Science and engineering are the most fundamental inputs to innovation in a business ecosystem. University and private labs are rich sources of ideas. Bringing an idea forward requires a team of experts including those who understand the discovery and those who know the industrial situation. The typical time frame for moving a discovery out of the lab to market is a decade or more. Our business ecosystems must have at their core processes of science translation — Bell Lab’s “reduction to practice” — where people can work closely, in secret, at the highest professional levels, for a decade or more. This is in fact, what does happen in our best ecosystems, an incredible feat little appreciated beyond the inner walls.
SOI Consortium, Nanoelectronics Research Initiative,
College of Nanoscale Science and Engineering SUNY Albany,
UC Berkeley Electrical Engineering Department
If you want to improve your business, find relevant fields of science and engineering and get closer to the research programs and results — read journals, trade publications and textbooks. Look for ways to incorporate the findings in your business or, better yet, get acquainted with professional researchers in the field and explore ongoing collaboration.
This, in simple form, is a leadership lesson I found at nearly every turn as I interviewed executives across the connected community. It was fascinating to realize the extent that university research contributes to the “Moore’s law” process advances in fabs. The connected community and its many business ecosystems do not live in isolation. Both levels, overall community and purpose-built ecosystems tap deep, active roots in the scientific and engineering communities.
Companies across the connected community routinely access advanced science and engineering research. This may involve just a rapid dip into a topic by way of reading published papers. There are many scientific papers, for example, on power saving techniques, published by university researchers as well as companies that have a long history of working on semiconductors such as IBM and Texas Instruments.
There are also large groups and organizations dedicated to fusing the talents of companies, independent research organizations, and universities when it comes to the core science that makes industry progress possible. For deeper fusion there are multi-year programs with an ecosystem of partners. The SOI Industry Consortium has been running a program on power-saving for many years. Its ecosystem has university affiliates including University of California Berkeley(US), Stanford University(US), Centro Universitário da FEI (Brazil), Kanazawa Institute of Technology (Japan), Ritsumeikan (Japan), Universite Catholoque de Louvain (Belgium) These universities are matched with EDA companies Cadence, Mentor Graphics and Synopsys, core library and IP companies ARM, IBM and Synopsys; and foundries Freescale, GlobalFoundries, IBM and UMC.[i]
Another cooperative institution, the Semiconductor Research Corporation, bills itself as “Pioneers in Collaborative Research.“[ii] Over 130 universities are members, as well as an almost endless list of companies in the semiconductor industry and many members of the connected community. Its Nanoelectronics Research Initiative, started in 2005, has identified roughly 20 new technologies that may be able to supersede today’s state of art technologies.
In addition, academic institutions are starting nanoscience departments to pursue issues of interest to the connected community. For example, the new State University of New York College of Nanoscale Science and Engineering, Albany[iii] was started in 2004 and is part of multibillion dollar co-investment between New York State, IBM, GlobalFoundries and others. Its goal is to create an advanced research cluster, with students and employees numbering in the thousands, to study, develop, and, at fabs being built in the region, manufacture next generation devices.
Tapping into these academic departments, consortia and institutes is simple in theory. Many are open and transparent public resources. Most consortia actively seek members and are open to companies across the connected community and beyond. On the other hand, translating research to practice requires dedicated effort often over a decade or longer.
The phrase “reduction to practice” appears in US patent law and means “embodiment of the concept of an invention.”[iv] This term can take on a broader meaning, over the years in members of the Bell Labs staff in particular seem to have come to use it as a kind of mantra emphasizing the value and the relative difficulty of getting a good idea to work well. There is a kind of genius in reduction to practice, similar to the experimental researcher with a knack for coming up with an elegant experiment. Reduction to practice is taking a discovery — for example a small change in transistor geometry that appears to be able to save power and increase speed — and turning that discovery into something that works in the world — and not just in our minds.
FinFET is a technology that is currently contributing to progress in Moore’s law. It is much in the news and is celebrated as an industry advance.
