This web page is an update of a laboratory development manual for university faculty. The manual, published by CEMD in 1994, is the "NSF Microfab Manual". Update 2007.
The subject is Microfabrication Technology, with emphasis on Integrated Circuit Manufacturing. The purpose is to assist in implementation of microfabrication experiments for undergraduates. The theme is that microfabrication experiments can be simple and inexpensive.
The first section is a copy of a published article that describes the Workshop project associated with this manual.
The section Workshop Discussion Items is the meat of the manual. Here you find the results of the workshops.
The two sections Books & Literature and Commercial Vendors & Dealers may provide useful references.
This article was published in IEEE Transaction on Education, Vol., 39, No. 2, pp 211-216, May 1996
Significant Update comments were last added to this web copy in July 1998. Minor update 2006 & 2007.
Abstract - This is a report on the NSF (National Science Foundation) Microfabrication Laboratory Workshops. This series of workshops was sponsored by the NSF, and held at San Jose State University. The theme of the workshops is that microfabrication experiments can be included in undergraduate laboratory curriculum even with a very limited budget. Several significant experiments were developed during the workshops. The most popular is a simple photolithography "artwork" exercise that demonstrates the techniques for fabrication of integrated circuits; the investment for this exercise is about the same as for a photography dark room. 165 undergraduate faculty attended the workshops. 142, or 86%, have implemented workshop curriculum improvements at their home universities. An estimated 10,000 undergraduate students have performed workshop related experiments. A surprise finding of the workshops was the large number of existing underutilized IC Labs at universities.
Microfabrication (also microfab, fab) here refers to the technology of making very small devices using photolithography along with other techniques. Microfab is used extensively in the manufacture of silicon wafers (or slices) for integrated circuits (IC) (also chips, die). Microfab is also used to manufacture discrete silicon devices, non-silicon electronic devices, displays, sensors, disc read-write heads, and other microelectronic devices. Microfab is used to manufacture non electronic devices such as micrometer scale sensors, actuators, electrostatic motors and mechanical devices. The word "microfabrication" has other meanings not associated with photolithography; these other meanings, and the technologies associated with the word "nanofabrication" are not considered here. A list of references on microfab, photolithography, and associated techniques is included in the "NSF Microfab Manual" [1], which was produced by the workshops described here.
The "NSF Microfab Manual has not been updated since 1994. This web copy serves as the update.
The focus is on simple, inexpensive undergraduate microfab laboratory experiments. There is a widely held notion that microfab experiments are necessarily expensive. One purpose of the workshops was to dispel that notion. In this article, that notion is briefly documented and dispelled..
The workshops and the associated project are funded by the National Science Foundation (NSF) through the Undergraduate Faculty Enhancement Program, Directorate for Education and Human Resources. The target audience is faculty with undergraduate laboratory responsibility in United States (US) educational institutions.
The workshops are held at the Center for Electronic Materials and Devices (the Center) in the College of Engineering at San Jose State University (SJSU). SJSU, a campus of the California State University system, is the largest supplier of engineers to the local "Silicon Valley" electronics industry. The Integrated Circuits Laboratories (IC Labs) at SJSU were used for the workshops.
Dr. Peter Gwozdz, the author of this article, the Director of the workshops and associated project, and maintainer of this web copy, worked sixteen years in the IC manufacturing industry before coming to SJSU as a professor in the General Engineering Department. Gwozdz retired from SJSU in 2003 but checked this web page in 2006.
Again, the primary subject is inexpensive experiments. A secondary subject of the workshops is inexpensive microfab techniques for research [1]. This subject, although quite popular at the workshops, is not covered in this article.
The Opportunistic approach [2] to educational laboratory management, which was named and developed as part of the workshop project, is described and discussed.
The workshops unearthed facts and insights regarding existing university IC Labs. Although existing IC Labs were not planned as a subject for the project, the related data is reported here.
The traditional use of IC Labs, for small classes of students fabricating IC devices, is not discussed here. But the subject is important, and came up often at the workshops. This report is restricted to description of simple, inexpensive experiments for large numbers of students. The experiments may be implemented in an existing IC Lab or in a typical university laboratory room.
A final objective of this article is to announce the availability of wafers. Silicon "test" wafers are supplied as donations by industry with restriction to use in undergraduate education. See Obtaining Silicon Wafers.
Fuller [3] reviewed the subject of microelectronics manufacturing engineering education. The result of a 1986 Florida workshop to define curriculum for microelectronic manufacturing education, sponsored by the Semiconductor Research Corporation (SRC), was published by Kerns [4]. For samples of specific programs, see Fuller [3], [5] for the program at Rochester Institute of Technology, and see Serra [6] for the program at the University of Victoria. This literature focuses on the education of elite microelectronics manufacturing professionals. Laboratory work is briefly covered. The number of students involved is estimated [3] at only 600 annually for the entire US. The cost per student is not discussed.
For larger numbers of students, with lower budgets, some example literature is available on comparable programs. Compton and York [7] used laser printed masks to fabricate printed circuit boards for microwave applications, in a laboratory exercise at Cornell similar to the photoresist exercise described here at SJSU. Haskard [8] briefly mentions simple Schottky diodes and solar cells that are fabricated as part of a larger laboratory curriculum at the University of South Australia.
An explicit count of IC Labs at universities in the US was not found in the literature. Fuller [3] estimates there are thirty. Dataquest Inc provided a list of thirty-four [9]. In the 1990 SRC Source Book [10] thirty-one university IC Labs can be found mentioned. This subject will be discussed below.
Our theme is that microfab experiments can be simple and inexpensive. Many of the participants come to the workshops with the idea that all microfab work is difficult and expensive. After performing a few experiments, all participants agree that meaningful and instructive student experiments can be performed simply and inexpensively. At the workshop discussion sessions, not one participant disagreed with the theme.
It is difficult and expensive to make 64 M bit DRAMs for a cost of a few dollars each. Modern microfab factories employ Ultra Large Scale Integration (ULSI) which is a difficult, expensive extension of the basic Planar Technology. The basic Planar Technology (photolithography plus oxidation of silicon plus diffusion in silicon plus aluminum metallization), is one simple example of microfab.
The scientific literature has never claimed that microfab is difficult. Submicron transistors were regularly produced in the 1960's in university laboratories and for manufacturing of discrete transistors. Only with the large image field of view required by ULSI is submicron resolution difficult and expensive.
The notion concerning the difficulty of Planar Technology is widespread. Most review articles on microfab emphasize the amazing achievements of ULSI without a reminder that ULSI evolved from a simple ancestor. Many workshop participants described faculty meetings where budget constraints were invoked to rule out microfab experiments. One objective of the NSF Workshops is to set aside this mistaken notion.
Each workshop session was five days, Monday through Friday. Sessions were held in June and in January. The home institutions of the participants covered travel expenses; all other expenses, including hotel and meals, were funded through the NSF grant. Eleven workshop sessions were held, June 1991 through January 1995. A typical session had fifteen participants.
Participants registered by writing to the Center. All US undergraduate faculty with laboratory responsibility who applied were accepted. Since the subject is highly interdisciplinary, faculty from many departments attended. Most participants were from Electrical Engineering, Materials Engineering, and Chemical Engineering. Because of recent interest in micromechanical fab, many participants came from Mechanical Engineering. A significant fraction also came from Materials Science, Physics, and Chemistry. One hundred sixty-five faculty attended.
Workshop time was about equally divided three ways, between laboratory, lecture and discussion.
Laboratory time, in the SJSU IC Labs, was for performing selected student experiments. Some faculty spent some of this time actually developing experiments. More on these experiments below.
Lecture time was in classrooms. The lecture format was varied for the first few workshops. It turned out that a general elementary review of microfab technology was almost universally favored by the participants; this was the bulk of the lecture curriculum. A few advanced IC topic lectures were offered. The laboratory and lecture content were very similar to a three day short course offered by the Center for industry professionals.
Discussion time was for design of student experiments, and for making these experiments portable. Most of the details of experiments and many of the ideas for experiments were created during the workshop discussion sessions. A few experiments are described below. Safety and environmental issues were also covered. Discussion sessions were held in meeting rooms, except one discussion per workshop was held in a redwood forest.
The workshops were evaluated using questionnaires. One, at the end of each workshop session, was for immediate evaluation of the session. Two others, by mail, counted the actual implementations of experiments, and measured other objective benefits of the workshops. Results are given below.
The word "experiment" is used here as a general laboratory term to include exercises and demonstrations. The most successful experiment by far is the photoresist imaging exercise. We call this our Paradigm experiment. Standard transparencies are used as masks, with manual exposure using a commercial mercury lamp. Transparencies are prepared with a laser printer or on a standard office copy machine. In this way, the paradigm avoids the expense for masks. The training required for an alignment machine is also avoided. (At SJSU, students use slide film and manual alignment to make transistors and IC devices at millimeter design rules. Although a sub culture of professors has spun off from the workshops to further develop this technique, it is beyond the scope of this article.) Artwork and text are used for the subject matter; many students bring their own transparencies.
The only significant piece of equipment for the Paradigm is the spinner. Commercial laboratory photoresist spinners are available for less than $2,000, but about $5,000 is recommended. A student project to build a spinner for a few hundred dollars was carried out as part of the workshop preparations; the resulting spinner is demonstrated at the workshops; documentation is available on request. Alternately, the spinner is an excellent candidate for use of a donated manufacturing type piece of equipment with automatic wafer handling; commercial factory spinners are usually easy to maintain.
For chemical containment and ventilation, standard chemistry style hoods are adequate. One participant designed a station using a hardware store kitchen exhaust.
Yellow lighting is commercially available. Red photographic lighting, if preexisting, works. In summary, the equipment for the Paradigm is equivalent to the equipment for a hobby photography dark room; the design can be expanded or scaled back to fit a reasonable university budget.
Several variations on the Paradigm are available, depending upon type of wafer, type of etch, and strip.
The simplest variation is the Silicon Souvenir, which uses unpolished silicon wafers, no etching, and no photoresist strip. Only spin, bake, expose, and develop are included. More than ten thousand donated unpolished silicon wafers are available at the Center. At SJSU, about six hundred students per year do this, including visiting high school students and all lower level engineering students (as part of a required Materials Engineering Laboratory course). The author of this article, who is an obsessed advocate of the Paradigm, at first doubted the academic value of such a simple exercise. But it is gratifying to overhear an engineering undergraduate exclaim "oh, now I get it" as he contemplates his finished print. This is a prime example of the power of "hands-on" work as a complement to classroom instruction.
