The Birth of ECD: An Invention Factory (1965

ECD’s goals and culture remained much the same but Ovshinsky’s role changed, especially after September 1967, when the laboratory became supported by publicly traded stock.² No longer the lonely independent inventor, he became the head of an ambitious research laboratory. His earlier inventions, from the Benjamin Lathe to the Ovitron and threshold switch, had been created directly by his own efforts with the help of only his brother Herb and a few employees, and from the barn on Chaffin Road to the storefront on McNichols Road, the places where they were produced had been necessarily modest. Now, with the move to Troy and ECD’s subsequent growth, Ovshinsky had larger and better facilities plus a growing research staff. The many inventions that came from ECD over the following decades were still his creations, produced under his leadership, and directed by his vision, but others now did the hands-on work. Like Edison’s lab, ECD became an invention factory with hundreds of scientists and engineers and supporting personnel.³ And like Edison and many other inventors, as well as many academic scientists, Ovshinsky became an administrator—but one who was always intensely engaged with ECD’s ongoing research programs and the science behind them.

Instead of working in the lab, Ovshinsky needed to devote more effort to representing ECD and raising funds. Meetings with licensees, accountants, and attorneys absorbed an increasing amount of his time. He now regularly reported on the company’s progress to major investors and to ECD’s new board of directors. For such activities, and eventually for all his work, Ovshinsky now dressed regularly in a handsomely tailored three-piece suit, a costume change that reflected this important transition in his life as an inventor.⁴

As Ovshinsky’s role changed, so our narrative of his life also changes at this point. Here and in the following four chapters, ECD becomes a major character, sometimes diverting our attention from Ovshinsky himself. Its technological achievements all stemmed from his inventive genius, and its aims and character all arose from his and Iris’s social vision. But tracing the history of the institution they created will occasionally require displacing them to make room for the many new figures who enter the story.

Physical Review Letters and the New York Times

On November 11, 1968, Ovshinsky’s name suddenly became widely known to physicists working on semiconductors, when his paper describing his chalcogenide switches appeared in the prestigious journal Physical Review Letters.⁵ Until then, Ovshinsky had described his switches only in trade magazines, for which the screening process of scientific peer review is unnecessary.⁶ But the 1968 paper describing the experimental behavior of his threshold switch passed peer review, although with some difficulty.⁷

On the same day the Physical Review Letters article appeared, the New York Times published a front-page story about the switches written by the well-known Times science writer William Stevens. Ovshinsky was surprised and happy to see the article while the family was visiting the Interlochen Arts Academy in northwest Michigan, where as a treat for Steven they were attending a benefit concert on November 10 by the Soviet-born violinist and conductor Isaac Stern.⁸ Iris recalled, “The next day we opened the New York Times and there was Stan’s picture on the front page!”⁹ They learned that an article on the switches had also appeared on the last page of the Wall Street Journal.

Iris recounted the steps resulting in the Times article. “Stan decided this was going to be quite earth-shaking” and called a press conference, to which he invited reporters from the New York Times and the Wall Street Journal. Unlike the Ovitron press conference (see chapter 4), this was a simple affair, held in ECD’s new conference room. After being shown around the laboratory, the reporters asked Ovshinsky to comment about the future implications of his switches and memory devices. He predicted the development of thin-film computers and flat panel displays, while Hellmut Fritzsche and his fellow ECD physics consultant Morrel Cohen did their best to explain the science. When Stevens phoned the respected elder physics statesman Sir Nevill Mott for his comments on the switch, Mott said, “it is the newest, the biggest, the most exciting discovery in solid-state physics at the moment,” adding that unlike the transistor, whose principles could have been worked out on the basis of existing knowledge, the discovery of the Ovshinsky effect was “quite unexpected” and represented “totally new knowledge.”¹⁰

Figure 6.1 Article about Ovshinsky’s invention in the New York Times, November 11, 1968.

What appeared at first to be a triumph of recognition and publicity quickly backfired. Many academic and industrial scientists were outraged. The very publication of an important scientific claim in a newspaper seemed to taint Ovshinsky as a self-promoting charlatan. Fritzsche countered by pointing out that unlike Ovshinsky, “all the scientists who were criticizing Stan had fixed salaries from universities, from Bell Labs, from General Electric” and didn’t need to advertise and seek support for their ideas. On the other hand, the contrary offense of a leading scientific journal accepting the work of someone who had no PhD was equally infuriating. Both the Times and Physical Review Letters soon came under strong pressure to withdraw their articles.

As the reaction unfolded, it became clear that besides Ovshinsky’s use of publicity and lack of credentials, the very fact of his discovery incensed scientists at major research labs, where from their point of view such a discovery “should” more appropriately have occurred.¹¹ Some of the critics pressed Stevens to withdraw certain statements from his Times article, but Stevens showed them his notes and did not retract anything.¹² When they tried to get Mott to call the Times and retract his statements, Mott didn’t either, but he felt extremely uncomfortable about the affair and expressed his concerns to Fritzsche in a private letter. As a young professor, Fritzsche also found the negative publicity embarrassing, but he promptly wrote back to Mott to explain the background of the press conference and to state his conviction that the criticisms of Ovshinsky were unjustified and rooted in jealousy.¹³ Some physicists also urged Samuel Goudsmit, editor of the Physical Review, to withdraw the paper, but Goudsmit refused because it had passed peer review. He later told Ovshinsky that he had never experienced such craziness in his scientific editing career. Fritzsche reflected that the objectors must have recognized something important about the discovery. Otherwise, “it would have been dismissed by everyone.”

