept shrinking to only 2 families attended, Dr. Lin along with four other families decided to join that small church. In ten years they rebuild that church and expand it to 100 attendees.

 

[user_2013101]

About Dr. Burn J. Lin

...

Thanks for the additional info on Dr. Lin. I was not aware of such details.

I thought Lin joined TSMC in the 1990’s, shortly after his retirement from IBM.

 

[hist78]

Thanks for the additional info on Dr. Lin. I was not aware of such details.

I thought Lin joined TSMC in the 1990’s, shortly after his retirement from IBM.

For about eight years Dr. Lin founded his own company Linnovation, Inc. after retired from IBM. After IBM, he thought he won't be an employee again and probably some bigger company will buy his start-up. Why he became an employee again, he told a reporter in one of the interview that he thinks it's all arranged by God. TSMC's headquarters is actually not too far from the high school he completed his senior year when he was only 17 years old. He said to become an employee of TSMC brought him back to his starting point.

 

[msporer]

Intel, along with Micron, deployed immersion lithography in the IMFT NAND Flash joint venture significantly earlier than when it was deployed in Intel fabs. Just another datapoint.

 

[user_2013101]

Another prevalent perception: FinFET is an innovation from Intel. After the introduction of Intel’s 22nm FinFET in 2012, foundries scrambled to follow.

Indeed, Intel is the first company to implement FinFET in volume production. However, Intel didn’t invent FinFET. It is invented in 1999 by a research team at Berkeley, led by Dr. Chenming Calvin Hu. The invention earned Dr. Hu an IEEE Solid State Circuits Award in 2002.

Intel jumped on the FinFET idea immediately:

Chenming Hu: It was an instant hit. I remember Cathy and I were invited to Intel Santa Clara just a couple of months after the publication, and in that same year, 2000, I was invited to Intel Oregon twice.

9 May 2011
The Origins of Intel's New Transistor, and Its Future

A Q&A with Chenming Hu, coinventor of both the FinFET and its likely competitor

The Origins of Intel's New Transistor, and Its Future - IEEE Spectrum

Note that Dr. Hu went on to work for TSMC, 2001-2004.

Chenming Calvin Hu - Wikipedia
Chenming Hu - Wikipedia, the free encyclopedia

That is, TSMC has been doing researches on FinFET at least since 2001.

Don’t be surprised if next year the 16FF+ rolls out smoothly and exceeds Intel’s 14nm.

P.S.
Intel uses the term “TriGate” instead of the commonly-known “FinFET,” perhaps to confuse the public as if it is Intel’s invention.

 

[user_2013101]

The US attracts talents from all over the world, and provides an environment that fosters numerous innovations.

However, unfortunately, a lot of times, the US fails to take advantage of such talents and innovations. The most infamous example in the semi industry is Morris Chang, the founder and chairperson of TSMC.

By early 1980’s, Chang had risen to become an executive, and a contender to the top position, at Texas Instruments, which was the world’s semi leader at the time before Intel rose to stardom. But, TI didn’t appreciate Chang’s fab-less model; later, TI let Chang go, perhaps having something to do with his ethnic background.

In 1985, Chang returned to Taiwan and founded TSMC in 1987. During the process of setting up the new company, Chang approached TI, Intel, and many other US and Japanese tech giants, for capital investments in TSMC. None responded. At end, due to the lack of interests, Taiwan government had to put up the final 20% capital for TSMC to start.

The rest is history.

Similarly, IBM didn’t treasure Dr. Burn Jen Lin’s groundbreaking works on lithography. Dr. Lin retired early from IBM and left behind a pile of “useless” research papers and patents. He was later recruited by TSMC.

Intel contacted Dr. Chenming Hu, the inventor of FinFET, first; but it’s TSMC that hired Dr. Hu.

And, Dr. C. Y. Lu, the inventor of BCDMOS and BE-SONOS; (similar story: numerous awards, now the head of Taiwan-based Macronix) …

Etc.

An exception: Jen-Hsun Huang remains in the US and started nVidia here, instead of returning to Taiwan.

P.S.
Intel turned down Steve Jobs’ request to develop a mobile processor for the iDevices.

 

[user_2013101]

More pitch sizes, from the 14/16nm processes, are revealed, at the IEDM on Dec 15-16.

One article declares "IEDM was a victory lap for Intel, which gave more details on the 14 nm finFET process." The article lists some selective items of Intel’s pitch sizes, but didn’t provide any numbers from TSMC’s 16FF+.

Another news story includes the pitch numbers: Intel from 52nm to 160nm, vs., TSMC’s 64nm to 90nm. The fin itself is 42nm from Intel, to TSMC’s 48nm. A writer can arbitrarily pick any number from the list and justify the superiority for his favorite company.

I don’t mean all these authors are part of Intel’s PR machine, although some obviously are. There is also some sort of unspoken and subtle political correctness at work here. If I have to write a formal article comparing Intel and TSMC’s technologies, I will certainly conclude that Intel still has leads, regardless, while knowledgeable people can see a more realistic picture from the included details.

