The philosopher Ludwig Wittgenstein wrote, “Knowledge is in the end based on acknowledgement.” The poet T. S. Eliot argued that the mature creative mind takes ideas from others and transforms them into something different, something better. These thoughts, posed by Wittgenstein and Eliot nearly a century ago, point to a long standing tradition in which artists, innovators, and academics alike have drawn material from the world around them—from the work of their contemporaries, from research being done in other disciplines, from music, paintings, articles in popular magazines, and applied that material to their own work in new, inspired ways.
Here at Portland State University this process is taking place in the Integrated Circuits Design and Test Laboratory (ICDT Lab) of Dr. Robert Daasch, Professor of Electrical and Computer Engineering, Maseeh College of Engineering and Computer Science. Dr. Daasch, along with his colleague and collaborator, Dr. Glenn Shirley, Adjunct Associate Professor of Electrical and Computer Engineering, has developed a new system and method to support decision-making in the manufacturing and testing of semiconductor components. The impetus of Doctors Daasch’s and Shirley’s new copula-based system and method for testing semiconductors came from an unlikely source: the worlds of finance and actuarial science.
So what exactly does a science that uses mathematical and statistical tools to predict risk in the finance and insurance industries have to do with manufacturing semiconductors, you may be wondering? The short answer is the copula-based system can help semiconductor manufactures design and plan for future electronic systems and integrated circuits, and optimize manufacturing tests of integrated circuits currently in production.
The idea came to Dr. Shirley when he and Dr. Daasch were attempting to analyze a beautiful set of data acquired from memory integrated circuits on ICDT Lab’s test equipment in an experiment designed by Dr. Daasch. One of Dr. Daasch’s students, Satoshi Suzuki, acquired the data. Dr. Shirley recognized the problem as an example of a larger class of problems involving modeling massive datasets for decision-making that he had been working on and managing during his 30-year career in the semiconductor industry. Doctors Shirley and Daasch knew how the copula function was used in other fields and they began to wonder if a copula method might be applied to this class of semiconductor test problems.
“We knew the copula was out there and that its structure had been used in finance and actuarial science,” Dr. Daasch said, “but it was untapped in relation to the semiconductor manufacturing and testing industry.”
“It has a broad applicability nobody in our field has touched,” Dr. Shirley added.
Modern semiconductor manufacturing tests measure and record many attributes of each integrated circuit produced. These attributes are correlated in various degrees. Copula-based models fitted to special test chip data can capture the full generality of any kind of correlation. The models can then be used to compute figures of merit targeted to key stakeholders, such as the producer (yield), and the end user (fraction of the end-use population which is defective) for products different in size, fault, and tolerance from the test chip.
The model will find applications at all stages of the semiconductor product development lifecycle; from initial evaluation of hypothetical products long before they are produced all the way to optimizing test settings in manufacturing of a mature product. Benefits of the improved decision-making enabled by the copula-based test method developed by Doctors Daasch and Shirley include cheaper, higher quality, and higher-performance integrated circuits.
To spearhead the introduction of their novel, new technology to semiconductor testing and manufacturing companies, Doctors Shirley and Daasch are publishing the results of their research and will be reaching out to some of the most well-known names in the industry. They have also partnered with the Office of Innovation & Intellectual Property at Portland State University to file for a patent and to promote the use and increase the impact of their new technology.
While the concept of the copula has been used in finance and actuarial science to model the behavior of two or more coupled attributes in a dataset, Doctors Daasch and Shirley have approached this concept with their own brand of creativity and forward thinking, a creativity honed by their expertise in semiconductor manufacturing and computer engineering and driven by the collaborative nature of their unique partnership.
It’s the kind of innovation Dr. Daasch likes to call “P.F.E.”
“It’s an acronym for, ‘proudly found elsewhere,’ which is the opposite of N.I.H., or ‘not invented here,’” Dr. Daasch said, cracking a broad smile. It’s also the kind of innovation the office of Innovation & Intellectual Property is excited to introduce you to.