Many critics seized on Ovshinsky’s predictions in the press conference about the future implications of his invention, which they considered greatly exaggerated. It was to be a recurring theme of his career, for he could envision the successful consequences of his inventions without worrying about the practical problems that needed to be solved. In this case, the predictions would be borne out. Some also objected to the name Ovonics, which seemed self-aggrandizing and promotional.¹⁴ But what most threatened researchers at Bell Labs or General Electric was the suggestion that technology based on amorphous and disordered materials would replace technologies based on crystalline semiconductors, then already the basis of a billion-dollar industry.

To make matters even worse, the notion that silicon was obsolete temporarily depressed the stocks of companies like Texas Instruments, Motorola, General Electric, and RCA, and caused ECD’s stock to shoot up briefly to nearly triple its Friday price.¹⁵ The incident, which looked like manipulation, “created very bad feelings on Wall Street that he was not to be trusted, despite that what he was saying was correct,” noted the Stanford chemist John Ross. Of the companies that expressed criticism, the sharpest attacks came from Bell Labs, which claimed “nothing new has been disclosed” in the paper.¹⁶ Bell Labs staff members were not permitted to visit ECD or invite Ovshinsky to visit for some years. That fact was driven home when not long after the appearance of the Physical Review Letters and New York Times articles he was invited and then promptly uninvited to give a talk at Bell Labs.¹⁷ Having idealized scientists as among the purest and most civilized members of society, Ovshinsky could hardly believe the way they were behaving. “I was naïve,” he reflected. “I wasn’t expecting the rejection and hostility.”¹⁸

Despite the uproar and enmity, publishing in a leading physics journal attracted many researchers to the area of amorphous and disordered solids, bringing the field “out of the backwater,” as Morrel Cohen noted, and the article became widely cited. The previous scattered work, in Leningrad, Bucharest, and elsewhere, now coalesced into an exponentially growing, coherent international research area. Just for fun, the ECD physics consultant David Adler would often show in his talks a graph of the number of publications on amorphous and disordered materials in academic journals over time. The graph showed a steady rise after November 1968, when Ovshinsky’s first Physical Review Letters paper appeared.¹⁹ That increase had been anticipated earlier when Ovshinsky finally visited the Soviet Union in 1967 to speak in Leningrad about chalcogenide switching and memory effects at the fourth Symposium on Vitreous Chalcogenide Semiconductors. There he met Boris Kolomiets, the Russian researcher who had for over a decade been examining the optical and electrical properties of chalcogenides (see chapter 5).²⁰ After the talk, Kolomiets drew a slowly rising line on the chalkboard representing progress in the amorphous field until then and predicted that because of Ovshinsky’s work, the line would now rise almost vertically. The prediction was fulfilled; the next international meetings saw between five and six hundred people attending, and the trend continued.²¹

The historical impact of Ovshinsky’s 1968 paper in Physical Review Letters was amplified by a second paper in that journal in 1969, nicknamed CFO, for its co-authors Cohen, Fritzsche, and Ovshinsky.²² This paper dealt with the electrochemical properties of amorphous semiconductors in relation to the switching effect. Eventually as much cited as the 1968 paper, CFO contained an account of the surprisingly sharp activation energy observed for the threshold switch, known as the “mobility edge,” which, as Cohen put it, went “completely against everything that we were taught and that we were teaching at that point.”²³

Growth of the Physics Department

The physics department was well supported in the early 1970s by a grant to survey the optical, electrical, and thermal properties of chalcogenide glasses. The support came from ARPA, the Advanced Research Projects Agency (now DARPA) responsible for developing technologies for the military. Bringing in roughly $300,000 a year, the grant allowed the group to flourish for some years. Ovshinsky soon hired other PhD physicists, including Ed Fagen, trained at the University of Pittsburgh, and Simon Moss, an x-ray radiologist from MIT who set up a structures lab, complete with an x-ray diffractometer as well as scanning transmission electron microscopes capable of seeing the phase changes from amorphous to crystalline and back.

John de Neufville had learned about Ovshinsky in 1968 while studying materials science under Harvard’s David Turnbull. What most attracted him to ECD was Ovshinsky’s question, “If you came here, what would you like to do?”³¹ De Neufville outlined a systematic study of Ovshinsky’s amorphous and disordered materials using the technique of sputtering to lay down combinations of elements. “If you came here, you could do that,” Ovshinsky replied.³² In addition to this De Neufville also contributed to the development of the optical memory. He remembered his five years at ECD as one of the happiest and most productive times in his career.³³

Besides these PhD physicists, Ovshinsky also added a number of junior scientists Richard (Dick) Flasck, who joined the technical staff in 1970 with an undergraduate degree in physics from the University of Michigan, was attracted by ECD’s recently added education program to support further training for staff members. By studying nights at Oakland University in Rochester, Flasck soon gained a master’s degree in physics. He recalled this period at ECD as “an exciting time,” but he also sensed the strain many scientists felt in dealing with Ovshinsky’s claims. “They tried to be as flexible as possible without breaking their ethical backbone.” Part of the problem, Flasck noted, was that “Stan did not think like a standard physicist or chemist—not in numbers and not in principles, but in pictures. And sometimes that gives insight that you can’t get from standard mathematics.”