I have to appreciate the fact that TSMC keeps a low profile and demonstrates modesty. Anyway, most of TSMC’s customers are US companies, including Intel; it is not a good business practice to violate the feelings of own customers, unlike Samsung’s attitude toward Apple.

Perhaps Morris Chang had learned, from his personal experience of being driven out of TI’s executive ranks in the 1980’s, how to work with the American pride.

P.S.
Samsung, whose 14nm is supposed to be far superior than TSMC’s 16nm and thus has won Apple’s A9 contracts, didn’t offer a presentation at IEDM. No way to gauge Samsung’s “superiority”; you just have to believe it.

 

[hist78]

"P.S.
Samsung, whose 14nm is supposed to be far superior than TSMC’s 16nm and thus has won Apple’s A9 contracts, didn’t offer a presentation at IEDM. No way to gauge Samsung’s “superiority”; you just have to believe it."

Did any one ask why Samsung is missing from the discussion?

 

[hist78]

I think you're probably talking abut this article at:

11 Views of IEDM | EE Times


"Samsung presented eight papers but none that described its 14 nm process. That’s a good sign it is probably coming soon, but development was still too immature when papers were due in late June."

WTH!

I think on this semiwiki.com we are doing much better. We won't see this kind of comment here. I hope.

 

[user_2013101]

"P.S.
Samsung, whose 14nm is supposed to be far superior than TSMC’s 16nm and thus has won Apple’s A9 contracts, didn’t offer a presentation at IEDM. No way to gauge Samsung’s “superiority”; you just have to believe it."

Did any one ask why Samsung is missing from the discussion?

hist78,

It seems you didn’t fully appreciate the sarcasm in my statement.

Had Samsung delivered a presentation, it would have exposed the nature of its inferior 14nm and destroyed the myth that it had won 80-100% of A9 orders.

 

[hist78]

hist78,

It seems you didn’t fully appreciate the sarcasm in my statement.

Had Samsung delivered a presentation, it would have exposed the nature of its inferior 14nm and destroyed the myth that it had won 80-100% of A9 orders.

OK, you got 6 stars out 5-star rating system.

 

[khaki]

Just a general comment. No way you can gauge the superiority of a given technology based on IEDM presentation anymore. Arbitrary units were used in record high this year going to the extreme of using x% and y% (without specifying x and y, of course) over already normalized charts.

 

[user_2013101]

Just a general comment. No way you can gauge the superiority of a given technology based on IEDM presentation anymore. Arbitrary units were used in record high this year going to the extreme of using x% and y% (without specifying x and y, of course) over already normalized charts.

If I understand correctly, the sizes of these pitches, gates, and fins, if used properly, at best, can show some crude and theoretical upper limits of density. That’s all.

Unfortunately, the marketing departments seem to have taken over and can pick whatever numbers in the range, from 42-160nm, to serve the PR purposes. It more or less has turned into some sort of futile excises of egos.

In the real marketplace, density advantage may not be so critically important, after all. Whatever the density, Intel is still struggling in the mobiles; and, ARM-based still cannot penetrate the servers. The current situation resembles a stalemate, which I don’t see will change meaningfully in the next 4-5 years.

Intel’s China Strategy

Perhaps Intel’s China strategy may help its mobile endeavor, but carries big risks.

In addition to partnering with China’s Rockchip and Spreadtrum, Intel has significantly upgraded its chip plants in China.

Dec 4, 2014
Intel to invest US$1.6 billion in China plant to push mobile chip development | PCWorld

In the short-term, this strategy may help Intel, but not necessarily so, longer term.

Once Intel, and other companies like Qualcom, helps build up the Chinese capacity in chip design and production, China is likely to throw the American companies out of China.

2014-12-18
China is Planning to Purge Foreign Technology and Replace With Homegrown Suppliers
https://www.bloomberg.com/news/2014...ing-shift-from-foreign-technology-to-own.html

Qualcom had helped China’s SMIC produce 28nm chips.

"Our collaboration with Qualcomm Technologies has been key in helping us to accelerate our 28nm technology development and achieve this key milestone"

Dec. 18, 2014
SMIC Successfully Produces Qualcomm Snapdragon 410 processor in 28nm Process -- SHANGHAI, Dec. 18, 2014 /PRNewswire/ --

 

[user_2013101]

The China strategy of Intel and QCOM has forced TSMC and MediaTek to consider similar moves that Taiwanese companies won’t do otherwise.

Is there any safeguard on the technology transfer to China?

These hi-tech icons profit on monopolies, slow to improve technologies, rarely deliver innovations; instead, they resort to dishonest PR. The best strategy they can think of is to transfer technologies to China, to temporarily gain some market shares, after sending the US manufacturing bases to China in the prior decades. It is a sorry situation.