After the IC: Jack Kilby’s Solar Misadventure

The Nobel Prize winner spent a decade trying to create a household solar energy system—and failed

12 min read
Photo: Texas Instruments records/DeGolyer Library/Southern Methodist University
Photo: Texas Instruments records/DeGolyer Library/Southern Methodist University

In the summer of 1958, a new employee at Texas Instruments, Jack St. Clair Kilby, had not yet accrued enough time off to take a company-wide vacation. Alone in the lab, Kilby pondered the then-current “tyranny of numbers” problem: the time-consuming and imperfect process of soldering connections between very large numbers of discrete solid-state components. Within a few short weeks, Kilby came up with a solution—the integrated circuit. This revolutionary invention, for which Kilby usually shares credit with Robert Noyce, launched the entire field of modern microelectronics and more than a half century of staggering technological progress.

Kilby’s integrated circuit helped make his employer a dominant chip manufacturer for much of the Cold War era. And over the years, Texas Instruments has worked hard—quite justifiably—to associate Kilby with the firm and to use his invention and subsequent honors to enhance TI’s reputation for high-tech innovation. As a result, however, few people are aware that Kilby was a full-time employee of TI for just 12 years. In 1970, he took a leave of absence to become a consultant and independent inventor.

Kilby’s post-TI career resulted in a number of inventions. But it was defined by one project: a household solar-energy and fuel-cell system that could generate electricity around the clock. Kilby had high hopes for his idea; he believed it would be just as revolutionary and society changing as the integrated circuit, and he worked for a decade to make it happen. To Kilby’s great disappointment, his technology didn’t catch on. Kilby is rightly better remembered for the integrated circuit—but that doesn’t mean we should forget his deep involvement in an equally important and challenging technology.

When Kilby left TI, he didn’t completely cut ties. He still received a generous retainer, in return for which he represented the company on various national committees and filed monthly reports on industry developments and his own projects. Formally, TI had right of first refusal to license Kilby’s inventions. Informally, he remained a frequent and generally beloved visitor to his old firm.

By leaving TI, Kilby gained the time and autonomy to become a classic American inventor. And he did in fact file a string of patent applications through the 1970s, though none came near the inventiveness or commercial success of the integrated circuit—or the later work he did before leaving TI on the handheld calculator. Kilby’s inventions in his first independent years included a reliable flashlight, a theft-proof cash register, and an electronic checkwriter.

He also invested significant energy, time, and money in developing a personalized, interactive electronic teaching machine. Although he attempted to sell the idea to numerous companies, including McGraw-Hill, Sony, Westinghouse, and Xerox, as well as TI, nobody agreed to commercialize it. Interestingly, his device bears strong similarities in aesthetics, operation, and objectives to TI’s later, wildly successful Speak & Spell line of computers for children. But while the Speak & Spell broke new ground in digital signal processing and electronic voice synthesis, Kilby’s machine did not.

Kilby was also interested in higher education. Throughout the 1970s, he worked with electrical engineers at Texas A&M University and with Jay Lathrop, a former TI colleague at ­Clemson University, in South Carolina, to develop new ways to teach undergraduates the practice of integrated circuit design and fabrication so as to better ready them for industry.

In late 1973, however, Kilby and Lathrop’s correspondence suddenly turned away from discussions of syllabi, textbooks, and course requirements. The Organization of Arab Petroleum Exporting Countries embargo on oil exports and soaring oil prices dominated the headlines, and Kilby proposed that he and Lathrop think about how to apply their expertise in silicon to lessen the world economy’s dependence on oil. (At the time, the United States used oil not only to heat homes and to refine into gasoline but also to generate a sizable fraction of its electricity.) ­Lathrop readily agreed. In taking this detour, the two men displayed a remarkable spirit for reinvention: They both jumped into new careers and would reinvent themselves as students—and eventually experts—in the physics, chemistry, and economics of solar energy.