At the same time that Feinleib and Iwasa were developing an optical phase-change memory, others were developing an electrical phase-change memory that would eventually prove more important. The work involved collaboration between ECD and the newly formed Intel, an outgrowth of the relationship that began when Intel’s founders, Robert Noyce and Gordon Moore, who had been members of William Shockley’s original team in his Shockley Semiconductor Laboratory, visited ECD in the summer of 1968.³⁴ They were attracted by news of Ovshinsky’s switching and memory discoveries and were especially interested in his technique for etching “down to the atom.” One of them said, “Stan, do you want to make a fortune? Use this for masks” (for etching computer chips). But at the time, Ovshinsky recalled, “I wasn't interested in masks, so I screwed that one up.” Noyce and Moore visited ECD several more times, and in 1970, Moore became a co-author of one of ECD’s early phase-change memory papers.³⁵

By then, Ovshinsky had established a separate division run by Ron Neale to commercialize electrical phase-change memory. Neale’s work with D. L. Nelson and Gordon Moore resulted in an important new kind of integrated circuit. Initially made in limited quantities by ECD, the 256-bit memory (the RM-256) was later manufactured by Intel and called the “read-mostly memory” (RMM). It avoided certain serious problems of the existing “read only” (ROM) and “random access” (RAM) kinds of memory. In particular, while the inflexibility of the ROM prevented its data from being changed and the volatility of RAM allowed data to disappear during power interruptions, the RMM could be programmed, read, and reprogrammed repeatedly, and it retained its data unless intentionally altered.

While the 256-bit RMM was not commercially successful, its development was an important moment in the history of nonvolatile memory, for it was the first time that a phase-change memory switch was integrated with a silicon chip. A long arc of development followed from it, leading through ECD’s later improvements of its electronic phase-change memory (see chapter 10) and up to the present (2016), when a direct descendent of the RMM is entering production (see the epilogue).

As a growing research laboratory, ECD needed a machine shop to make experimental equipment and prototypes. In 1969, Harley Shaiken, a returning veteran from the storefront, set up the shop in a steel garage that Ovshinsky had erected in a corner of the parking lot to store the Bentley he had been given in England. Between working in the storefront for some months in 1960 and returning now to ECD for about five years, Shaiken had earned his journeyman’s card as a machinist.³⁶ He recalled working with Herb to design and produce various machines as “the most satisfying work years I’ve ever had.” He especially enjoyed the fact that everyone was treated equally at ECD, and he admired its “pliable” hierarchy, designed to draw out the best in people by rewarding them “more on merit and engagement than any place that I’ve worked or studied.”

As a fellow machinist Shaiken had a special rapport with Ovshinsky, who had not lost touch with the early experience that contributed to his discoveries. Shaiken recalled how Ovshinsky occasionally came by the shop, and they “would do things together.” On one occasion when Shaiken was making a particularly complicated and time-consuming cut on the lathe, Ovshinsky suggested he use a different approach. Knowing that Ovshinsky hadn’t worked on a lathe for many years, Shaiken said something like “that sounds good but I think maybe this would be a safer setup.” Ignoring the suggestion, Ovshinsky proceeded to remove the part and put raw stock in the lathe. “Let me show you,” he said. Turning up the speed much higher than Shaiken felt comfortable with, Ovshinsky cut the part in a way that the younger man considered “brilliant.”

Another addition to ECD with a shared history was the anarchist-metallurgist couple from Connecticut, Laurence and René Pellier. Laurence was a metallurgical engineer, and her husband René was a machinist who prepared the specimens for her to study with their electron microscope. Longtime friends of Iris’s family, the Pelliers worked for ECD as consultants while living in Connecticut. The attractive glass models of atoms and molecules that René made were on display in Ovshinsky’s downstairs office at home throughout his life, an embodiment of the way he himself visualized the structure of materials.

Figure 6.4 The Pelliers with Stan, Iris, and glass models.

Theories and Models of Chalcogenide Semiconductors

In the efforts to explain the Ovshinsky switching in terms of fundamental scientific principles, one controversial issue was whether it was a thermal or electronic effect. The debates polarized the physics group. (According to Dick Flasck, “there were almost fist fights.”) Ovshinsky felt that the mechanism behind the switching was electronic but was unable to explain why in scientific terms. Some believed that he preferred the electronic explanation simply because a heat-based phenomenon would be less reliable and so not suitable for commercialization, or because electronic switching would also be a more fundamental discovery. Feinleib, who on his first encounter with Ovshinsky had declined a challenge to prove the electronic explanation, would get annoyed when Ovshinsky argued by analogy for electronic switching, noting that the device’s behavior seemed similar to that of electronic phenomena, and different from thermal phenomena, where there is a time lag. Feinleib considered this way of arguing unscientific, but Fritzsche and Adler accepted it.

Not until the early 1980s was the controversy resolved. Melvin Shaw, a professor at Wayne State University, who joined ECD in 1970 as a consultant, showed with numerical simulation techniques that the Ovshinsky switching is first electronic but then also thermal. The initial event, Shaw explained, “the thing that breaks it down, is electronic. After it breaks down, it channels and gets hot.”³⁷ Shaw recalled that when he reported his findings, “Stan thought I was going to show that it was thermal. When I told him it’s electronic, he said ‘I knew it all along,’” a characteristic response of Ovshinsky’s that exasperated some of his colleagues.

Other ECD consultants worked to explain the Ovshinsky effect at the atomic level. An important contribution came from Marc Kastner, one of Fritzsche’s graduate students. Drawing on his chemistry background, Kastner noted that of the four outer p-electrons of the chalcogen atoms in Ovshinsky’s materials, only two are normally used in bonding. He suspected that the two remaining (normally non-bonding) electrons, which are called lone pairs, determine the special properties of Ovonic materials.³⁸ Ovshinsky was “enormously excited” when he heard about Kastner’s idea. “Once I knew it was a lone pair, it all came together,” he explained. “In my mind, I saw exactly the whole form of it.” Ovshinsky’s enhanced power of visualization, seeing the positions of the electrons in the atomic structure of his materials, gave him an alternative to scientific calculation. “You figure out the bonds,” he told Fritzsche. “In my mind I see them clearly, but I stutter when I try to describe them.”³⁹ Introducing the concept of the lone pairs was a crucial moment in the process of science catching up with Ovshinsky’s intuitions and then feeding back into their further development.

