Thursday, May 3, 2012

Funding the future: How best to spend limited R&D dollars?

One of the big issues now facing the industry is to how best spend limited funding for research and development, when so much needs to be done. Continued scaling to 20nm and below, a transition to 450mm, 3D integration, device new structures, a bevy of new materials with unknown integration challenges - these are all needed.

These issues are top of mind since I've been asked to moderate a panel on this topic at the upcoming Advanced Semiconductor Manufacturing Conference (ASMC), to be held May 15-18 in Saratoga Springs, NY. The panel, "Competing for R&D Dollars: Funding the Future," will be held on May 16, 2012 from 4:30 to 5:45. The panelists will be:

-- David Bennett, VP Alliances, GLOBALFOUNDRIES;
-- Noreen Harned, VP Marketing, ASML;
-- Subramanian (Subu) Iyer, IBM Fellow, IBM Systems & Technology Group;
-- Nag Patibandla, Sr. Director (Office of CTO), Applied Materials; and
-- Risto Puhakka, President, VLSI Research.

Each panelist will give a brief overview of their perspective on the topic with no more than two slides, and then we'll move into the panel discussion.

I think it's fairly clear that, with the industry's present model, it's going to be difficult to fully finance everything on the wish list. In fact, that's been one of the main arguments against going to 450mm: Wouldn't precious R&D dollars (or euros/yens/RMBs) be better spent on some other endeavor, such as 3D integration?

This isn't a new dilemma, of course. In a SEMI-funded reported published in 2005, Ron Leckie analyzed all the things that were noted in the International Technology Roadmap for Semiconductors (ITRS) and concluded that the funding gap could reach upwards of $9.3 billion by 2010. It's now 2012 and, two years later, Moore's Law is holding firm. On the other hand, there has been a fair amount of consolidation, and a shift to fabless and asset-lite (or capital-lite) business models.

In this blog, over two years ago, I reported that some believed that a massive restructuring is underway that will leave only a handful of companies producing devices at the leading edge. Bob Johnson, VP of research at Gartner, predicts that by 2014 there will be only 10 companies operating at the leading edge: 1-2 non-memory IDMs, 4-5 memory companies, and 3 foundries (speaking at SEMI's 2010 ISS). I haven't seen much to contradict that thinking. The percentage of revenue that chip makers are spending on R&D is fairly low, around 14%. It was at a historical low in 2010, only 12%

The way the industry has been able to get around the funding challenge is through collaboration. Consortia such as SEMATECH and imec are great models of collaboration, enabling companies to work on "pre-competitive" research to share costs and reduce risks. The Common Platform model employed by IBM/Samsung/GLOBALFOUNDRIES is another great example of collaboration. The entire CNSE/Albany Nanotech/G450C effort is perhaps the best example of university, industry, consortia and government working together (as I'm sure will be discussed at ASMC).

What's intriguing to me, though, is if the semiconductor industry will go the way of the automotive and aerospace industries, which have very different models when it comes to cost-sharing and risk-sharing. Mike Splinter of Applied Materials, Steve Newberry of Lam and Paolo Gargini of Intel all made references to those models at this year's ISS.

I don't have the answers, but if you have questions you'd like to pose to the ASMC panel, and aren't going to make it in person, let me know: peters@pennwell.com.

Wednesday, May 2, 2012

Thirty years ago today

Today marks an interesting milestone for me: 30 years ago, I started work as an editor covering the semiconductor industry. I reported to Don Levinthal, publisher of Semiconductor International, at 222 W. Adams in downtown Chicago. The magazine was produced by Cahners, which had recently acquired Kiver Publications. Milt Kiver, as I recall, had won fame and fortune by writing a book on color television repair (Television Simplified, first published in 1946), and went on to start Electronic Packaging & Production (EP&P), Electro-optical Systems Design (EOSD) and Semiconductor International (SI), as well as the NEPCON tradeshows. SI was spun out of EP&P in 1979, drawn to the same market covered by another successful magazine, Solid State Technology (SST).

Two weeks later, I was attending my first Semicon West writing stories for our show daily. I have attended every Semicon West since, plus many other Semicons in Europe, Japan, Korea, Taiwan and China, plus Austin and Boston when there was a Semicon Southwest and Semicon East. I tried to count how many days I got up, put on a suit and badge (a show badge) and reported to work on a trade show floor. Easily a year’s worth. But it was that first trip to California, dining at Ondine's in Sausalito, looking out over San Francisco bay, when I thought, “hey, I really like this job!” And, of course, covering the most dynamic industry in the world, during the information age, has been a lot of fun too.

My first feature article was about photomask inspection and repair. Several companies, such as Florod, offered laser-based tools that could trip off a bit of chrome or add a bit of chrome to repair a pinhole defect. Today, eliminating defects on masks continues to be a challenge.

What I remember most from the early days was trying to piece together the overall process flow of an integrated circuit. Today, the 300 or so process steps -- cleaning, etching, deposition, doping, etc. – are pretty well known, but that was not the case in the ‘80s when companies such as Texas Instruments and IBM jealously guarded their information. One Eureka! moment was when I saw the process flow of a 16K DRAM. Another was when I read my first set of papers from the International Electron Devices Meeting (IEDM). The conference organizers were kind enough to send abstracts of all the articles well in advance of the actual conference – it was always one happy day when that foot-high stack of paper arrived (now they send a link but it’s still a thrill). One of those papers was by Lou Parillo of Motorola (he’s now with Unity Semiconductor). I asked Lou to write a paper for SI on CMOS Trends, which he did in 1988. I think it was a bit of a challenge for him, probably due to time constraints – I clearly recall him saying I had him “screwed to the wall” on it – but he came through with an excellent piece. I also remember when the first International Technology Roadmap for Semiconductors (ITRS) came out, although I don't remember the exact year (1994?). That pretty much filled in any missing blanks when it came to process flow.