Within a few months Kilby and ­Lathrop had acquired a third collaborator, A&M’s W. Arthur “Skip” Porter, and the three came up with a basic idea for a residential solar energy system. At its core, it consisted of a set of very shallow tubes partially filled with fluid—hydrogen bromide in most of the configurations they discussed. Lining the bottom of each tube were thousands of silicon spheres the size of BB-gun pellets, embedded in a translucent matrix. Solar photons would strike those silicon pellets and knock off electrons, creating an electric potential that would break down the fluid around them into hydrogen and bromine. These constituents would be siphoned off and stored until electricity was needed, at which point they would be recombined in a fuel cell and then circulated back into the tubes.

/image/MjgxMjQ1Mg.jpegCells of Silicon: The heart of the solar generation system that Kilby and his colleagues developed consisted of small pieces of silicon embedded in translucent material.Photo: Texas Instruments records/DeGolyer Library/Southern Methodist University

Kilby, Lathrop, and Porter weren’t alone in entering the solar field in the wake of the oil embargo. The 1970s saw a boom in solar energy. From 1974 to 1979, federal funding for renewable energy development rose from US $32 million per year to $1.36 billion. Private-sector interest mirrored this growth. By 1976, Solar Engineering Magazine listed 88 companies with some involvement in solar energy that were traded on the New York Stock Exchange. Those large companies were joined by small firms, academic laboratories, and amateur enthusiasts. Oil, electronics, and
military-industrial firms were all heavily involved in solar in the 1970s. Texas Instruments, as an oil-field instrumentation company, semiconductor manufacturer, and defense supplier, had roots in all three of those industries.

Kilby’s system differed from most other solar technologies in that it answered the eternal question, What do you do when the sun’s not shining? Because the system stored the chemical constituents separately and combined them on demand, the moment when photons struck could be temporally separated from the moment when the system created electricity. Kilby and his colleagues also believed that their system would be considerably easier to manufacture than other silicon-based solar energy schemes. At the time, fabricating large sheets of single-crystal silicon for photovoltaic cells was difficult and expensive. But Kilby’s liquid-fuel system used silicon pellets with less-exacting qualities. The tiny spheres could be produced using a method invented in the late 18th century to make shotgun pellets. In this fiendishly simple technique, molten lead (or silicon) is poured from the top of a “shot tower,” forming spherical droplets that cool and solidify as they fall.

By the spring of 1975, Kilby was ready to present the idea to TI. For once, he got a positive response. TI agreed to shepherd Kilby, Lathrop, and Porter’s invention through the U.S. patent office and to seek patents in dozens of other countries, too. The firm also provided Kilby and his colleagues with generous research funding, and it formed its own development team, code-named Project Illinois after ­Kilby’s alma mater. As the use of a code name implies, TI wasn’t yet ready to go public with the project. This way, the firm could take advantage of Kilby’s leave of absence. He could attend solar power conferences and approach federal funding agencies under his own name without revealing that he was also acting on TI’s behalf.

TI was cautious in part because Project Illinois was an unproven technology with significant risks. It was unclear how efficient the system would be. What’s more, hydrogen bromide is extremely toxic and corrosive—a rather unlikely chemical for homeowners to keep on their roofs in large volumes. And one of the key components of the system was a fuel cell, a notoriously finicky technology.

Closed-Loop Solar Energy Conversion System

A more fundamental problem, however, lay in the project’s business model. TI expected it would first need to sell the system at high cost to very wealthy homeowners in order to generate enough revenue, knowledge, and economies of scale to bring the price down for the middle-class market. But there was little evidence that high-end homeowners would embrace costly and arguably ugly solar panels on their property.

If only there was some entity with very deep pockets that could subsidize development of the technology or provide a captive market to help drive down the cost of deployment—or both. Of course, such an entity existed: the government. Indeed, the U.S. military had already played that role in the development and commercialization of TI’s integrated circuits.

Conveniently, the U.S. government was investing heavily in renewable energy. In 1973, President Richard Nixon had announced Project Independence, an “Apollo program” for energy that aimed to make the United States energy self-sufficient by 1980. TI executives believed they could find federal partners who would help the company bring solar energy to the masses.