An argument can be made that plastic plates instead of wafers and painted photoresist would work for the Paradigm. It is the consensus of workshop participants however that this would push the idea of simplification too far, and that the Paradigm as taught is the minimum meaningful microfab experiment.
The favorite variation on the Paradigm is the MOS Souvenir, which is documented [1] with tips for laboratory instructors. All workshop participants (and all Center industry short course participants) print artwork on an aluminum film over silicon dioxide. The experiment includes aluminum etch and photoresist strip. The entire exercise is accomplished in the first hour or two of the course, and provides dramatic evidence of the simplicity of the Paradigm. Later in the workshop or course, participants thermally oxidize silicon wafers and perform aluminum evaporation, thereby preparing wafers for the next session.
Another Paradigm variation includes HF etch of oxide. Additional safety instructions are required for the use of HF [1].
The safety instructions for most variations can be accomplished in a few minutes. Safety glasses and gloves are used to protect against broken silicon chips and splashes of hydroxide developer (and phosphoric acid for aluminum etch).
Special supplementary implementation funds were made available by NSF. In the spring of 1992, the Center organized a project for implementation of the Paradigm by seven prior workshop participants. Each of the seven wrote a proposal. The group of proposals were accepted. NSF provided 50% of the cost of spinners, at $5,000 each, for each of the seven home institutions. NSF also provided $2,000 each for initial laboratory supplies. The Center packaged the combined purchases of spinners and supplies. The Center also arranged donations of wafers and photoresist by industry. Shipments were complete in the summer of 1993; students are already using five of the seven at this writing, winter 1994. This special supplementary project provided experience on implementation of the Paradigm as a "canned" experiment; institutions interested in an identical implementation can contact the Center for details on how to do it for about $8,000.
The Paradigm has been offered here in detail as an example of a simple experiment, explained in a fashion useable by most laboratory faculty. The following experiments are more briefly described; they may require further reference [1].
Oxidation can be performed in any laboratory furnace. Air ambient is recommended for simple inexpensive undergraduate experiments. Workshop participants report success using metallography furnaces and ceramic kilns. James Masi of Western New England College reported satisfactory oxidation using a dental table top crown annealing furnace; the results were repeated at SJSU; these dental furnaces can be purchased for about $500 [1]. Temperature uniformity is excellent for one to ten wafers in just about any high temperature furnace, because of the high radiant heat flow. Several technically meaningful experiments have been designed and implemented to study oxidation kinetics; the subject of silicon oxidation is very amenable to experimental design. Oxidation rate is dominated by air humidity; this should be treated as an experimental parameter, not as an inadvertent problem. Controlled gas mixtures of oxygen, nitrogen, and water vapor can be added at reasonable extra cost.
Aluminum Evaporation is simple and inexpensive in any laboratory vacuum system that has an electrical feedthrough for filament heating and which has room to set wafers near a filament. Any reasonably pure wires of tungsten for the filament and aluminum for the source give satisfactory results.
IC chip Opening is a useful demonstration that can be done in any classroom with a microscope. Ceramic packaged chips, which are easily identified by their layered structure, are easily opened with a pair of pliers.
Fringe Counting is a particularly popular experiment with workshop participants because of the powerful visual demonstration of Bragg's law. The simplest implementation is to expose a photoresist covered wafer while slowly sliding aside a piece of paper that covers the photoresist. After development, a wedge or terraced photoresist thickness gradient is observed; the "dead rainbow" of interference colors is observed in white room light by students. Any optical filter can be held up to the eye to count interference fringes and quickly use Bragg's law to calculate thickness. Another implementation uses an HF dip etch to produce a similar oxide thickness gradient across an oxidized silicon wafer. At the SJSU IC Labs, a box of oxidized wafers of several different thicknesses (and thereby different colors) is always kept on hand for demonstration.
Spin on Dopant is recommended for undergraduate diffusion doping experiments and for fabrication of pn junctions.
These and many other experiments are briefly described; a few have detailed documentation [1]. Documentation has also been performed on an ad hoc basis by prior workshop participants; copies are available from the Center on request. There are three kinds of documentation: student Handouts, implementation Details, and instructor Tips.
Student Handouts are available, but most faculty prefer to write their own. Implementation Details are the documents most often recommended during workshop discussion. Details of a specific implementation include a list of equipment and supplies along with vendor address and phone number. Interestingly enough, Details are rarely requested when available. It seems each faculty implementer designs a variation of an experiment to fit what she has available or to mesh with his way of doing things. On the other hand, the extensive general list of vendors [1] has been appreciated and used.
Instructor Tips are the most successful style of documentation for undergraduate microfab experiments. Tips include alternative ways of performing various steps in an experiment. Tips provide a trouble shooting list for when things go wrong. Future issues of the NSF Microfab Manual [1] will concentrate on Tips.
The Opportunistic approach [2] to educational laboratory management is used by the Center and is taught in the workshops. The Opportunistic approach differs from the traditional Requirements approach. In the Opportunistic approach, laboratory experiments are defined in terms of existing equipment and in response to donations. In the Requirements approach, experiments are defined in terms of the academic curriculum; equipment and supplies are purchased to satisfy requirements. The Opportunistic approach is offered in this article as a model to be used by other university educational laboratories.
The Center is particularly qualified to teach the Opportunistic approach. All Center activities are supported by donations, grants and short courses. There is no SJSU budget for the Center. In fact, even the six room IC Labs and facilities were donated through funds from local industry during construction of the new engineering building in 1988. SJSU has an advantage; the Silicon Valley location makes Opportunistic activities particularly effective. Few universities can reasonably expect to fund an entire IC Lab operation through donations and grants. The style of Center operations was explained at the workshops to inspire discussion and elaboration of general techniques for Opportunistic management of fab experiments. The detailed recommended techniques are documented in the NSF Manual [1].
The Opportunistic style is not new. A professor who takes her students on an informal field trip is using the Opportunistic method. Most teachers are eager to use recently donated equipment. It may seem pedantic to formally assign the name "Opportunistic" to something that faculty have always done, and to teach it as a technique. However, the first four workshops had extreme difficulty in discussions on equipment procurement for microfab experiments. Many faculty participants expressed frustration describing department meetings where they became side tracked on issues of curriculum requirements and equipment costs. There was no resolution of these issues in the first four workshops. The author of this paper personally experienced frustrations failing to communicate his vision of how to inexpensively set up microfab experiments. After the name "Opportunistic" was introduced [2] at the fifth workshop, the situation dramatically changed. Discussions became focused and fruitful. Participants vocally expressed the conviction that they could deal with objections in future department meetings. Perhaps the rhetorical trick of a "new technique" is required to overcome the aura of difficulty associated with microfab.
The Opportunistic approach is a complement, not a replacement, for the Requirements approach. Most university student experiments are designed to teach techniques and concepts required by the curriculum. Those few universities fortunate enough to have a budget for an IC Lab should go ahead and buy equipment as required to teach microfab according to a well thought out educational plan. But for most undergraduate laboratory faculty attempting to teach microfab, the Opportunistic style is the only way to go.
This author is an advocate of the Opportunistic approach. The objections to the Opportunistic approach will be mentioned, but with rebuttal: The Opportunistic approach sounds like "working without a plan"; a vision in some cases is better than a detailed plan. The Opportunistic approach means that sometimes an experiment needs to be canceled because donated materials are no longer available; yes, backup Opportunistic experiments are always available, and second choice is better than nothing. The Opportunistic approach may cause faculty to become sidetracked from important experiments that are academically required; do not let that happen to you. The Opportunistic approach is inconsistent and non uniform, providing different experiments to students during different semesters; there is nothing wrong with that.
The NSF Microfabrication Workshops were designed for faculty with responsibility for typical undergraduate laboratories. We have been surprised and flattered that 10% of participants have been microfab experts with IC Labs responsibility. They profess to have learned many valuable tricks for simple, inexpensive experiments. They have contributed impromptu lectures and lab demonstrations. Their presence and testament add credibility for the other participants. They tend to be most active in follow up documentation and development. So the original objectives of the workshops have been expanded, to include the development of simple inexpensive experiments that use existing IC Lab equipment for large numbers of students.
We were also surprised at the number of existing IC Labs at universities, and at the low level of usage. Our observations and conclusions are documented here:
The traditional use of IC Labs for undergraduates is to support a course, usually in a department of Electrical Engineering, wherein students fabricate complete IC devices, or at least a set of test transistors. The surprise is the small number of students. The typical number is about ten per year. Twenty students per year is unusual. Many university IC Labs are not used at all. It is difficult to communicate our confidence in the accuracy of these findings. A statistical analysis is not available. Most data was orally disclosed by participants only after a few days of informal discussions at meal time. A list of IC Labs with number of students would be an inappropriate breach of confidence. A literature search turned up no contrary claims. If the published estimate of annual microfab student production, 600 [3] is divided by the published estimate of the number if IC Labs, 30 [3], the result is 20. Not all the 600 students actually take an IC Lab course; we estimate only half do.
This disclosure is not intended to disparage the traditional use of IC Labs. The intensive training of a select few is a benefit to society at large. The author of this paper is one of the beneficiaries of such intensive university laboratory education. The small number of students per IC Lab course was a surprise to most workshop participants during discussions, so the conclusion was summarized here.
Another surprise is the large number of IC Labs at US universities. The literature search [3], [9], [10], mentioned above, puts the number at about thirty. We estimate the correct number is about one hundred. Again, the estimate is based on informal oral disclosures at the workshops. Many oral disclosures concerned IC Labs that were not operating at all. In some cases, the faculty member who designed them had left the institution. In many cases, money was available for equipment but not for operations. It is understandable that such facilities are not documented in the literature.
It is difficult to define an IC Lab at a university. In the estimate of one hundred, an IC Lab is defined as a contiguous set of rooms with at least some air filtration, which were designed with the intent of fabricating modern IC devices for research and/or education, and which have essentially all the required equipment in place, even if not hooked up. This definition is consistent with Fuller's [3] "Basic Educational Facility" (Fuller goes on to two more restrictive definitions). If the definition were extended to any university building containing a spinner, a good camera, a furnace, and an evaporator (sufficient to fabricate 100 micron pMOS IC devices) the number would be closer to one thousand.