Fritzsche suggested that Kastner write a paper on his model under his own name, even though he was still a graduate student. That, Kastner said, “really helped my career get off to a start.” Later, after Kastner found a position at MIT, Ovshinsky invited him to join Dave Adler on a visit to ECD for a workshop where Mott presented a model of the role of dangling bonds in amorphous semiconductors. Listening to Fritzsche point out problems in the model, Kastner “suddenly had a glimmer of an idea of how to make this work based on the idea of lone pair semiconductors.” On the plane back to Boston, he explained his idea to Adler, “and by the time we landed, we had a draft of a paper.” A week or so later, at a conference in Williamsburg, the two of them huddled with Fritzsche to work out the consequences, and their new model was soon published.⁴⁰ Adler and his colleagues developed this model further into a fuller account that became generally accepted.⁴¹ The Adler model left some aspects of the Ovshinsky effect unexplained, however, and subsequent attempts have still not resolved all the issues. As the physicist Steve Hudgens recently observed, “The ‘deceptively simple’ two terminal devices that Stan described forty-four years ago still provide us with a fascinating mystery.”⁴²

Ovshinsky himself was struggling to convey his own conception of amorphous and disordered materials and decided that he needed to have actual physical models to show what he saw. The opportunity to create them arose in 1971 when a young Indian biophysicist, Krishna Sapru, moved to the area to be with her husband, who worked in Detroit. She had a research fellowship at Wayne State University and planned to work on DNA replication, but when her professor moved to California, she applied for a position at ECD. She felt an immediate rapport with Ovshinsky, who showed interest in her work on DNA, and she was especially impressed when he remarked, “There's no distinction between physics, chemistry, and biology.” It was a view that few scientists would have expressed at that time but which reflected Ovshinsky’s consistent disregard for disciplinary boundaries.

Sapru’s first assignment was to create “some real models” of the amorphous and disordered materials. After studying the electronic structures, she and her young daughter sat on her patio and constructed the models using several hundred small Styrofoam balls, which they colored and connected with colored pipe cleaners. She told her daughter, “We will pretend that yellow is tellurium, brown is sulfur, green is germanium, and so on.” Over a weekend, working with formulas that Ovshinsky supplied, they assembled thirty to fifty models of the switching and memory materials.

As Sapru worked on the models, she developed a feel for Ovshinsky’s perspective on the “personality” of each atom and molecule based on their number of protons and neutrons, and particularly the distribution of the electrons in their quantum-mechanical orbitals.⁴³ She began to recognize that atoms always try to form the strongest bond. For example, when they combine to make lithium fluoride they are “two really happy atoms. And that’s ionic bonding, when one electron goes mostly over to the other one, whereas in the case of covalent bonding, they share electrons.” In chalcogenides, the lone pairs in the p orbital (represented by two pipe cleaners) “are not happy in the sense that they are dangling bonds, not paired up with anything,” and so “anxious to make a connection.” This anthropomorphic and visual kind of thinking, with which Sapru became adept in the course of her model-building work, was more like the way chemists, rather than physicists, typically think about atoms and molecules.

Figure 6.6 Ovshinsky lecturing with Styrofoam models.

Ovshinsky was excited when he saw the models that Sapru had made; he told her, “We are writing a paper!”⁴⁴ He would keep her models in his office bookcase for the rest of his life, using them in presentations, and indeed whenever possible, in his efforts to explain his insights about the switching and memory materials to his befuddled listeners. In watching him talk about his work while manipulating his models, Joi Ito (whose parents Masat Izu and Momoko Ito were ECD employees) recalled, “Stan would talk about science in a sort of artistic way.” He would be “holding these models of Styrofoam and pipe cleaners, shaking them and saying, ‘See these dangling bonds. There’s energy here.’” Ito added, “It took Nobel laureates to translate what Stan was feeling.”

OIS and OMI

As ECD grew and developed new applications for amorphous and disordered materials, Ovshinsky found that the demands of managing its business made it hard for him to stay as closely involved as he wanted in the work of research and development. To get help, in November 1969 he asked the lawyer and accountant Keith Cunningham, then a senior executive at the accounting firm Touche, Ross, and Co. to become ECD’s president and chief executive officer while Ovshinsky remained its chief operating officer.⁴⁵ As Chet Kamin, who would later become Ovshinsky’s attorney and adviser, explained, “Stan brought Keith in because he knew he wasn’t good at financial stuff. He wanted somebody that he could lean on to basically run the business part of the company so that he could spend more time and energy on research.” By early 1971, Cunningham had raised enough money to set up two subsidiary companies aimed at commercializing ECD’s technologies: Ovonic Memories, Inc. (OMI) in February 1971, and Ovonic Imaging Systems, Inc. (OIS) in April 1971, both located in Southern California.⁴⁶ Even with the funding Cunningham had arranged, the financial basis of the two companies seems to have been precarious.⁴⁷