TI executives and Kilby began looking up old friends in government to see who might contribute to Project Illinois. The answer turned out to be the newly formed Department of Energy (DOE). In January 1978, after several years of small-scale research, George Heilmeier, a TI vice president who had recently led the Defense Advanced Research Projects Agency (DARPA), sat down with John Deutch, the DOE’s director of energy research, to talk about Project Illinois. In a memo, Heilmeier later reported that Deutch said the DOE’s expertise in solar energy was limited, and that he would therefore be willing to fund the project. Not long after, TI entered a four-year, $18 million cost-sharing arrangement with the DOE to accelerate development of what was now more openly being called the Texas Instruments Solar Energy System (TISES). With that funding, Kilby and colleagues built a prototype system, boosted the efficiency of the solar-­electric conversion process, increased the capacity of the fuel cells, and extended the lifetime of the components.

Kilby and TI still had to find a market and customers for the system. Here they were optimistic because the federal government and some states had started offering tax credits for the adoption of residential solar equipment. But they made some critical missteps.

From the start of Project Illinois, TI had borrowed R&D templates from the military-­industrial complex. Kilby and his colleagues relied on networks of contacts who moved freely between the military and its suppliers. TI insisted that its contract with the DOE contain intellectual-­property language lifted from TI’s Pentagon contracts, and Deutch explicitly modeled the TI grant on the high-risk, high-reward funding approach favored by DARPA. TI even tried to figure out its strategy for developing TISES using a nuclear war–­gaming technique called scenario planning. Drawing on military-industrial practices made sense, of course, since TI had a great deal of experience with the military. Moreover, the company’s success with integrated circuits showed that it could, quite profitably, spin off large civilian markets from its military work.

/image/MjgxMjUzNw.jpegPhoton Power: These photos show the front and back of a TISES prototype. According to a document in the Southern Methodist University archive, this system is the “TISES ‘C’ Prototype Module.”Photos: Texas Instruments records/DeGolyer Library/Southern Methodist University

But that approach had a less salutary feature: a reluctance to trust solar power experts who did not have strong ties to the military. Despite Deutch’s claim, the DOE actually had a fair number of people working on solar energy. In fact, it had an entire Solar Energy Research Institute (now the National Renewable Energy Laboratory), headed by Earth Day organizer Denis Hayes. But Kilby and his TI colleagues found little traction for their idea among the group, and they were dismissive of points raised by SERI personnel in turn. When Ronald Reagan’s presidential transition team wrote to Kilby after the 1980 election to ask which DOE solar energy officials should be kept on by the new administration, Kilby’s blanket reply was, “All of the people who have been involved with the solar program should go…. In general it is my feeling that the level of competence of the people within DoE is a couple of notches below those in DoD.”

Instead, Kilby sought out military colleagues. In late 1978, for instance, he pushed for Richard D. Alberts, the U.S. Air Force grant officer who funded TI work on integrated circuits, to take over the DOE’s photovoltaic section. A year later he asked Alberts to introduce him to someone in the Air Force who would consider adopting TISES as the energy source used at MX missile silos. That application never panned out, but Kilby’s preference for military experts over civilian solar energy advocates such as Hayes is telling.

Ironically, because of their military-industrial bias, Kilby and TI were looking for allies for TISES among a group of people who were unlikely to support it. At the time, the U.S. national security establishment’s imagined energy future emphasized nuclear power rather than solar, which was considered an immature technology redolent of the counterculture. There’s no trace in Kilby’s papers that he ever considered partnering with those “countercultural” solar advocates, who might have been able champions for TISES.

One thing such advocates might have offered was a persuasive environmental or social vision in which TISES would play an important part. Lacking that, Kilby and TI were left to argue for the system on narrow economic and national security grounds. Those grounds were persuasive so long as the Middle East was in turmoil and the price of oil was high and rising. But in 1980, the price of oil began to drop. Soon after, the Reagan administration pulled back on federal support of solar energy.

In the new political and economic environment of the 1980s, TI’s optimism about TISES waned. In late 1982 or early 1983, the program was put on probation. TI told Kilby and program manager E.L. “Pete” Johnson that the company would stop funding the project, but that they could continue developing the technology if they could find a partner to foot the bill.