I did have the opportunity to interview Bob Noyce before his untimely passing in 1990. He always took time to talk to the press, which was nice. He likened semiconductor manufacturing to a swamp full of alligators. “You have to drain the swamp, and when you see an alligator, you shoot it,” he said (or something like that). That's hilarious.. and I say that with 4 syllables like my 6 year old nephew does, hill-air-e-us.

Another article I wrote was on intrinsic gettering. Oxygen in silicon causes dislocations, and those dislocations trap or “getter” defects and impurities that tend to move around in the silicon (the oxygen is introduced into the silicon by the melting of the interior of the quartz crucible in which the silicon ingot is grown or pulled). At the time, I had no idea that there was so much action going on in solid materials. Or that they were so porous. In the case of ultra-high vacuum, small atoms such as hydrogen will move right through stainless steel, making 10-11 Torr, "ultra-high" vacuum difficult to achieve.

Some of the biggest trends and advances I covered over the years: the move from aluminum to copper, the switch from silicon dioxide to low-k dielectrics, the introduction of chemical-mechanical planarization (CMP), high-k dielectrics, strained silicon and, of course, continued scaling (we did a special supplement, Breaking the One Micron Barrier, in the late '80s). And the transitions to larger wafer sizes. I don’t remember the transition to 200mm being a big deal, probably because IBM financed most of it (Paul Castrucci specifically) but I know the 300mm transition was pretty rough on many suppliers. Some are no longer around today as a result, I believe. It was really CMP, though, that was shocking. To take a wafer that had been handled so delicately, when it had acquired so much value, to turn it upside down and grind it down with a slurry... that was unthinkable, at least to me. But IBM proved it worked and the rest is history.

Other cool moments: walking the halls of IBM’s Almaden research center, where I met with Grant Willson. My degree is in EE, so organic chemistry hurts my brain a bit, but Willson made it hurt less. I wrote an 11 page article on photoresist around that time, and had to figure out what diazonaphthoquinone novolac resist was all about, and Grant was very helpful. I'll never forget my first visit to Bell Labs in Murray Hill, where the transistor was invented. I stood in the parking lot for a bit, staring at that awesome building, totally awestruck! I later, just a few years ago actually, had dinner with Steve Hillenius (now with the SRC), who worked at Bell Labs, and he said his office was the same room where Shockley, Brattain and Bardeen worked their magic. The reverence in his voice was clear. A good read on the invention of the transistor is Shockley's "The Path to the Conception of the Junction Transistor". Shockley turned into a nut later in life, but it's still a good read (jeers to IEEE; this used to be free but now they want $10).

Another Eureka! moment came when I was reporting on some kind of technology award. I believe it was AT&T. I called someone up and my questioning was along the lines of what research led to this, what did you discover? His answer, which I'll never forget, was simple: It was really the people. It wasn't the company, it wasn't the technology, it was just a bunch of smart and dedicated people. Everything this industry has accomplished is truly amazing, but it's really the people behind it that deserve the credit.

If you’ve made it this far into this snoozefest, you may be wondering what happened to Semiconductor International. Well, Cahners was acquired at some point by Reed, which at some point merged with Elsevier to form Reed Elsevier (RE). If you think the combination of British managers and Dutch accountants is a good idea, then you’ll like RE. Anyway, RE decided to get out of the business-to-business (B2B) publishing market and, in early 2008, before the biggest downturn since the Great Depression, announced that it was selling the Reed Business Information (RBI) division, which was largely based in the U.S. 30 different magazines or “brands” were on the block. It was the worst timing ever, and by early 2009, the decision to divest was cancelled. But, the damage was done. After selling a few and closing a few magazines in 2009, in April 2010, RE closed 23 magazines, including SI. Seven of those were eventually sold, but the majority simply ceased to exist. I had moved to PennWell (a fantastic company) and Solid State Technology in April 2008, so thought it almost comical, except for the fact that many good friends lost their jobs on that day.

Looking back over the last 30 years, what has clearly changed the most is media. With the evolution of the internet, ironically made possible by semiconductor technology, people are “consuming content” in every way imaginable, including blogs, tweets, podcasts, videos, e-newsletters, e-mail. An editor today also has to be tuned into all the social media: Facebook, LinkedIn, Twitter, YouTube, GooglePlus, Pinterest, etc. Search engine optimization know-how is also a must (I'm not complainin' I'm just sayin'). Most people use social media and Google to stay up on the news and trends, but sometimes very technical information is best absorbed by hearing someone explain it. That’s probably the best part of being an editor for 30 years: Some wonderful and very smart people have taken the time to patiently explain things to me, and I have done my best to collect that and pass it on.

I can't imagine what the next 30 years will bring -- the Internet of Things and some major advances in personal medicine enabled by semiconductors, no doubt, which I will hopefully need later rather than sooner, but, as always, send me your ideas: peters@pennwell.com.