These disclosures are not intended to disparage university administrators. Underutilized laboratories are a resource, not a problem. The original objectives of the workshops have been expanded, to include the inexpensive use of existing underutilized IC type equipment.
Implementation was measured via annual questionnaire, in April 1993 and again in April 1994. This section VIII was written November 1994, for publication, based on 135 participants to date. Only this one paragraph is updated, last time July 1998. The estimated number of participants, 165, turned out correct. The number of implementers, 142, is still a good estimate. Some who planned to implement did not, but the Center continues to receive correspondence from individuals who did not attend the workshops, but implemented on the basis of the NSF Microfab Manual and this web site. The number of students who performed workshop experiments was doubled from 5,000 below to 10,000 in the abstract, as an estimate, to adjust for an additional 3 years time.
88 of the 135 participants responded to the April 1993 questionnaire.
57 of the 135 participants responded to the April 1994 questionnaire.
Participants who do not respond to questionnaires are phoned, but not every year. April, 1994 data is available for all 135 participants, but in a few cases the data is three years old. Phone call updates were obtained in April 1994 for 21 implementers, including 14 of the top 20.
In the questionnaires, faculty are asked to estimate the number of students who have performed NSF Workshop related experiments. In a separate question, faculty are asked to estimate the number of students who attended lecture courses wherein the curriculum had been improved based upon the NSF Workshops.
38 of the 135 participants implemented laboratory experiments. An additional 64 participants report plans to implement NSF Workshop laboratory experiments.
96 participants implemented lecture improvements.
116 participants, or 86%, report either lecture or lab implementations or lab implementation plans.
80 individuals who did not register for a workshop requested a copy of the NSF Microfab Manual [1]. These also receive the NSF Microfab Newsletter. 58 of them also responded to the 1993 and/or 1994 questionnaires. 8 of them implemented on the basis of the manual.
3604 is the total reported count of students who participated in workshop related experiments.
7168 is the total reported count of students whose lectures were improved by workshop material.
2621 of the laboratory students performed the Paradigm experiment.
2372 of the laboratory students performed their experiment at SJSU. 1800 of these are SJSU engineering students who did the Paradigm experiment as part of Materials Engineering 25, a required course for all engineering majors. Perhaps many of these students would have used the IC Labs at SJSU even if the NSF Workshop Project had not existed, but because of the synergy between Center activities and the NSF Project, most would not have.
167 Paradigm experiment students came to SJSU from local junior colleges because their lab professor attended the NSF Workshop and elected to implement his/her laboratory activities as a field trip to the SJSU IC Labs.
66% of the student laboratory experiences to date have been at the SJSU IC Labs. This percentage will decrease significantly in the next few years, because implementation at SJSU was accelerated by work that occurred before the NSF project. Most non-SJSU student experiments occur more than two years after faculty workshop participation. Furthermore, the student usage rate seems to accelerate for at least three years after workshop participation. For example, Thomas Bingham of St. Louis Community College, who attended a workshop in 1991, has been the most active follow up participant in terms of documenting experiments and innovating techniques. However, his count of students is zero in the numbers, because he has just this semester finished preparations for including his work in courses.
The true numbers are greater than the reported totals for two reasons. First, much of the data is old, as mentioned above. Second, many questionnaires under report student numbers. Phone conversations with participants who have reported data usually cause the reported number to increase.
The paragraphs above are based on April, 1994 data. The paragraphs below were written in November, 1994, at the time of acceptance for publication.
14 faculty attended the June, 1994 Workshop. 17 are registered for the January, 1995 session. Assuming 1 no-show in January, the final count of faculty participants will be 165.
Applying the above 86% implementation rate to the total, we calculate that 142 faculty will have implemented by publication time.
The April, 1994 student implementation numbers, 3604 Lab and 7168 Lecture need to be updated and corrected for under reporting. The September, 1995 (estimated publication date) implementation numbers are estimated to be 5,000 Lab and 12,000 Lecture.
The rate of student performance of workshop related laboratory experiments is estimated at 2,000 per year. The acceleration is about 1,000 per year per year.
Estimate of 90% confidence level: 10% for the 5,000 student lab experiments; factor of 2 for the 2,000 student current rate; factor of 4 for the 1,000 student acceleration rate.
266 is the total count of faculty who benefited from the project to date either by attending a workshop or requesting a copy of the NSF Microfab Manual [1]. Although the NSF Workshop is complete, the Center will continue to distribute copies of the NSF Microfab Manual on request. The Center continues to offer Microfab short courses, also.
Many participants pointed out that their research activities, and research activities by their students, were improved by participation in the NSF Workshop. This benefit has not been measured. Simple, inexpensive research should be the subject of a separate report.
Many participants insist that there is subjective value in the academic pleasure of attending the NSF Microfabrication Workshops. This benefit was not measured.
The NSF Microfabrication Laboratory Workshops were successful. United States faculty obtained information and learned laboratory skills by attending this series of workshops. Simple, inexpensive microfab student experiments have been set up at many universities and colleges. A large number of undergraduate students have benefited from this. Many if not most existing IC Labs at universities are underutilized. Additional large numbers of students are being provided with simple experiments from the NSF Microfabrication Laboratory Workshops for work in existing IC Labs. University and college faculty continue to implement simple microfab experiments after studying the NSF Microfab Manual, which is updated on this web site.
[1] P. Gwozdz, NSF Microfab Manual. The manual has not been updated since 1994. This web copy serves as the update.
[2] P. Gwozdz, "Semiconductor Manufacturing Education at San Jose State University," IEEE Trans. Semiconductor Manufacturing, Vol. 5, no. 2, pp. 153-156, May 1992
[3] L. Fuller et al, "Microelectronics Manufacturing Education", in 1993 IEEE / SEMI Advanced Semiconductor Manufacturing Conference & Workshop, pp. 26-33, Boston, October 18-19, 1993, published by SEMI, Mt. View, CA, 1993
[4] D. Kerns, "Microelectronic Manufacturing Engineering Curriculum Development", IEEE Trans. Educ. Vol. 32, no. 1, pp. 4-11, Feb. 1989
[5] L. Fuller et al, "Microelectronic Engineering at RIT - Ten Years of Industry Partnership", in Proc. 10th Bienial University Government Industry Microelectronics Symp. p. 23, 1993
[6] M. Serra et al, "New Advances in Microelectronics Education: A Canadian Model", IEEE Trans. Educ. Vol. 36, no. 1, pp. 141-147, Feb. 1993
[7] R. Compton & R. York, "A Hands-On Microwave Laboratory Course Using Microstrip Circuits", IEEE Trans. Educ. Vo. 33, no. 1, pp. 161-163, Feb. 1990
[8] M. Haskard, "Low Cost Hand-On Training in Microelectronic Technologies", IEEE Trans. Educ. Vol. 36, no. 3, pp. 295-301, Aug. 1993
[9] M. Morales, database query printout, personal FAX communication, 11 Feb 1994
[10] 1990 SRC Source Book, Higher Education Publication Inc., Falls Church VA
Send your ideas to the Center. Consider writing a paragraph for the "Implementations" section. Consider writing a full subsection for the "Workshop Discussion Items". An email attached document will work well. On the other hand, if your time is limited, please phone in your ideas.
Mail: Center For Electronic Materials & Devices
College of Engineering San Jose State UniversitySan Jose, CA 95192-0080
Phone (408) 924-3931; leave a recorded message if no answer
FAX (408) 924-3818email pgwozdz@email.sjsu.edu
Basic Integrated Circuit Technology Reference Manual
Editor Richard Skinner
Integrated Circuit Engineering Corporation (ICE)
15022 N. 75th Street
Scottsdale AZ 85260
This $100 book is produced by ICE, a consulting firm. It provides a simple overview of IC technology. The material is simalar to the "orange book" by Gise & Blanchard that we use at the workshops.
CORD
Semiconductor Manufacturing Technology Training Aids
CORD/SMT 7030
Center for Occupational Research and Development
601 Lake Air Drive
Waco TX 76710
800-231-3915
This is a listing of videos and computer programs for technical training, including safety, chemistry, etc. Robert Stone, 3801 Campus Drive, Waco TX 76705, 817-867-1237, a participant in the first workshop, has information on the CORD training programs for the IC industry.
Fortino
Fundamentals of IC Technology
Reston
A book like Gise & Blanchard, recommended by a participant in the first workshop.
Gise, Peter & Blanchard, Richard
Modern Semiconductor Fabrication Technology
Prentice Hall 1986
This is the text used for workshop lectures. It is also used for short courses at the Center. It is probably about as easy reading a text on fab as you can get for engineering students without dropping down to general education level. It has equations, but no calculus. It is used for the workshops not because the lecturer believes it is the best book, but because the lecturer has experience with it. It is good for classes with a wide spectrum of knowledge on fab. It serves the purpose of providing a format with lots of illustrations for open discussion style lectures. Also, it has a nice glossary, which justified not putting a glossary in this NSF Manual. You may have difficulty finding copies in bookstores, because it is out of print; Dick Blanchard (415-948-3073) promises to get it back into print soon. Meanwhile, he gave me permission to make copies for you until it gets back into print. This book was originally written for a class for technicians at Fairchild Semiconductor in the 1970's. Updates have not modified the basic old technology foundation. So it is not modern, just basic and easy.
Grove, A.S.
Physics and Technology of Semiconductor Devices
John Wiley 1967
The bible. But no photolithography. It's out of date, but most of the dozen or so books available on the subject paraphrase this one, in some cases almost chapter for chapter. Find this one in the book store or library and there will be several modern versions next to it by other authors. The modern ones are of course a better purchase. There are very few books exclusively on fab; Grove and most others cover fab in the first few chapters and spend most of the book on devices. Many of the books in this genre have OK fab chapters, including photolithography. No particular recommendations.
IEEE Transactions on Semiconductor Manufacturing
One of the IEEE family of journals. ISSM (International Symposium on Semiconductor Manufacturing)
This annual symposium (formerly called ISMSS) is sponsored by SEMI and IEEE. A Proceedings is published. See SEMI, next section. The first non California symposium will be June 21, 1994, Tosho Hall, Tokyo. Next year, back to California.
Jaeger, Richard C.