Briefly FinFET is a design approach wherein the transistors on a chip, field effect transistors (“FET”), are made with a 3D structure that has what can be loosely called a “fin.” The finned transistors leak less electricity and run faster that flat “planar” transistors.
Now, to whom should we credit FinFET? We can thank academic research and the same US government agency that brought us the Internet: the Defense Advanced Research Projects Agency (DARPA).
“FinFETs have their technology roots in the 1990s, when DARPA looked to fund research into possible successors to the planar transistor. A UC Berkeley team led by Dr. Chenming Hu proposed a new structure for the transistor that would reduce leakage current.”[v]
In order to facilitate the reduction to practice of FinFET, Chenming Hu himself became Chief Technology Officer of the Taiwan Semiconductor Manufacturing Company from 2001 to 2004.[vi]
“Leading foundries estimate the additional processing cost of 3D devices to be 2% to 5% higher than that of the corresponding Planar wafer fabrication. FinFETs are estimated to be up to 37% faster while using less than half the dynamic power or cut static leakage current by as much as 90%.”[vii]
In a tech world obsessed with power saving, battery life, and the massive electrical demands of cloud computing, FinFET is a no brainer. FinFET is just one of many advances drawn from scientific and engineering work done in academic or independent research centers. For the connected community, these discoveries are accessible and can form seeds of new commercial technologies once they are reduced to practice.
I had a set of particularly enlightening conversations with Phil Dworsky and Rich Goldman at electronic design automation company Synopsys. In order to implement FinFET, Synopsys partnered with foundry TSMC, the Berkeley research group founded by Chenming Hu and colleagues, and ARM research and development professionals. Rich gave me a sense of what the reduction to practice required:
“There is no way forward without collaboration, at least with us in our industry. And not only is it collaboration between two companies, but it’s collaboration between three or more companies, and they all have to come together.
“For example, to implement FinFET at TSMC we [Synopsys] worked closely with ARM very early. We all have to work together very early on to prove that it can work. Then once you get to that point that you’re able to make something that works, you then have to work together to develop a [design tool methodology] that the end user to can use to do the same thing.”
In addition to making a process work, the success of an electronic design automation tool company — like Rich’s company Synopsys — has to be on two sides. It has to work in the fab, in light and silicon and electricity — and it has to work for design engineers who want to spec out a final chip, push a button, and have their chip in hand not long after.
As Rich says, you have to have design tools that make the end user, the design engineer, efficient and successful.
“If you can’t deliver effective tools then all that work has been for naught. To deliver that requires, again, an additional period of close collaboration with at least three parties, and probably even a fourth party, which is the lead customer.”[viii]
Phil Dworsky, Rich’s colleague at Synopsys, came in at this point to describe making the tools, which involves creating models of the light and physics of the specific processes being used, models that address a specific chip architecture and layout, and models to guide the machines that will do the manufacturing. Layout requires respecting design rules for managing issues such as heat, radio-frequency interference, or the “quantum-tunneling” problem that is a leading cause of power loss. Phil added,
“Absolutely all of these collaborations require multiple years. Just going from conventional design to a well-known fabrication process using proven design tools can take five years from idea to volume production.”
Now factor in new ideas for new processes having new design rules and requiring new types of design software.
“It requires a process of understanding and modeling, experimentation, multi-way tuning and optimization –– there is just an awful lot that goes into making that happen. Then finally we get it into the advanced designers’ hands – and they’re starting to do their pilot projects. Then we get into early production and then finally we get into mainstream. If we look at the lower power work we did –– together with ARM – it took about seven years.”[ix]
The major lesson of this chapter is to draw deeply from advances in science and engineering. The lesson of these cases for the connected community is that it is a long-term, patient and demanding process—but one that is essential to the advances of the community. Without that science and engineering input, the whole community stagnates.
Drawing deeply requires cooperative teams that can stay together for years, and do so with very open, cross-company sharing among themselves, while maintaining secrecy toward the outside world—sometimes among the companies of the members themselves.