TI’s decision sprang from two sources. The company was recovering from a failed effort to transform itself into the leading personal computer manufacturer and had decided to focus on things it was already good at. Upper management now frowned on expensive gambles in new markets. New cost projections also reinforced the idea that TISES was too big to chew. By late 1982, the team was projecting a small revenue stream against significant costs to build manufacturing plants, with the project forecast to consume $275 million to $400 million from 1983 through 1990 (around $600 million to $900 million today). Most of that investment would be pure loss for almost a decade.

Kilby wasn’t quite ready to give up. For the next nine months or so, he and Johnson—sometimes accompanied by Heilmeier—made the rounds of potential corporate partners: oil and chemical companies, electric utilities and equipment suppliers, building materials firms, and even—according to Johnson—tobacco companies. But the timing was bad; other microelectronics firms, such as RCA, were selling off their solar energy divisions to oil companies, usually for cut-rate prices. Kilby and company couldn’t find a partner for TISES. By September 1983, TI’s patience and money had run out, and the program was canceled.

Kilby was crushed. According to Kilby’s friend and biographer Ed Millis, the cancellation of TISES “was a tough blow,” especially as it followed not long after the death of his mother in 1980 and his wife in 1981: “The steady stream of inventions that he logged in his notebooks had ceased during the latter part of the solar energy program, and never started again” after TISES was canceled. Kilby quit TI for good and settled down to serve on boards and committees and to collect awards: “Jack the Giant Killer Inventor,” Millis wrote, “seemed to be content to hang up his slide rule.”

With that, Project Illinois disappeared. Today, it’s an almost-­forgotten coda to the career of a legendary engineer. The basic idea behind the technology does still come up from time to time. In fact, while researching this article I was contacted by a team at a major technological institute because they noticed we had both been looking at the Kilby papers at Southern Methodist University, in Dallas—myself as a historian, they as technologists looking to revive the system.

/image/MjgxMjU4OQ.jpegSmall Silicon Spheres: The size of the silicon spheres in the solar energy scheme evolved over time. These measure roughly a quarter of a millimeter in diameter.Photo: Texas Instruments records/DeGolyer Library/Southern Methodist University

What should we make of Kilby’s great disappointment? Does it tell us anything about alternative energy or high technology in general? Maybe. It’s possible that the TISES system had flaws that justified its cancellation. But plenty of other solar energy projects were canceled at the same time. It seems likely that the solar bust of the early ’80s killed at least some technologies that, with further development, could be competitive today.

High tech, especially in the United States, often proceeds in booms and busts that can inhibit development over the longer term. Surviving those busts depends on a long-range vision and significant investment. But Kilby and TI’s arguments for Project Illinois were grounded in the economic climate of the moment, and that justification faltered when the price of oil dropped. It didn’t help that they pitched the project to the national security community, which was more receptive toward nuclear energy. A more long-range vision, motivated perhaps by a concern for the environment or the desire to build a more diverse and thus resilient portfolio of energy sources, might have resulted in slower development but more stable support from both public and private sources.

It may also be that microelectronics is a poor analogy for other technologies. Kilby and TI believed that their expertise in manufacturing chips from silicon would translate easily to manufacturing power systems that relied on silicon. They believed that their experience with the Department of Energy would be just as collegial and productive as it had been with the Department of Defense. And they believed that their experience with consumer markets for calculators and pedagogical toys would translate to selling durable goods for homes. None of those ideas proved valid. That’s not to say that people who know something about integrated circuits can’t or shouldn’t reinvent themselves as experts in solar power. But it does show that reinvention is tough.

The prospects for solar power look much better now than they did in 1983. Solar cell efficiencies are higher, and costs are much lower. Enthusiasm for solar is today buoyed by widespread anxieties about sustainability and climate change. History won’t repeat itself here—solar’s future will be different from its past. But we can use stories like Kilby’s to help us plot the best path forward over the long term—whether solar’s short-term outlook is cloudy or bright.

This article appears in the October 2016 print issue as “Burnt by the Sun.”

About the Author

Cyrus C.M. Mody is a historian of science and technology at Maastricht University, in the Netherlands.

The Conversation (0)