Introduction to Microelectronic Fabrication
Modular Series on Solid State Devices, Volume V
Addison Wesley, 1990
A very good book, on a slightly higher level than the Gise & Blanchard book that we use in the Workshop. Several workshop participants have recommended this book as a text for undergraduate or graduate courses. I agree. Modular Series is a series of about ten volumes.
Semiconductor International
Box 5700
Denver CO 80217-9889
Very similar to Solid State Technology. Same comments as for the latter.
Silicon Magic
Silicon Run
Silicon Run II
Semiconductor Equipment and Materials International (SEMI)
805 East Middlefield Road
Mt View CA 94043
415-964-5111
These are three videos on Microfab. Silicon Magic is shown on Monday at each NSF Workshop. Participants always agree it should remain on the agenda. At SJSU it is also shown to all engineering undergraduates as part of a Materials Engineering 25 lab. It is also shown at Center short courses. Silicon Run is the older version, but Silicon Magic differs only in that newer models of equipment are showcased, and there is more flavor of the "magic" of silicon. Silicon Run is more technical and longer. The 1994 Silicon Run II emphasizing packaging.
Solid State Technology
Box 3689
Tulsa OK 74101-9928
A good source of technical process technology articles, particularly as a first reference for professors and students who are new to the technical details of a particular process. A bit more technical than Semiconductor International. This is a free monthly magazine; you need to fill out the form that is included in most issues. The form is used to "qualify" you for this privilege; in fact it is used for the mailing list, which is sold. The Center uses the mailing list for short course advertisement.
Sze, S.M.
VLSI Technology, 2nd edition
McGraw Hill 1988
This is the best text on microfab. It is used for an undergraduate and a graduate level course at SJSU. Use this one to learn subjects not covered by the lectures in the NSF Workshops. Not for beginners. Chapters by individual authors. Edited by the same Sze from Bell Labs who wrote the thick blue book on device physics.
van Zant, Peter
Microchip Fabrication
McGraw Hill 1990
Covers not just microfab, but all of microchip fab, including packaging. Great for advanced high school or beginning junior college.
This section is a mini buyer's guide for your reference and convenience. It is limited by time available to collect data. Please exercise the usual caution in purchasing; treat this like you would any other buyer's guide. Your library may have a multivolume buyer's guide, like the Thomas Register. For a buyer's guide with emphasis on microfab, see the magazine Semiconductor International; they do an annual "Source Book" issue.
Allied Signal Inc
1090 S. Milpitas Blvd
Milpitas CA 95035
408 562 0300
Spin on glass for simple single wafer diffusion doping
Ashland Chemical
East of the Mississippi: Easton PA; 215-258-9135
West of the Mississippi: Dallas TX; 800-228-3031
Etchants, photoresist strippers
Bid Service
Box 128
Bradley Beach NJ 07720
908-775-8300
A large national used equipment dealer. Never used by the Center.
Black Jack
Omega Specialty Instrument Co
4 Kidder Rd
Chelmsford MA 01824
508 256 5450; FAX 8015
Blow guns, for cleaning wafers.
Capital Equipment Connexion
546 Division Street
Campbell CA 95008
800 777 1766
Used equipment dealer. Never used by the Center.
Catalyst
3784 Fabian Way
Palo Alto CA 94303
415-856-1800
Used equipment dealer. Never used by the Center.
Darby Dental Supply Co Inc
Rockville Centre, NY 11571
800-448-7323
Jim Masi (January 13, 1992) uses their crown sintering furnace, $500 table top model, for oxidation of silicon wafers.
Gift-In-Kind Clearing House
Box 850
Davidson NC 28036
704-892-7228; FAX 3825
Non profit organization. Clearing house for 170 (e) (3) donations.
Headway Research
3713 Forest Lane
Garland TX 75042-6928
214-272-5431
Lab table top spinners. Rodney Turner sent a couple catalogues to the first NSF Workshop. He quoted $4300 for cheapest spinner, P/N 3-07032, with no options. Ask about 6% university discount. We ordered 6 of these for the NSF supplement; see Implementations.
Hoechst Celanese Corp
AZ Photoresist Products
50 Meister Ave
Somerville NJ 08876
908-231-5800
Photoresist, developer
HPS Hoosier Photo
8020 Zionsville Road
Indianapolis, IN 46268
317 875 9000 ext 502
Jerry Wagner (June 10, 1991) ordered a 10 X reduction camera from them, for printing masks on emulsion plates directly from DRAW PERFECT output. A double reduction will give 100 X.
KTI
KTI was a well-known supplier of photoresist, developer, etchants, strippers, etc. They are out of business. The photoresist and developer product lines were sold to OCG. The etchant and stripper lines were sold to Ashland.
Laurell Technologies Corp
Lansdale PA
215 699 7278; FAX 4311
Another vendor of inexpensive laboratory style spinners. WS-200-C series. This information is from an advertisement; this vendor was never contacted by the Center.
Machine World
45277 Fremont Blvd, Suite 1
Fremont CA 94538
510-659-1911
Lab table top spinners, under $2,000. Small recent private job shop that also rebuilds used spinners. Ask for Glenn Krone. He visited the Center, but I did not contact any prior customers.
E. McGrath Inc
35 Osbourne St
Salem MA 01970
508-744-3546; FAX 741-4020
Used equipment dealer. Name from S. Burkhard. Center never used them, but Pete phoned in 1994 and verified that they have used vacuum systems, microscopes, etc.
Meadow Lake Corp
25 Blanchard Drive
Northport, NY 11768
516-757-3385
Tom Bingham (June 10, 1991) purchases transfer sheets from this company, for PC board fabrication. The sheets look like copy machine overlays and are used that way to copy artwork. A heat transfer method is used to transfer the image to the copper on PC boards, for photolithography.
Mouser Electronics
800-34 MOUSER
Materials for student lab PC board, including etching. Never used by the Center; this info from a local high school, who dumps chemicals at the IC Labs.
MRL
440 North Central Avenue
Campbell CA 95008
408-370-3838
Furnace manufacturer. Small company. They have nice small furnaces for labs. They specialize in furnace elements, for replacement of burned out units, or for building your own. I toured their factory in 1988, and was impressed by their efficiency in design & manufacturing.
Nova Electronic Materials
2655 Villa Creek, Suite 131
Dallas TX 75234
214 620 1463; FAX 7505
Wafers. Prime, test quality, seconds, etc.
OCG Microelectronic Materials Inc
5 Garret Mountain Plaza
West Paterson NJ 07424
201-977-6002; FAX 6110
Photoresist, developer
Precision Photoglass
Palo Alto CA
Supplier of mask blanks, emulsion & chrome.
Shipley Co Inc
2300 Washington St
Newton MA 02162
617-969-5500
Photoresist, developer
Quintel
2431 Zanker Road
San Jose CA
408-435-1995
Rebuilds used Kasper contact aligners.. (Kasper, Cobilt, and Cannon are popular used contact aligners.) During the 1/6/92 workshop, Ward Collis visited Quintel to check on an aligner that they fixed up for his IC Labs. Ward recommends them as an inexpensive source for contact aligners.
SEMI
(Semiconductor Equipment and Materials International)
805 East Middlefield Road
Mt View CA 94043
415-964-5111
SEMI is an excellent source of vendors. Most IC suppliers belong to SEMI. SEMI sponsors trade shows, the big one, SEMICON West was held until recently at the San Mateo Fairgrounds, now at Moscone Center in San Francisco, next one July 19, 1994.
Semiconductor International
Good advertisements by vendors. Also annual vendor reference issue. See previous section.
Silicon Quest International
(Formerly Ziti)
2904 Scott Boulevard
Santa Clara CA 95054
408-496-1000
Wafer distributor. Good source for inexpensive silicon wafers for the lab. They have new wafers (misc boxes from many manufacturers), test wafers (seconds, usually with a few particles each), and reclaims (used wafers ground and repolished). There are many such wafer dealers; this one is a large outfit that I have used. Check your local dealer, too. See Workshop Discussion Items, Obtaining Silicon Wafers.
Solid State Technology
Good advertisements for vendors. See previous section.
Specialty Coating Systems Inc
5707 West Minnesota Street
Indianopolis IN 46241
800 356 8260
Another vendor of inexpensive laboratory style spinners. P-6204 model. This information is from an advertisement; this vendor was never contacted by the Center.
Source
2200 Martin Ave
Santa Clara CA 95050
408 988 0200
Used equipment dealer. A big one; owned by GE. Visited, but never used by the Center.
Vector Technical Group
3080 Olcott Street, Suite 105A
Santa Clara CA 95054
408-727-1966
Tweezers, wands, etc. Specialize in epi parts & rebuilds. Also gas jungles. Never used by the Center; this info from a mailer.
Wafernet
1641 N. 1st St #120
San Joe CA 95112
408-437-9747
Wafer distributor. Good source for inexpensive silicon wafers for the lab. They pass their excess inventory on to SJSU, so if you need free wafers, call SJSU, not Wafernet. For purchases, ask for Shannon John; she attended a short course at SJSU IC Labs. See Workshop Discussion Items, Obtaining Silicon Wafers
Yellow Pages
Don't forget to look here for vendors & dealers. Local is best.
Zircar
110 N. Main St
Florida NY 10921
914-651-4481
Jerry Wagner (June 10, 1991) ordered a furnace from them, for $2050. They do not make the controller and thermocouples, but they gave Jerry a list of six suppliers; Jerry ordered for $1,000. It looks like an IC Lab can get a complete furnace for under $4K.
This section describes some major implementations by workshop participants. This section was written in 1994.
NSF provided supplementary funds in summer 1993. The supplements were used to fund supplies and 50% of the cost of a simple commercial spinner. The intent: Set up our so-called "Paradigm" experiment. For details, see V. The Experiments. Seven prior workshop participants benefited from this $30,000 supplement. The seven are: June 10, 1991: Bingham, Shaban; June 24, 1991: Hsu; January 6, 1992: Garabagi, Joshi, Sveum; January 13, 1991: Disney.
Tom Bingham (June 10, 1991) has been very active. Tom returned to San Jose as an instructor for the June 24, 1991 workshop. He set up a tripod and camera in the IC Labs. He worked out the details on how to photograph a circuit layout hanging on the wall and how to use the negative as a mask. He used a Polaroid 35 mm Autoprocessor, a hand-crank developer. We did not actually make any circuits that week, but many of the participants etched reduced copies of their workshop completion certificates in aluminum on oxidized wafers, at about 10 microns resolution. Recently, Bingham finished documentation of his method; see page 39.