There is an important additional leadership lesson as well: The human side of this process requires intellectual and emotional intelligence — what Phil called in this context “maturity.”
“The attributes [our team members need] are maturity and a sense of what’s needed, because it’s almost like living in another foreign language. You need to be able to translate it always in terms of someone else’s objectives and be able to demonstrate to one person what the other needs and, and why that makes sense for both to do that.”[x]
Rich and Phil spoke of the management systems required, starting with respect for all members of the collaborative team — scientists, engineers from the foundry, software tool and IP, and device side — combined with the need to maintain silence and secrecy about the content of the work. Often the mere fact that a particular company is considering a new fabrication is a competitive secret.
Choices frequently have consequences that affect members of the cross-company team differently. As Phil put it, the team’s job is to find solutions that both work and “create the most area under the curve” of positive results for all involved.
“… we basically have to strive to convince others, even internal to our company of the righteousness of a certain project. We need to create the business case that faces all directions.”
Phil makes much of the benefits of 360 degree business cases. Carefully written out cases help each representative articulate what is important to that person’s company, how it can be measured, and what the representative and company might trade off to achieve higher priorities. By making these cases available to all, other members become skilled over time at understanding each others’ positions when collectively evaluating a choice. As Rich stated the overall goal:
“Our job is to understand the others’ challenges and be able to translate between them, because we’re trying to pull together multiple parties within and outside of the company along a solutions path that requires each to give and take. It’s a negotiation on all sides.
“The other side of this is to realize just how pragmatic and practical the whole electronics industry is. We all take what we can get and then we have got to solve the problem somehow. Those who have been exposed to all sides have the understanding that things are sometimes imperfect and you have to adapt, move and change to facilitate more productivity. EDA is all about productivity, making sure that our customers can take what we give them and use it on this really difficult challenge to quickly and reliably get to a result.”[xi]
All across the connected community people are taking ideas — game changing ideas in many instances — and carrying on the down-at-the-workbench tasks of reducing them to practice. In this chapter I hope to have highlighted those who dip into advanced science and engineering and how much they mean to the connected community and the ecosystems within it. With the specialized and open-to-all business models of the connected community the advances made possible in reduction to practice become widely available. Thus the benefit of this work is leveraged many times over. Tapping into advances in science and engineering proliferates through and very much serves the connected community, the ecosystems, and the customers.
NOTES[i] SOI Industry Association, Ecosystem, http://www.soiconsortium.org/about-soi/ecosystem.php
[ii] Nanoelectronics Research Initiative, Semiconductor Research Corporation http://www.src.org/program/nri/
[iii] State University of New York at Albany, – College of Nanoscale Science and Engineering, http://cnse.albany.edu/PioneeringAcademics/Constellations/Nanoengineering.aspx
[iv] United States Patent and Trademark Office, Manual of Patent Examining Procedure, 2138.05 “Reduction to Practice”
[v] “FinFET: The Promises and the Challenges, Synopsys Insight Newsletter, Issue 3, 2012 http://www.synopsys.com/COMPANY/PUBLICATIONS/SYNOPSYSINSIGHT/Pages/Art2-finfet-challenges-ip-IssQ3-12.aspx
[vi] Chenming Hu, University of California at Berkeley, http://www.eecs.berkeley.edu/Faculty/Homepages/hu.html
[vii] “FinFET: The Promises and the Challenges, Synopsys Insight Newsletter, Issue 3, 2012 http://www.synopsys.com/COMPANY/PUBLICATIONS/SYNOPSYSINSIGHT/Pages/Art2-finfet-challenges-ip-IssQ3-12.aspx
[viii] Rich Goldman, Vice President, Corporate Marketing & Strategic Alliances, Synopsys, personal communication, April 2013
[ix] Phil Dworsky, Director, Strategic Alliances & Publisher of Synopsys Press, Synopsys, personal communication, March, April 2013
[xi] Rich Goldman, Vice President, Corporate Marketing & Strategic Alliances, Synopsys, personal communication, April 2013