Industry Insights: Cohesive education programs for IC manufacturing in need
12/01/1999
Stanley Wolf
At the height of the last boom in the semiconductor industry, we heard an industry-wide cry about the lack of properly educated and trained people entering the industry at all levels. Part of the concern was that US academia was not turning out enough engineers in general and, more specifically, with a few exceptions, was not offering programs with a strong emphasis in IC manufacturing.
Now, we've gone through a classic downturn and are entering what we anticipate is another boom. But it appears that IC manufacturing education at the university level in the US today has not changed appreciably. While engineers and scientists are well educated in specific disciplines, such as electrical engineering, chemistry, physics, materials science, mechanical engineering, and optics, they still enter the industry untrained for IC manufacturing. By-and-large, even at the engineering level, training in IC manufacturing is still occurring on the job. There has been no fundamental change in engineering education to support what is a cornerstone industry in today's society.
My best perspective of this trend comes from seeing the shortage of recent, up-to-date semiconductor textbooks.
Only ~2000 textbooks specifically about semiconductor manufacturing are delivered to university bookstores each year and ~15% are returned because they are not purchased. Yearly enrollment in such classes is about 1500. IC fabrication courses are offered at about 100 US universities, with an average enrollment that sells 15
Four of the five top IC manufacturing textbooks, which constitute most of the 2000 mentioned above, have copyright dates circa 1990. This indicates that book publishers do not see a demand to keep these books up-to-date with the rapid pace of IC manufacturing technology. The current edition of Sze's VLSI Technology (McGraw Hill), which originated as notes for an internal training program at Bell Labs, was published in 1988. Sze, who is now at National Chiao Tung University, Hsinchu, Taiwan, republished his book in 1996 as ULSI Technology (McGraw Hill), but this volume eliminated chapters on oxidation, diffusion, and ion implantation, reducing its effectiveness as a textbook.
Other older textbooks include Runyan and Bean's Semiconductor Integrated Circuit Processing Technology (Addison-Wesley), published in 1990, and Gandhi's VLSI Fabrication Principles (Wiley), which was re-issued as a second edition in 1994, although 90% of its references still pre-date 1985. Campbell's Science and Engineering of Microelectronic Fabrication (Oxford) was published in 1996 and currently owns ~35% of the textbook market. However, it contains almost no references more recent than 1991 and therefore does not describe the hottest process technologies of the 1990s, such as CMP, copper metallization, dual-damascene interconnects, step-and-scan alignment, and phase-shift masks.
Despite the fact that ~90% of IC manufacturing is CMOS-based, there is no single textbook in print today on CMOS technology. Chen's CMOS Devices and Technology for VLSI (Prentice-Hall, 1990) is out of print. This may be because only a very small number of classes are available at universities on the device physics of MOSFETs
There is a shortage of engineering graduates, especially those from the pioneering and renowned IC manufacturing program at the Rochester Institute of Technology (RIT). Numbering about 35
Certainly, we need to recognize that implementing an IC manufacturing engineering program at a university is no small task. Establishing an IC fabrication laboratory is very expensive and such a facility requires significant resources to maintain. It is further complicated because IC manufacturing is so interdisciplinary. Existing single classes tend to fall under EE departments. However, some would argue that they should be taught under chemical engineering or materials science, as there is really less of electrical engineering in wafer fab technology than of other technical disciplines (except in metrology and test). But then there is lithography that comes from optics and polymer chemistry; process control that comes from statistics and mathematics; robotics from mechanical engineering; software programming from computer science; and vacuum technology and ion implantation from physics. Each of these disciplines is from a particular academic turf and has professors who traditionally don't like to cross boundaries.
As a result, students who are in a single department do receive a very in-depth education in their specialty, but get very little exposure to other areas. This may work for the automotive, steel, and oil and gas industries, but these don't require as broad a scientific background as does IC manufacturing.
While there are a few schools that are attempting to bring instructors from various disciplines together to teach microelectronics manufacturing courses
Engineering enrollment in the US hit its trough in the mid-1990s. The US economy is strong and the labor pool is tight. Companies from all technology sectors are actively recruiting the relatively small graduating classes of new engineers. The IC industry had some breathing room during this last downturn to strengthen the course infrastructure on IC-manufacturing in academia. The fact that this opportunity was not exploited may come back to haunt the industry in the near future.
 |
Stanley Wolf is the author of the three-part series Silicon Processing for the VLSI Era and owner of Lattice Press, PO Box 352, Sunset Beach, CA 90742, ph 714/840-5010, fax 562/592-1976, e-mail [email protected].