35 mm film is a cheap method to make masks. We have also used two other methods: sending artwork to the campus photography shop, ordering high resolution slides; and sending artwork to a local (yellow pages) microfiche service. See also Salmon, below. All these give down to 50 microns resolution with minimal effort; with experimentation, you might get down to 10 microns. The copy machine overlays which we use are about 200 microns resolution.
Bingham also reports that he has oxidized silicon wafers using a gas fired metal heat treating furnace. Oxide uniformity is excellent for his one wafer oxidation process, because of the high local heat transfer by radiation. Metal contamination is likely, but these wafers are great for photolithography classes.
Bingham is active with more documentation of workshop items. He sent me a copy of his package for the Alpha-Step, a surface profilometer built by Tencor that we used in the workshop for film thickness measurement. See Meadowlake in the Vendor section of this manual for his heat transfer overlays.
Robert Stone (June 10, 1991) of Texas State Technical Institute, Waco Texas is the program coordinator for the Semiconductor Manufacturing Technology program at TSTI. He shared the program plans with the June 10, 1991 workshop participants, and he reported back plan changes based on his workshop experience. See CORD in Books & Literature section for Bob's list of training aids.
Jerome Wagner (June 10, 1991) is building an IC Lab with a minimal budget at Rose-Hulman Institute of Technology. See the Vendor section, Zircar and HPS, two of Jerry's sources, for furnaces and reduction cameras.
Linton Salmon and Richard Woodbury (June 10, 1991) from Brigham Young University also have an IC Lab. As a result of the workshop, Linton set up a routine for producing inexpensive masks for fabrication. They send personal computer Postscript based artwork by campus network to the BYU press, where a commercial system for desktop publishing is used to produce a film negative. Linton reports 50 micron resolution; he makes contact prints onto emulsion glass masks from the film.
Robert Minniti (June 24, 1991) of Nortre Dame phoned his students during the workshop. He told them to try to make MOS transistors and diodes using copy machine transparenciess for masks. The students made the masks and started wafers before the workshop was over. Unfortunately, there were too many pinholes at contact mask, so the transistors, which were finished a few weeks later, shorted out. Diodes and resistors worked. Bob also tried an ink jet approach but was not satisfied with the quality of mylar he was using. He'll switch to paper print out of student designs followed by photographed slides for masks. (Slide quality film negatives are used at SJSU for transistor fabrication at millimeter design rules.)
By the way, Salmon and Minniti both say that the workshops are well worth while for experienced professors with IC Labs of their own. I'm flattered and surprised. The workshops were designed for beginners. Experienced IC profs were not expected at these workshops, but each workshop has had at least one. The June 10, 1991 workshop was unusual with four of them. Comments and impromptu lectures by these experts are appreciated by all. Minniti offered his IC Labs for a follow up workshop.
Robert Warrington (June 24, 1991) of Louisiana Tech was already working on the design of the building and facilities for the Institute for Micromanufacturing before he attended the workshop. The Institute will use the LIGA technique and other techniques for research on micromechanical devices (like tiny gears). Based on his workshop experience, Bob modified the plans to include simple fabrication to complement the advanced fabrication being planned for the Institute.
Mustafa Guvench (June 24, 1991), University of Southern Maine, saw a professor at the University of Istanbul use a microscope backwards. I had been kicking this idea around for some time, so I mentioned it at the workshop. Mustafa verified that yes, artwork can be placed where the film usually goes in a microscope camera. Then, using UV illumination behind the artwork, which is either glass or plastic with dark images, the microscope optics can be used backward to project a reduced image onto the photoresist covered wafer or glass mask. If filtered yellow light is used before hand, the microscope stage can be used to align the wafer to the image. In other words, a direct step on wafer aligner can be built for a few thousand dollars, the cost of the microscope. Who will be the first to do this and document the details for this manual?
Karen Avanessian (January 6, 1992) of Diablo Valley College returned to the IC Labs three times. He brought 65 students from the three semesters. They each did the Paradigm experiment, going away with a souvenir wafer.
James Masi told the January 13, 1992 Workshop about a very simple oxidation method. He uses a dental furnace for oxidizing silicon wafers. These $500 table top furnaces are intended for sintering crowns, but work great for IC fab. See Darby in Vendors. His results have been repeated by Green of SJSU. Jim sent me samples of his mask films; he has no trouble doing about 30 micron resolution with plastic films for masks. Jim also sent a box of color slides that he took during the workshop (not necessarily for wafer printing); the group photo is great.
Katy Disney and Charlotte Behm (January 13, 1992) have a tiny budget. They applied for the supplement because they need a microscope. They are building their own spinner, using the plans that were distributed by Ramana Balagani at the workshops. (Ramana of SJSU built one as a student project; page 39.) Katy also plans to build her own exhaust hood by adapting a home oven unit from an appliance store. When she is finished, I will use her as an example: no excuses about inadequate budgets!
Fred Padilla (June 22, 1992) mailed me a copy of a 1962 Bell Labs booklet, Energy From the Sun. This 90 page booklet, written by D.H. Chapin, originally came with a kit for construction of a solar cell. The construction is intended as a high school experiment. Boric acid, diluted with alundum, is used as the p type dopant for n type silicon. A furnace is constructed by surrounding an electric heating element by bricks.
Lynn Fuller, Rochester Institute of Technology, is not a participant in the NSF project, but we often exchange notes. I'm mentioning him in this section for want of a better place. Lynn runs the IC facility at RIT. He offers a one week fabrication course. I recommend that course to the NSF workshop participants who are interested in a short course at a higher technology level than our NSF workshops.
Malcolm Haskard, from the University of South Australia, mailed me copies of his lab experiments. email ETMRH@Levels.UniSA. Malcolm has independently been doing simple, inexpensive IC experiments down under. He is not a participant in our project, but we welcome his documentation. Haskard's student handouts are listed in List of Experiments, below; one of them is attached at the back of this manual.
Please email write or phone with your implementation news, so it can be included in the next version of this Manual. See Communication
This section contains topics discussed at the NSF Workshops. Opinions of the author that did not achieve near consensus at the workshop are indicated as such. Opinions expressed without reservation represents near consensus of the participants. Naturally, opinions of the author (Peter Gwozdz) that received near unanimous rejection are not documented.
This section, written in 1994, has needed only minor modifications since. It is still current.
These same discussion items are briefly summarized in the published article that is reproduced as the first section of this web site.
Our theme is that Microfabrication experiments can be simple and inexpensive. A discussion subsection in a web site such as this cannot prove the validity of the theme. The theme is proved by the NSF Workshops. Many of the participants come to the workshops with the idea that all Microfabrication work is difficult and expensive. After performing a few experiments, all participants agree that meaningful and instructive student experiments can be performed simply and inexpensively. At the workshop discussion sessions, not one participant disagreed with the theme.
It is difficult and expensive to make 64 M bit DRAMs for a cost of a few dollars each. Modern Microfabrication factories employ Ultra Large Scale Integration (ULSI), which is a difficult, expensive extension of the basic Planar Technology. The basic Planar Technology (photolithography plus oxidation of silicon plus diffusion in silicon plus aluminum metallization), is one simple example of Microfabrication.
The notion concerning the difficulty of Planar Technology is widespread. Most review articles on Microfabrication emphasize the amazing achievements of ULSI without a reminder that ULSI evolved from a simple ancestor. Many workshop participants described faculty meetings where budget constraints were invoked to rule out microfabrication experiments. The primary purpose of the NSF Workshops is to set aside this mistaken notion.
Probably the simplest experiment is chip examination. (The word "experiment" is used in this web site as a general laboratory term to include demonstrations and exercises.) In the workshop, we pull or clip chips with ceramic packages from scrap printed circuit boards. Then we use pliers to break open the package for chip examination in a microscope. This is an example of an experiment that requires only objects available in almost every university.
Other simple experiments require instruments that can be obtained with a modest budget and which require almost no maintenance. Examples are: V/I measurement, thickness measurement, and step height measurement.
When fume hoods with sinks for acids and solvents are available, many more simple experiments are possible. Photoresist processing and acid etching require hoods and sinks, which are widely available in universities.
The so called Paradigm experiment for photolithography is described in the next subsection. The author of this manual is obsessed with the Paradigm experiment and would like to see every science and engineering undergraduate in the country perform it. The obsession was not shared by all NSF Workshop participants, but most participants expressed pleasant surprise on how simple and inexpensive a significant photolithography experiment can be. Participants are actively implementing the Paradigm at their institutions; see the Implementations section.
Many (but not a large percentage of) universities have expensive microfabrication labs. About 10% of the NSF Workshop participants work in such labs. These labs are great. We wish everyone had one. This manual is not intended to disparage the good work going on in these labs. But such labs are expensive. These experienced participants all expressed frustration at the small number of students that can be accommodated by the sophisticated experiments that are traditionally performed. Usually, a single class of 5 to 20 students per semester perform a complete fabrication of a simple integrated circuit. These experienced NSF Workshop participants are well positioned to add laboratories for large enrollment classes. Most said they will do so. At SJSU, about 300 students per semester do the Paradigm experiment.
A word about magic in closing this subsection. I don't really believe in supernatural magic, but the basic planar technology fits the instinctive human notion of magic. Most of us have the experience that high technology is difficult; most of the technological things we try do not work; most of the things that work do so only with constant attention. ULSI is a very difficult extension of the basic Planar Technology. But the basic planar technology is pervasive precisely because it was easy to set up thirty years ago. Think of it as unusual (magic) because it passed through the fine sieve of practicality. That means we have to set aside our vast experience when dealing with student lab experiments in this subject. If your implementation of one of the basic experiments in this manual does not work, please disregard your trained methods of careful study and research! Temporarily forget experimental design. Just tweak something and try again. Believe me, that's the fastest way to get these experiments running. I have seen many highly trained scientists get outperformed by naive technicians in the IC industry when dealing with the basic planar technology, which is pervasive because it works like magic for untrained technicians. The only technology more magical than the basic planar technology, in my experience, was my 1956 Chevy; it also responded nicely to naive repair. In my opinion, the only element more magical than silicon is carbon.