The aim of OIS was to commercialize instant imaging and non-silver films for microfiche records.⁴⁸ This technology seems to have been the first ECD application of amorphous materials beyond switching. The idea itself was not new: using photoconductive amorphous materials to copy documents was already the basis of xerography. In the Ovonic system, however, the image was reduced and transferred to a microfiche card, which could not only be read but also revised before being stored again. It was thus the analog equivalent of a digital rewritable memory, an ingenious technological advance. The key to making the non-silver film work not only for copying images but also for writing and rewriting text was activating it with an electron beam, an idea proposed by the longtime ECD scientist Peter Klose.⁴⁹

To help with manufacturing the film, Cunningham offered a position to Herb Ovshinsky, who was Cunningham’s friend and neighbor. In Ovshinsky’s earlier years in Akron, New Britain, and Detroit, he had often relied on his brother in designing and building machines; after joining ECD in 1971, Herb resumed this role. Working with his colleague, Al Adominis, Herb built a machine for roll-coating the new instant imaging film. It was ECD’s first use of a continuous production method, the kind that would later be used for making thin-film solar cells (see chapter 8).

To work with Herb on the instant imaging technology, in 1972 Ovshinsky hired the young chemist Masatsugu (Masat) Izu, who had taken his doctorate at Kyoto University under the great theoretical chemist and later Nobel laureate Kenichi Fukui. Izu was then a postdoc at the University of Waterloo in Canada; to recruit him, Ovshinsky offered to double his postdoc salary. Izu decided to move to Michigan, along with his wife Momoko and their two small children, Joichi (Joi) and Mizuko (Mimi). Ovshinsky also hired the even younger chemist David Strand, who came to ECD at this time with a bachelor’s degree in chemistry from Michigan State University. Strand, who over the following decades would become a mainstay of ECD’s research programs, helped develop the coating of the film, a tellurium-based organo-metallic compound, which when exposed to light formed a latent image that became visible when heated and could be printed on paper or shown on a display.⁵⁰

Unlike other microfilm systems, the Micro-Ovonic Fiche (MOF) allowed users to revise and save the stored information. It was therefore considered a revolutionary technology, and ECD, as the OIS parent company, entered an agreement with 3M to commercialize it. Despite this promising beginning, however, the timing was wrong. The venture with 3M never materialized, and OIS failed because a market for the microfiche retrieval technology could not be found.⁵¹ In any case, the MOF analog imaging system, however sophisticated, would eventually prove unable to compete with the emerging electronic digital technologies.

The other company, OMI, also soon failed. It had been established to commercialize the optical memory developed by Feinleib and Iwasa. It aimed to manufacture and market a disc drive for IBM computers, promising a prototype with a capacity of 64 billion bits by spring 1972. Named the 4440, it would provide ten times the storage of IBM’s popular 3330 with the same average access time of 30 milliseconds.⁵² As with the microfiche, however, there was no market for the technology—in this case because computers did not yet require that much storage.⁵³

Ovshinsky felt the failure of OIS and OMI also owed much to Cunningham’s mismanagement and the inflexibility that made him a bad negotiator.⁵⁴ In any case, Cunningham’s commitment of so much of ECD’s resources to the two companies was clearly, as John de Neufville said, “a very high-risk approach,” and their failure “almost broke the back of ECD.” Over the next three decades Ovshinsky would continue trying to find people to relieve him of more routine management responsibilities, but he would never again delegate so much authority.⁵⁵

After Cunningham left the company in late 1974, ECD’s fortunes continued to decline, and it was necessary to reduce the staff to about twenty-five.⁵⁶ The few remaining employees were called on to juggle several tasks. Dick Flasck, one of those few, “more or less inherited the materials research lab/physics lab, the analytical lab, the bomb room and a number of other departments.” In addition to his research, Flasck was sent on business trips so often and on such short notice that, he reported, “I kept a packed suitcase in the trunk of my car. He had me flying close to 200,000 miles a year.” Flasck also recalled how Ovshinsky, refusing to be daunted by the downturn, continued to give one-hour talks every other day expounding his ideas. Sometimes, when there were no consultants visiting, Flasck alone made up the entire audience. Like many others, he was usually baffled by Ovshinsky’s explanations.

In the midst of ECD’s struggles to survive, Ovshinsky got some help for developing his electrical phase-change memory. Late in 1973, he and Iris came to see Robert Johnson, then senior vice president of engineering at Burroughs Corporation in Detroit, a business equipment company that had become an important computer manufacturer. They told Johnson they needed $100,000 quickly.⁵⁷ Johnson recalled having met Ovshinsky, who many considered “a wild-eyed inventor.” Ovshinsky “didn't sound crazy” to him, although it was immediately clear that he “wasn't particularly good at explaining things.” Johnson was able to arrange a $100,000 advance on a contract for licensing ECD’s read-mostly memory (RMM) to Burroughs.⁵⁸ Collaborative work between ECD and Burroughs continued for several years, and in 1978 resulted in an RMM memory array of 1,024 bits, not only much bigger and better-performing than the 256-bit array but also faster and with a somewhat lower programming current.⁵⁹ A few years later, Johnson would join ECD’s staff (see chapter 10).

Figure 6.7 The ECD-Burroughs 1,024-bit integrated memory array.

The Beginning of ECD’s Photovoltaic Program

Of all the programs ECD undertook in its early years, developing the means to produce cheap and efficient solar power was the most important for advancing Ovshinsky’s goal to replace fossil fuels. The enormous amount of energy radiated by the sun made it a promising source (in one hour the earth receives enough to meet all human needs for a year). But the photovoltaic program did not begin to make real progress until the later 1970s.