A simple photoresist experiment is an excellent educational laboratory experience for undergraduates. Experiment S1PHOTO is offered for that purpose. Experiment S1PHOTO is strongly emphasized by the Center at SJSU; more than 600 students do the S1PHOTO exercise annually. S1PHOTO is coupled with S2ETCHAL to produce an aluminum etched MOS structure souvenir artwork. This MOS artwork experiment is performed by more than 150 short course participants annually. In addition, all the participants of the NSF Workshops produce at least one MOS souvenir artwork.
The word "paradigm" is used generically as a name for the set of all simple photolithography experiments with or without etch. The paradigm includes S1PHOTO, S2ETCHAL, and many other variations.
One objective of the NSF Microfab Manual is to get you to set up a paradigm experiment at your university. Remember that the word "experiment" is used in this manual to mean experiment, or exercise, or demonstration. S1PHOTO is an educational laboratory exercise. Some variations, such as contrast optimization experiments, are data analysis experiments.
This subsection is not an argument. Your discussion is solicited. If you think the paradigm should be significantly modified, or if you think it is not that great an educational tool, please say why. So far, the consensus at the NSF Workshop discussion sessions is that a paradigm experiment is well worth setting up in as many universities as possible. The experience at SJSU is that many undergraduates, even in engineering, quickly obtain a fresh intuitive understanding that photolithography is the essence of microfabrication by performing it. Freshmen and sophomores do S1PHOTO as part of Materials Engineering 25, required for all engineering undergraduates at SJSU. They say S1PHOTO provides a real enough modern fabrication experience for them, at an early stage of their education, and that it is fun.
At SJSU, all students get to take home their patterned resist on bare silicon souvenir. The Center has thousands of wafers, donated by local companies, available. Request some for your students. Alternately, wafers may be stripped of photoresist and used again.
There are many variations in the details of how to do S1PHOTO. See the last subsection below. Please try your own variations, and send a description of your successes to the Center for future editions of the Manual.
Many of you will implement S1PHOTO plus an etch exercise. Great. A prior thermal oxidation experiment ties in nicely. Aluminum over oxide is great. Many other thin films will do.
Some of you will use existing equipment, such as an aligner or an automatic developer. Fine. In this regard, S1PHOTO is typical of most of the NSF Microfab Manual experiments. It is not an exact recipe for you, the faculty member in charge of the lab. Use whatever is available to save money, add whatever is available to jazz it up. Write your own recipe for use by the students, but by all means feel free to copy sections from the example student handouts at the back of this Manual.
The MTI spinner at SJSU has three nozzles for auto dispense of three different kinds of resist from reservoirs under computer control. It takes a couple hours to clean the lines when the system is not used for several days, so we never use the auto dispense option. When 250 students use the spinner during the one week experiment for Materials Engineering 25, the clean up time would justify use of auto dispense. However, we judge that students gain a better hands-on appreciation for what photoresist is by pouring or squirting the liquid onto the wafer themselves. So we have an interesting combination of an automated spinner with computer programmed process and mechanical handling of wafers from a cassette, plus manual dispense of resist by the student!
The spinner is the only significant piece of equipment. The next subsection discusses the acquisition of a spinner. All the rest of resist technology has been reduced to inexpensive standard equipment in S1PHOTO. It could be argued that the spinner should be eliminated. A paint brush would do the job. So would a spray gun. After all, if a million dollar stepper-aligner-exposer can be replaced by a lamp and piece of wood, why can't the spinner be replaced? Well, the aligner must be replaced. It takes hours of practice to learn how to use even a relatively inexpensive contact aligner. Having each student learn the details of how to use an aligner-exposer is not justified for S1PHOTO. Having a technician expose the students' wafers as a batch would detract from the hands-on tone of S1PHOTO. Replacing the spinner, on the other hand, would make S1PHOTO appear not enough like real technology.
The previous two paragraphs may sound like a subjective apology for the way we do things. Maybe it is. But the subject was discussed in every workshop, and participants agreed that the points make sense. It is a workshop consensus that S1PHOTO is a significant and minimal distillation of the essence of photolithography for undergraduate students.
This subsection will discuss three different styles of spinner for acquisition by a university lab. The approaches are: manufacturing spinner, lab table top spinner, home made spinner.
A manufacturing spinner these days always has automatic wafer handling and computer control of the process. Cost of a new one is minimum $50K, typically $200K. Rebuilt spinners, from used equipment dealers, typically cost about one fourth the new unit price, with a minimal guarantee.
As with most semiconductor manufacturing equipment, there are models that are unpopular because of system unreliability. For the case of spinners, the unpopular models can be easily avoided by making a few phone calls for recommendations from other users.
Unlike most other pieces of semiconductor manufacturing equipment, popular spinner models are easy to set up and maintain. Students do this at SJSU with no training. Even vacuum and exhaust facilities were done by students at SJSU for the spinner. For the popular models, down time, spare parts, and maintenance tricks are not tough issues.
Junk spinners are widely available as donations. Don't try to rebuild a junk spinner. Reconditioning them is not difficult in principle, but a used equipment dealer with experience and inventory in that particular model can rebuild more efficiently. At SJSU, we obtained three junk spinners as donations and traded them to Bay Area Technology, who rebuilt one at no cost and kept the other two as payment, OKed by the donor. (Bay Area Technology has since gone out of business.)
If you receive a donation of a manufacturing spinner that works, verify that it is a popular model (reliable) and take it. If your IC Lab budget allows more than $10K for a spinner, buy a new or used spinner.
A manufacturing spinner is not necessary for lab work. If you only plan to do S1PHOTO, a manufacturing spinner does not make sense. If you plan to set up several fab related experiments, you will want some experiments that involve examples of automated wafer handling. In this respect, your choice of spinner is related to other equipment in the lab. If you have no other modern factory equipment to showcase, obtaining a manufacturing spinner is a simple way to provide equipment related experiments. At SJSU, for example, we have an experiment on resist uniformity vs programmed spin acceleration ramp (not documented yet in the S series).
A lab table top spinner can be had on a budget of less than $10K. Most of you will go this recommended route. These are small spinners that allow one wafer at a time to be manually loaded onto a vacuum chuck. The spin motor has electronic control. Time and RPM are set by hand. These spinners were used in manufacturing in the 1960s and early 1970s.
There are established vendors for lab table top spinners, starting about $4K: Headway, Macronetics, Solitec, Ultratech. Incomplete list. Some are listed in the Commercial Vendors & Dealers section. Rebuilt units are available from equipment dealers.
You may line up a donation of a lab table top spinner. Take it, even if it does not work. These guys are easy to fix and maintain; students can do it.
Home made spinners are an option. Ramana Balagani, a graduate student in EE at SJSU, was assigned the student project of building one from scratch under the NSF grant. This was in part to demonstrate feasibility as a student project, in part to provide a spinner for mask making demonstrations. Ramana did the job in two months, spending about 200 hours, and less than $1,000. The design includes electronic control of manual set spin speed, with two time cycles. The project included design, specification of parts, purchase of parts, construction, and documentation. He demonstrated the spinner as part of the NSF Workshops. Ramana's documentation is available through the Center. It can be argued that it makes more sense to buy an off the shelf spinner from a reputable manufacturer for $4K. But if your budget is severely limited, and you have student projects courses, this may be the way to go. With used motors, minimal electronic control, and student shop wafer chuck, you can get the cost under $100.
Consider jobbing out your spinner to a machine shop. See Machine Word in the Commercial Vendors & Dealers section.
Someone reported using a drill! (Sorry, my notes from the 1/5/93 workshop do not have the name.) A hardware store fly cutter was chucked into a Black and Decker drill. Wafers were spun OK at 3 KRPM for photoresist application.
Fab experiments involve chemicals. So safety precautions are about the same as for a chemistry laboratory. HF is the most hazardous chemical used by the S series experiments.
Documents S3 and S4 are the student handouts used by the Center. Use them as a checklist in planning your safety program. Please feel free to copy them in whole or in part.
We do not do formal safety training for the simplest experiments in the IC Labs at the Center. For hundreds of students annually, this would be impractical. Before a group goes into the IC Labs, the instructor tells them about safety glasses and gloves in about one minute. Then a lab instructor at each station gives safety instruction to each sub group of less than ten students.
So there are three stages of safety control in the IC Labs: The lowest stage is no formal instruction for simple experiments. Lab instructors must be present at all times, with a maximum of ten students per instructor. The lab instructor covers a checklist of safety items that apply to the particular station. NSF Workshop participants are treated this way on Monday morning.
The second stage is for regular student users of the IC Labs. They each must sign the S3SAFE sheet after formal instruction. S3 is attached to this Manual. NSF Workshop participants sign this sheet at Noon on Monday. In workshop discussions, the participant consensus is that the multiple stage safety approach is a good model for other IC Labs.
The third stage is for students who have a key to the IC Labs. They each must sign S4SAFE. Acid neutralization is the only environmental issue in the S series of experiments. The IC Labs has a scrubber for exhaust from CVD systems and plasma etchers, but this is not necessary for the S series. The small amount of vapor produced by the exhausted spinner is well within the usual allowance for unscrubbed university laboratory fume hood exhaust.
Oxidation of silicon to make uniform thin films of silicon dioxide (oxide) is easy. Silicon is the only material that oxidizes to make a nice uniform passive film a few tenths of a micron thick. 1000C is the nominal temperature; 800 to 1200C works. The radiant nature of heat transfer at 1000C makes temperature control easy in a small region. Silicon oxidizes in any combination of oxygen and water vapor. Ambient inert gasses, including nitrogen and argon, only slow down the oxidation.
In other words, you can set up a satisfactory student oxidation experiment in a room air furnace. It is worthwhile for students to experience silicon oxidation. Perhaps the most serious academic caution is that the students will draw the mistaken conclusion that many materials oxidize this nicely.
For controlled, predictable oxide thickness, the partial pressures of oxygen, nitrogen and water vapor must be controlled with flow meters. For volume production, temperature must be precisely controlled over a large furnace region. But for a lab situation, oxidation responds to feedback. You can shoot for low thickness, measure, and oxidize again for short times and get any thickness you want. The discussion of magic in Simple, Inexpensive Experiments applies.
For diffusion doping, it is nice to minimize the oxygen with inexpensive gas bottles and reasonably air tight construction. At SJSU, we have done diffusion doping in air ambient. Almost all our oxidation is done in air.
Furnaces can be purchased from several established manufacturers. Thermco is a popular source. Mellon Furnace Co was recommended by a participant in the first workshop. MRL is listed in the Commercial Vendors & Dealers section as one of several suppliers of small lab type furnaces. Many labs make their own. See Implementations for reports on Bingham and Masi using inexpensive metal and dental furnaces for silicon oxidation. A suggestion was made that a "red hot heater" from hobby shops could be used to oxidize wafers.
Used IC manufacturing furnaces are recommended. They are widely available for under $5K from dealers. The cheapest need work, but fixing a furnace is well within the technology of student projects. After all, a furnace is nothing more than a sophisticated assembly of toaster wire and brick, with thermocouples, electronic temperature controls and a quartz tube.
Our primary theme is that Microfabrication experiments can be simple and inexpensive. There is a secondary theme: Submicron device fabrication can be simple and inexpensive. About 20% of workshop participants expressed interest in laboratory fabrication for device education and research. The other 80% can read this subsection for entertainment.
The purpose of this subsection is to justify the secondary theme. First, an argument will be presented to convince the reader that micron resolution in photoresist (resist) is simple and inexpensive. Next, the argument will be extended to the basic planar technology.
There is a mistaken notion that resolution of one micron requires advanced, expensive equipment. Review articles on Microfabrication often give graphs of minimum line width, or resolution, vs year. These graphs show manufacturing line widths decreasing from about 25 microns in 1960 to about 0.7 microns in 1994. The companion text usually emphasizes how expensive and difficult the progress has been. It is rarely pointed out that this progress refers to the low cost manufacture of circuits with the largest possible number of transistors. People forget that in the 1960's submicron resolution was routinely achieved in university laboratories. Factories producing discrete high speed transistors have always worked below one micron. It is difficult and expensive to make millions of 4 M bit DRAMs for a cost of a few dollars each, but it is not difficult or expensive to make just a few bits for purposes of research or education. There are two key technical points that will drive home our message: field of view and uniformity:
Field of view is a measure of the size of an optical image. Photographers know that wide field camera lenses are more difficult to make and more expensive than narrow field lenses. Similarly, microscope buyers know that for a given magnification, microscope cost increases with the size of the field of view. In fact most readers of this paragraph have probably sat at an inexpensive microscope at one time or another and personally resolved about one micron albeit with a field of view much smaller than a modern ULSI chip. Modern ULSI requires a submicron image field of view greater than a centimeter; that's what's difficult and expensive. If you wish to make isolated submicron devices for research purposes, simple, inexpensive optics is sufficient. If you wish your students to make single transistors that are modern enough for educational purposes, 1 micron resolution should not be a big deal. Just do not try to make 16 M bit DRAMs.
Now that means you should not just buy masks. There is no market for inexpensive one micron masks for research purposes. You need to make your own. That's not trivial, but neither is it terribly difficult or expensive. I should write a separate discussion subsection on ways to make your own masks. Alternately, you can go ahead and buy expensive IC quality masks and restrict your savings to inexpensive contact alignment (separate future subsection).
Uniformity is a measure of the variation of an output parameter when one tries to keep input parameters constant. For a specific example consider CD (an acronym for "critical dimension", used for any resist printed dimension). CD is the output; inputs are exposure time and intensity, soft bake time and temperature, develop time and temperature, developer concentration, resist thickness, and so forth through the legion of things that affect CD. CD is intimately correlated with resolution.
The control of uniformity is related to the practice of selection and feedback. Two examples: If a process gives a CD which varies across a wafer, one may select the section of the wafer which gives the desired CD. If a process gives a CD today which is different than the CD for the same (attempted to be the same) input parameters one month ago, the exposure time can be varied a little today ("tweaked" ) to produce the desired CD on the next wafer. The goal of manufacturing engineering is to minimize selection (maximize yield) and to minimize feedback (maximize process control). For ULSI competitive manufacturing, the goal is difficult and expensive. But selection and feedback are OK in a laboratory environment. (An argument can be made that selection and feedback are instructive for students.)
It turns out that resist is particularly amenable to tweaking of exposure time. That fact does not surprise photographers, who regularly experiment with exposure times. Resist on wafers is also amenable to rework, which refers to stripping unacceptable resist results and repeating the resist process. In standard resist technology, test wafers are used to determine the current status of a process, and to adjust exposure. These three techniques: tweaking, reworking, and running test wafers, are easy and inexpensive to implement in a laboratory situation.
That was the resist argument. To summarize: one micron resist patterning is possible in a university laboratory situation with minimal trouble and expenses. Submicron patterning is possible with a little extra trouble. Resolution of a few microns is easy. Now we need to extend the argument to the rest of the planar technology.
Wet (aqueous acid) etching is an easy and inexpensive way to transfer resist patterns to wafers. With feedback, measuring dimensions with careful re-etching, isolated CD's in the micron range can be easily achieved in a laboratory environment; undercut is compensated by mask design (oversize or undersize by undercut amount). For adjacent lines and spaces, resolution is degraded by undercut, but only to the extent of etched film thickness. Again, a conventional wisdom, that plasma etching is required for micron CD's, applies only to ULSI manufacturing. With feedback and tweaking at etch, it is easy in the laboratory to achieve sub micron isolated features and true resolution of about one micron with wet etching.
Manufacturing type plasma etchers are expensive. They are difficult to keep running. There are lab type plasma etchers available, which are more cost and trouble than wet etch but adequate for true submicron etching. Future subsection.
Oxidation and diffusion are the easiest part of the planar technology. Try oxidation of silicon in air to convince yourself about how easy this part of the technology is. Don't forget feedback and tweaking.
Aluminum evaporation is easy and inexpensive. ULSI uses sputtering (expensive & difficult) for optimum step coverage. Step coverage is a big issue for yield and reliability of large circuits, but can be compromised for university lab purposes.
Air filtration is a big issue for acceptable ULSI yield. But for fabrication of discrete laboratory devices, and for student experiments, air filtration is not a big issue; reasonable housekeeping, inexpensive room air filters, clean boxes for storage, and perhaps one or two laminar flow filtered work stations are sufficient. Standard commercial deionizers are adequate to provide residue free rinse water with minimal MOS threshold drift.
That finishes the basic planar technology! Most extensions of the basic planar technology are expensive and difficult to maintain. Examples: epitaxy, sputtering, ion implantation, CVD. Meaningful devices can be built either avoiding these processes, or limiting their application to one or two steps. Since these processes are widely available as wafer services (future subsection), it is easy and cheap to send the wafer out to a service company. Also, remember that local commercial manufacturers are often happy to run a few educational wafers through a single process step for a university as a favor.
Please understand that making a transistor is not as easy as S1PHOTO. We still do not make an actual device as part of the NSF Workshop. Phil Hoff built a solar cell during the week of his workshop, but it did not work (mask design error). I plan to add fabrication of a simple pMOS transistor to the workshop in 1995 after the process bugs are all found through a one semester course at SJSU. The primary purpose of the workshops is to inspire university faculty to have their students perform simple, inexpensive experiments like S1PHOTO. This is at about 200 microns resolution. However, once you do this, little extra effort and expense are required to make 10 micron devices. Micron and submicron devices require still additional effort and expense, but are well within reach for university laboratories.
The simplest and cheapest way to get wafers is to ask us for some. We have 10,000 available, from industry donations. These are restricted to student experiment use. We are commited to ship you what you need. No charge. I hope we will still be doing it while you read this. Write, phone, or email, page 9. You must name or briefly describe the experiment which is scheduled or planned, with date, number of students, and course number / title.
Another simple way is to buy a box of 100 from a wafer distributor. Most distributors deal with inexpensive test wafers as well as expensive prime quality wafers. The term "test wafer" is used generically here, synonymous with the purchasing term "seconds". (Technically, the term "test wafer" is sometimes used generically to mean any inexpensive wafer, and sometimes specifically, to mean a specific type, as distinguished from a "dummy wafer", or a "monitor wafer".) Test wafers are used by IC manufacturers to test or monitor a process. There are many distributors which specialize in marketing such inexpensive wafers. Silicon Quest and Wafernet are listed in the Commercial Vendors and Dealers section. There are many more; shop around.
If you request wafers from us, we will send test wafers. Test wafers are ideal for simple, inexpensive student experiments. Test wafers are discontinued spec wafers, out of spec wafers, end of run wafers, excess inventory, and reclaimed wafers. Specifications (specs) such as resistivity (doping concentration), doping type, orientation, thickness, flat location, and flatness are extremely important to ULSI manufacturers but unimportant to simple student experiments, such as S1PHOTO, oxidation, etc. Probably the only spec of importance for simple experiments is wafer diameter if manufacturing equipment is used, but even then wafers of the correct size but outside of strict tolerance limits are available as test wafers. Unpolished wafers are used at SJSU for S1PHOTO. Ask for wafers out of spec for particles (a few "specks" per wafer visible only in a strong light); these are great for student experiments.
Reclaimed wafers are wafers which have already been used for tests, or wafers which have already had defective IC circuits manufactured on them. These are lapped (thinned by grinding with abrasive) in order to remove device structures and repolished to look as good as new.
"Coin roll" wafers are inexpensive wafers stacked like coins for storage and shipment. They typically have some scratches, but should be OK for most student applications.
A typical low price for test wafers would be $500 for 100 four inch diameter wafers, shipped to you. This compares with typically $2,000 for 100 prime wafers from a new wafer manufacturer. All prices in between are possible, depending upon quality and availability. Smaller diameter wafers are generally cheaper.
Try calling a local manufacturer. Call your alumni who work there. Very often, these companies have hundreds of wafers sitting in a corner because it is not worth selling a "small" quantity to a used wafer dealer. Remember to say that you do not require wafers which are identifiable by type; a box of used oxidation test wafers, for example, are great for student experiments. (See reuse, below.)
It could be argued that wafers are not needed. A cheap substitute would be glass plates for photolithography experiments. Logically that is true, but most workshop participants agree with the subjective proposition that wafers should be silicon in order to stimulate student interest.