Understanding the problems ECD had to solve requires some understanding of the technology. A conventional solar cell is basically a diode, a one-way electron valve that converts light into electricity by means of the photoelectric effect, in which light dislodges electrons. Photovoltaic material is arranged in layers whose electronic structure has been altered (doped) by adding small amounts of other elements (boron and phosphorus) to create P- and N-type materials.⁶⁰ When these layers are brought in contact, an electric field forms at the P-N junction, causing dislodged electrons to move toward the N-layer, while the holes left behind behave like positive charges and move toward the P-layer, building up a voltage. Inserting the device in a circuit allows current to flow and do work.

Up to this time, solar cells had all been made of crystalline silicon because it was believed that only crystalline material could be used to make P- and N-type material. Such cells, however, were costly, heavy, rigid, and fragile. Ovshinsky wanted to make amorphous thin-film cells that would be cheaper, light, flexible, and more robust. These had three layers, with P-type and N-type material separated by a thicker, undoped layer (an intrinsic, or I-layer) through which the electric field passes. Again, the dislodged electrons (now mostly from the I-layer) flow toward the N-layer and the holes again move to the P-layer. The difficult technical problem of making a thin-film solar cell was thus essentially that of using amorphous material to make an efficient PIN structure.

Ovshinsky knew that he would have a tremendous patent advantage in developing thin-film solar cells if ECD could use his amorphous chalcogenide alloys to create such PIN structures. That, however, depended on doping them to make P- and N-type material. But when his consultants analyzed the atomic structure of Ovshinsky’s materials, they concluded that such doping would not work with them.

Further progress required switching to a different amorphous material, whose photovoltaic potential had already been demonstrated. In 1975, Walter Spear and his student Peter LeComber at the University of Dundee published a paper describing a process for making hydrogenated amorphous silicon with high photoconductivity. Using a new method (the glow-discharge plasma decomposition of silane gas pioneered by R. C. Chittick and colleagues), they found that amorphous silicon prepared with hydrogen could be doped like crystalline silicon. The addition of hydrogen furthermore neutralized the defects and dangling (unsatisfied) bonds that are numerous in amorphous silicon and would otherwise capture electrons and reduce the current produced in the deposited film.⁶¹

These results suggested that ECD should shift its photovoltaic research to amorphous silicon.⁶² But Ovshinsky was not yet ready to give up on his chalcogenides. With new funding from United Nuclear Corporation between August 1976 and August 1977, he turned to what he would call “chemical modification,” mixing larger quantities of elements from other groups with the chalcogenides to try to get the same effect as doping. Presenting this work at a meeting in Edinburgh in June 1977, Ovshinsky announced that it had increased conductivity by up to nine orders of magnitude.⁶³ He considered this feat of “atomic engineering” as “one of the most powerful things I’ve done,” but it did not solve the problem of making chalcogenide solar cells.⁶⁴

Ovshinsky was at last ready to turn to hydrogenated amorphous silicon. Fritzsche recalls persuading him to make the change at the March 1977 American Physical Society meeting in San Diego. (By then, David Carlson at RCA had announced making the first solar cells from amorphous silicon.)⁶⁵ Vincent Canella, a physics professor at Wayne State University who joined ECD in 1976, recalled, “Stan came back and said, ‘We should do silicon.’” Ovshinsky then hired LeComber’s student Arun Madan and directed the physics group to experiment with plasma-deposited amorphous silicon and germanium.

There was immediate tension however between Ovshinsky and the young Indian-born physicist. Madan wanted to do detailed, small-scale research aimed at fully understanding the physics, while Ovshinsky was eager to move forward and make complete cells that could be quickly brought into production. He therefore created a second photovoltaic group headed by Masat Izu to build and test cells. But the reorganization did not resolve the tensions, because Ovshinsky was still intent on developing a unique approach he could call his own. Having agreed to work with amorphous silicon, he wanted to substitute fluorine for hydrogen. In theory, this made a certain amount of sense, for like hydrogen, fluorine’s outer shell lacks one electron, and Ovshinsky believed it would form a stronger bond with silicon. Fluorine would thus not only neutralize the dangling bonds but also give ECD the proprietary advantage he had failed to gain by using chalcogenides.⁶⁶

But in practice, substituting fluorine was a failure. “It never really worked out, but we spent a lot of time finding out,” Vin Canella said. The highly reactive fluorine would not only join with the dangling bonds but would also break desirable bonds, and as an etchant it would often break down the film as it was deposited. Stan’s insistence on trying to use fluorine infuriated Madan. Dick Flasck recalled, “Arun was just frustrated as hell. His attitude was, ‘I don't want to waste my time trying to find some way around the situation by not using hydrogen and using something else that's not quite as good, especially if, from the technical standpoint, there's no good reason to do that.’” Eventually, after these time-consuming efforts to develop an alternative, Ovshinsky went ahead with using hydrogenated amorphous silicon, and ECD’s photovoltaic program became highly successful. But to the end of his efforts to make better and cheaper solar cells, he never completely gave up on using fluorine (see chapter 12).⁶⁷

A Bold Concept and Major New Funding

Ovshinsky did not work directly on studying or designing the thin-film solar cells, but by the late 1970s he was very much engaged with producing them quickly and cheaply because he knew that was the key to replacing fossil fuels with solar power. Solar panels were then produced slowly and expensively, one at a time. Ovshinsky instead imagined a machine for manufacturing thin, flexible solar panels roll-to-roll, or as he liked to say, “by the mile,” like film or newsprint. The basic concept for such production was not new (ECD had already made imaging film that way), but applying it to making solar cells was radically new. Ovshinsky envisioned a machine that would use plasma deposition to produce the thin layers of hydrogenated amorphous silicon on a moving stainless steel substrate, producing miles of solar panel. Based on only recently available concepts, the plan presented daunting technical problems.

Ovshinsky called a number of meetings with his scientists to explain his roll-to-roll concept. Fritzsche recalled those present shaking their heads with skepticism but also not offering much resistance to the concept, whose advantages were obvious: the thin and flexible panels could cover large areas, promising to bring down the price of solar energy. But the proposal was such a huge leap from the small experimental samples of less than a square centimeter the researchers had been working on that, as Fritzsche recalled, “it left us speechless.” To the scientists, the problems involved in going from these small experimental cells to continuous rolls appeared overwhelming. Like his attempts to use chalcogenides and fluorine, it was yet another instance of Ovshinsky’s asking for what seemed impossible. But in this case, his long-range vision would be vindicated, and the problems of building the roll-to-roll machine would be solved, as detailed in chapter 8.

Ovshinsky’s search for support to fund this ambitious scheme succeeded, and succeeded in a way that dramatically increased the scope of ECD’s operations. The support came from ARCO (Atlantic Richfield), then one of the seven or eight largest American oil and chemical companies. Such a partner might at first seem surprising, but in the later 1970s many major oil companies were developing alternative energy programs because the 1973 Arab oil embargo had made foreign supplies seem unreliable and because scientists had projected that the world’s oil reserves would be depleted by the end of the century.⁶⁸ Indeed, this was to be the first of several such partnerships between ECD and an oil company. The energy crisis and economic slowdown of the 1970s encouraged an interest in new technologies that favored ECD’s growth.

ARCO was already involved in solar energy, having in 1977 acquired Solar Technology International, which became ARCO Solar. It was (and, as its successor, SolarWorld, still is) primarily involved in making crystalline solar cells, but at the time it was also looking at other materials. The physicist Richard Blieden, who would later join ECD’s staff, was working in ARCO’s R&D program at the time he and Ovshinsky met in 1978 at the American Solar Energy Society conference in Boulder. Blieden had read Ovshinsky’s 1968 Physical Review Letters article when he was teaching physics at Stony Brook and continued to hear about him and ECD later, when he directed solar programs at the National Science Foundation, the US Energy Research and Development Administration, and the Department of Energy. Ovshinsky’s presentation at Boulder piqued Blieden’s interest, and persuaded him to look more closely at what was happening at ECD. Blieden’s visit after the Boulder meeting was followed by another from his boss, Robert Chambers, the head of R&D at ARCO, and then by negotiations resulting in May 1979 in a $3.3 million grant for ECD’s photovoltaic research.⁶⁹

Less than a year after the initial ARCO grant, the opportunity arose for a much larger one. This time the connection came from Ovshinsky’s old union colleague, Jack Conway, then on the board of ARCO Solar.⁷⁰ As an experienced activist and administrator, Conway appreciated the social implications of Ovshinsky’s vision for solar energy, and he brought ARCO president Thornton Bradshaw to see him. A highly successful manager, Bradshaw believed in the social responsibility of corporations. Ovshinsky recalled him as “an American business executive who was interested in the great problems of the country and the world, a wonderful guy.” Excited by Ovshinsky’s concept for making solar panels by the mile, Bradshaw told him ARCO would fund ECD’s energy research for three years. He could not guarantee support beyond that time because he was planning to leave ARCO then, but for three years ECD would have unrestricted scope for developing alternative energy systems.

Now Ovshinsky had to decide how much money to request from ARCO. He turned to Nancy Bacon, a successful accountant handling ECD’s accounts at Deloitte Touche whom he had persuaded in 1976 to join ECD as chief financial officer. Bacon was to play a critical role in all of the company’s financial transactions. She remembered Ovshinsky coming to her office after his meeting with Bradshaw and asking, “What do you think we should do?” She suggested asking for $15 million, which at first shocked him. But after much consideration of how they would structure the proposal, “Stan decided to ask them for $25 million,” Bacon recalled, “and made the thing stick.” The second agreement, in January 1980, provided not only an additional $6 million for the research and development of solar energy but also $19 million for other alternatives to fossil fuel, including thermoelectric and hydrogen.

The ARCO grants were a turning point in the story of ECD’s ascendance as a major energy laboratory. Ovshinsky’s boldness in envisioning roll-to-roll production and then in getting such a large increase in funding led to dramatic expansion. As Dick Blieden said, “For the first time at ECD they had enough money to build up the laboratory facilities they needed, to hire the patent attorneys to generate the IP, and basically to do all of the things Stan wanted to do to explore the opportunities in these new materials. ARCO really was a lynchpin.”

Notes

11. The reasons for the controversy have been perceptively analyzed from a science studies point of view by Michael Gibbons and Philip King, “The Development of Ovonic Switches: A Case Study of a Scientific Controversy,” Science Studies 2, no. 4 (October 1972): 295–309. Gibbons and King describe the controversy as a striking example of the hostile response likely to meet an outsider who challenges scientific norms: “Ovshinsky had mixed the exchange system of the entrepreneurial-industrial community with that of science, and in so doing violated the norm of disinterestedness.” Yet, as they note, researchers in large industrial laboratories, who are not seen as outsiders, mix those systems with impunity. See also Philip M. Boffey, “Ovshinsky: Promoter or Persecuted Genius,” Science 165 (August 15, 1969): 673–677.