Reuse will be important to most of you. At SJSU, students keep their wafers as souvenirs, thanks to generous local donations. Most of you will reuse them. Photoresist can be stripped using solvents or oxidizing agents. Boiling sulfuric acid is a standard. Careful! Add hydrogen peroxide (when the acid is not hot!) or ammonium persulfate for complete oxidation. A beaker of room temperature sulfuric acid "grows" by absorbing atmospheric moisture, so careful about leaving it around. Such a beaker made a mess at SJSU on October 17, 1989. For wafers which have been used for oxidation tests, or which have had oxide lithographically etched, strip the oxide with hydrofluoric acid (HF) (plastic beaker, of course). Read S3SAFE, rule 5. Buffered HF, required for oxide lithography, may be used for striping oxide. Also, a dip in HF after stripping photoresist is a good way to remove the 100 Angstrom oxide produced during the photoresist stripping. (Without HF dip, silicon wafers stripped of photoresist are hydrophillic; with dip they return to hydrophobic.) For fine lithography, photoresist may lift from wafers which had been rinsed in water (or left out in the air for a day); for best results, bake at 150 C for an hour to remove adsorbed moisture.
Stripping for reuse is very inefficient in quantities of less than 25 wafers. There is another meaning of reuse; see "reclaim", above. Save your diffusion experiment wafers. Ask local manufacturers for IC rejects. (You should use reject IC wafers as student demonstrations, anyway.) Manufacturers get next to nothing selling these to reclaim companies, so most will be glad to let you come and get them for students. Once you have a few hundred, grinding and polishing would make a great student project.
This subject was briefly covered in prior subsections; see the last few paragraphs of Paradigm Experiment, and the following Spinner. We call your attention to the proposition, covered there, that it is very valuable to have at least one "real" manufacturing piece of equipment in your lab. Most of the workshop participants agreed to this subjective value. Many of you will obtain used manufacturing equipment.
Used semiconductor manufacturing equipment is widely available at reasonable cost. Free donations to universities are common.
Two points from the Spinner subsection are elaborated now: reliability and rebuilding. Reliability issues are notorious in the IC manufacturing equipment business. I believe this is due to the extreme marketing pressure of the push to increased number of devices per chip. The benefit of replacing four chips on a PC board by just one chip is extremely great; the price advantage is obvious; there is also a speed advantage; in addition, the reduced pin count gives enormous reliability advantage; finally, products which simply cannot be sold profitably with older technology become possible with increased device density. For this set of reasons, unreliable manufacturing equipment has been tolerated where such equipment was believed to make possible a new process for increased device density. Sometimes unreliable equipment is worth the effort in manufacturing at least for a few years; sometimes it does not deliver as promised in addition to being unreliable. Not everyone agrees with this analysis. Tons of paper have been used to discuss how much better the Japanese are at making equipment. Suffice it to say everyone agrees that there is a lot of unreliable equipment out there.
Most of this unreliable equipment is available to you as a donation. Stay away from it. One of these would be a millstone around your neck.
One example is the Zylin etcher. Scores of these aluminum anisotropic plasma etchers were sold at over $200,000 each. Some were actually used for manufacturing for awhile, but none (that I know of) are now. I get at least three calls per year asking me to come and get one as a donation. They are worthless for scrap because the cost of cleaning out the acid is greater than the value of the reclaimed materials. Some of the workshop participants also report such calls. In each case, the pumps, which do have value, had been removed.
But do not be scared off. There is a simple technique to identify which equipment models are reliable. Ask someone. Phone three or four people. Call the Center and ask me. Equipment reputations are just like people reputations; they are widely known. If someone is unfamiliar with a particular piece, ask him for a name of an expert on that piece, and call her.
How about examples of reliable equipment? Spinners, furnaces, contact aligners, and evaporators are generally reliable. Reliable evaluation gear includes spectrophotometers, CV plotters, profilometers, and dimension measurement equipment. But even the most reliable category, spinners, has exceptions. The GCA Wafer Track spinners, for example, designed for factory automation, are widely available as university donations, but are not recommended.
The concept of unreliable blends continuously into the concept of very complicated. Take a stepper aligner, for example. Every model of stepper is very complicated. These guys are very expensive to maintain; they require dedicated experts to keep them running. Expensive repairs are frequent. No one will disagree with this, not even the stepper manufacturers. Since steppers are required for modern ULSI manufacture, users tolerate the complication. A Cannon stepper is a different beast than a Cannon 35 mm camera, although they both sort of do the same job. My theory of tolerated unreliability applies also to tolerated complication. So when asking about reliability, ask also about practicality for university use. Steppers, epi reactors and ion implanters are not appropriate for simple inexpensive experiments. Sputter systems and CVD systems are marginally appropriate. Ask about others.
At SJSU, workshop participants use the 8110 Applied plasma etcher. Ordinarily, I would consider this model marginally appropriate for a student lab. In this case, however, Applied Materials donated not only the etcher, but free parts and maintenance. This is unusual.
Rebuilding nonfunctional equipment should be considered with caution. Again, ask someone with experience for advice. Sometimes rebuilding makes a great student project. In general, however, it is better to use the services of a company which specializes in rebuilding that particular model. At the workshops, a few specific participants reported success, particularly with rebuild of furnaces. But more participants reported frustration because the confidence and competence were not available to rebuild a nonfunctional donation.
You need names of used equipment dealers. You need names of specialists at equipment rebuild. Consult the Commercial Vendors and Dealers section for a start. Then consult the vendor list references which are in that section. The list is not repeated here to facilitate expansion and updating.
Equipment auctions are valuable. Bob Minniti of Notre Dame discussed the pros and cons of equipment auctions at the June 24, 1991 workshop. Bob has lots of equipment experience and confidence. I have never been to one, but I surely will someday, and I recommend auctions for about 5% of the workshop participants.
A word about tax advantages for donors. Most of you know that the donor gets to subtract from profit the book value of a piece of equipment donated to a university. You may not realize, however, that the subtraction comes "below the bottom line". Profits before taxes are reported without this subtraction. In other words, a donor may prefer a tax loss to an operations loss. For example, suppose a furnace originally bought for $100,000 has only $50,000 depreciation to date. Let's say it cannot be used because the wafer size has been increased. But used equipment resale value is only $10,000. So resale would result in a $40,000 deduction from reported profits. That looks bad. In many cases, it increases losses, or worse yet turns a profit into a loss. But donating that furnace to your lab leaves reported profits untouched, and, at 50% tax rate, reduces profit after taxes (or future taxes) by only $25,000. So don't look a gift horse in the mouth. Don't be suspicious that "the furnace must be worthless, or why would it be offered?" Remember, too, that the donor does not want to discuss this subject.
On the subject of taxes, your potential donor may not be aware of "gift-in-kind" deductions, allowed by IRS tax code 170 (e) (30). This law allows donors to deduct their cost plus 50% of their usual mark up. It is restricted to new goods, usually materials and supplies from inventory. The recipient must keep records for two years showing the donation was not resold. I'm sticking gift-in-kind in this section because it goes with the subject of donations.
Gift-In-Kind Clearing House, see Commercial Vendors and Dealers, handles code 170 donations, mostly for east coast institutions. There is a fee to join this organization. About 100 of their clients are universities. As written, the law does not specifically name universities as allowed beneficiaries, but in practice code 170 donations are often made to universities.
Remember too, there is such a thing as altruism. One of my biggest and most valuable donations, of perfectly good gear from a closed plant, came from LSI Logic. The guy in charge of getting rid of it was a SJSU alumnus. He called me early to be sure SJSU got it before other sections of the company found out about it.
Consider trading. Most used equipment dealers will take your valuable nonfunctioning equipment donation and trade it for something else they have which works. Always remember to mention this plan to the donor. Some donors have specific rules against trading, and you do not want to get a reputation as a rule breaker. At most institutions, equipment is tagged with inventory numbers and trading tagged equipment is not allowed. So trading needs to be executed without tagging, preferably without your taking physical possession, and the local university official in charge of equipment should be notified. It is surely unethical to execute a trade knowing the local official does not allow such transactions.
That brings us to ethics. Workshop participants told scandal stories. Individuals have lined their pockets through donations in various ways. Some universities have adopted very stringent rules concerning donations. The following is my analysis; most participants did not express opinions on such rules and I did not push for a consensus: The potential for abuse with donations is no different than the potential for abuse in purchasing, or in petty cash. Everyone knows horror stories about purchasing kickbacks. I tell the story (from first hand experience) of the secretary who copied petty cash forms and pocketed the receipts from extra submissions. When the company caught her, $30,000 too late, the petty cash procedures were tightened up very strictly. If I were an administrator, I would surely tighten up rules after a scandal, because two scandals in a row are infinitely worse than one scandal. That does not mean we should avoid donations because of possible scandals. It probably means that if there is a donation scandal, donations at that institution will be avoided for a few years. All this is moot, because it is beyond my comprehension how someone with a proclivity for stealing gets into donations, or how someone with a proclivity for donations gets into stealing. I know it sometimes happens, but it does not make sense.
Caveat emptor. Buyer beware. A related ethics problem concerns the purchase of worthless equipment. Workshop participants discussed this. There is a widespread fear of being taken by unscrupulous used equipment dealers. On this subject, there were plenty of opinions expressed by participants, but no consensus. Again, I give my analysis, although it did not receive a majority consensus. I'm more on the side of the used equipment dealers than most. I've seen many of them get burned by buying a warehouse full of neat equipment and ending up selling it for less than they paid, usually loosing their house in the bargain. My favorite used equipment dealer went bankrupt in 1993. Most used equipment dealers were hard working, competent, maintenance people who rose to manager at a large company before going off to get rich on their own. In spite of being well intentioned, most of them end up being hated by many of their buyers because of unrealistic expectations by the latter. I will concede that the used equipment market acts in a Darwinian manner, allowing survival, on the average, of the least scrupulous, or the most prone to empty promises. But I see as the driving force the insistence by buyers on empty promises. Please do not ask for a one year warranty on used equipment, that just guarantees you will end up buying from a liar. Understand what you need, ask for advice beforehand, and expect the worst when dealing with used equipment. Establish a good give and take relationship with a dealer. Send him your junk donations for whatever he can get from them and you'll find him fixing your broken equipment for free on occasion. For most universities the prior inquiry, the risk, and the later aggravation involved in buying used equipment is more than compensated by the benefit to the student laboratory.
How about junk? Many donations are of marginal value, or worthless. I never flatly refuse a donation, although I often end up saying no thank you after discussion and perhaps inspection. In most cases, I accept marginal donations, and in many situations most of the materials, supplies, and