16. See the editorial in Electronics Review 41, no. 24 (November 25, 1968). This dismissal was based on earlier work at Bell Labs that had been reported in A. David Pearson, W. R. Northhover, Jacob F. Dewald, and W. F. Peck, Jr., “Chemical, Physical, and Electrical Properties of Some Unusual Inorganic Glasses,” in Advances in Glass Technology, Technical Papers of the VI International Congress on Glass, Washington, DC, July 8–14, 1962 (New York: Plenum Press, 1962), 357–365. Working with a compound of tellurium, arsenic, and iodine, Pearson’s group had observed the current-voltage characteristic that later marked Ovshinsky’s memory switch, but they did not report observing threshold switching and did not seem to recognize the important implications of their findings. Moreover, the material they used was unstable and its behavior irreproducible; Bell Labs therefore discontinued the work. Ovshinsky learned of Pearson’s work only after he published his own discoveries. Fritzsche investigated, and in a letter concluded that Pearson’s “memory device might be considered a close miss. He might have discovered the right one if he had worked on it for a few years more.” Fritzsche, letter to Ovshinsky, February 1969. Ovshinsky papers). For Ovshinsky’s account, see S. R. Ovshinsky, “An Introduction to Ovonic Research,” Journal of Non-Crystalline Solids, 2 (1970): 99–106, 100–101. For Pearson’s account, see A. David Pearson, “Memory and Switching in Semiconducting Glasses: A Review,” Journal of Non-Crystalline Solids, 2 (1970): 1–15.

22. Morrel H. Cohen, H. Fritzsche, and S. R. Ovshinsky, “Simple Band Model for Amorphous Semiconductor Alloys,” Physical Review Letters 22 (May 19, 1969): 1065–1068. Later in 1969, there was a memorable symposium in New York City on semiconductor effects in amorphous solids. John de Neufville said that Ovshinsky and his colleagues created the meeting to “assert the scientific legitimacy of what they were doing.” Afterward, he gave a fancy dinner for his team at the Plaza Hotel. The tradition of an Ovshinsky dinner continued well into the 1980s at the annual American Physical Society meetings held in March, always devoted to solid-state physics. Richard Zallen later deposited in the Virginia Tech archives records documenting the March meetings he attended. The Ovshinsky dinners, which were arranged mainly by David Adler, consisted of 15 to 20 physicists (always including Adler, Fritzsche, and Cohen), and took place “at the best restaurant in the city in which the March meeting was being held,” Zallen said.

24. American industrial research laboratories divide between those established by individual inventors to support their own inventive work and create new industries and the typically much larger laboratories established and supported by corporations to advance their goals. ECD fits the former model, for which Edison’s Menlo Park (1876) was the prototype, aimed not at supporting an existing industry but at creating new industries. See Paul Israel, Edison: A Life of Invention (New York: John Wiley & Sons, 1998). Labs that fit the other model include those established by General Electric (1900), DuPont (1902), Eastman Kodak (1913), and Bell Labs (1925). Those interested in the rich existing literature on early industrial research in American can begin with Leonard Reich, The Making of American Industrial Research: Science and Business at GE and Bell, 1876–1926 (Cambridge: Cambridge University Press, 2002); Thomas P. Hughes, American Genesis: A Century of Invention and Technological, 1870–1970 (New York: Viking, 1989); Jon Gertner, The Idea Factory: Bell Labs and the Great Age of American Innovation (New York: Penguin, 2012); David A. Hounshell and John Kenly Smith, Science and Corporate Strategy: Du Pont R and D, 1902–1980 (Cambridge: Cambridge University Press, 1988); and Lillian Hoddeson, “The Emergence of Basic Research in the Bell Telephone System, 1875–1915, Technology and Culture 22, no. 3 (July 1981): 512–544.

49. Klose was recruited by Hellmut Fritzsche, who had known him as an outstanding technician in Germany and helped him come to Purdue, where he earned an MS in solid-state physics. Klose started in the storefront in 1964, where he worked on making numerous switching alloys and was soon traveling widely to represent the company. His technical skills were extremely important in the company’s early years, when he worked to turn Ovshinsky’s ideas into functional devices, e.g., the first threshold switch that could actually be used as an electronic component. Serving as project manager for the microfiche camera, he went with the project to California. See Peter Klose, “High- and Low-Lights from My Time at Energy Conversion Devices (1964–1983)” in Reminiscences and Appreciations: Presented to Stanford R. Ovshinsky on the Occasion of his 80th Birthday, electronic manuscript, Ovshinsky papers.

61. W. Spear and P. LeComber, Solid State Communications 17 (1975): 1193–1196. (Ironically, as Marc Kastner noted, Spear himself “refused to accept the idea that his films had hydrogen in them.”) Although Spear and LeComber are generally given the credit for this discovery of the high photoconductivity of hydrogenated amorphous silicon, the first to see it were R. C. Chittick and his colleagues at the Standard Telecommunication Laboratory. See R. C. Chittick, J. H. Alexander, and H. F. Sterling, “The Preparation and Properties of Amorphous Silicon,” Journal of the Electrochemical Society 116, no. 1 (1969): 77–81. But Chittick’s company then chose to focus on crystalline silicon, so he and his team discontinued their amorphous silicon research and transferred their equipment and know-how to Spear. See also the retrospect article of Chittick and Sterling, “Glow Discharge Deposition of Amorphous Semiconductors: The Early Years,” in David Adler and Hellmut Fritzsche (eds.), Tetrahedrally-Bonded Amorphous Semiconductors (New York: Plenum Press, 1985), 1–10.

6 The Birth of ECD: An Invention Factory (1965–1979)

Physical Review Letters and the New York Times

Creating a Research Laboratory

Optical Phase-Change Memory