Historical Interlude: From the Mainframe to the Minicomputer Part 1, Transistors and Integrated Circuits

So now its time to pause again in our examination of video game history to catch up on the technological advances that would culminate in the emergence of an interactive entertainment industry.  As previously discussed, the release and subsequent spread of Spacewar! in 1962 represented the first widespread interest in computer gaming, yet no commercial products would appear before 1971.  In the meantime, computer games continued to be written throughout the 1960s (which will be discussed in a subsequent post), but none of them gained the same wide exposure or popularity as Spacewar!.  Numerous roadblocks prevented the spread of these early computer games ranging from the difficulty of porting programs between systems to the lack of reliable wide area distribution networks, but the primary inhibitor remained cost, as even a relatively cheap $120,000 PDP-1 remained an investment out of the reach of most organizations — let alone the general public — and many computers still cost ten times that amount.

The key to transforming the video game into a commercial product therefore lay in significantly reducing the cost of the hardware involved.  The primary expense in building a computer remained the switching units that defined their internal logic, which in the late 1950s were still generally the bulky, power-hungry, temperamental vacuum tubes.  In 1947, John Bardeen and Walter Brattain at Bell Labs demonstrated the solution to the vacuum tube problem in the form of the semiconducting transistor, but as with any new technology there were numerous production and cost issues that had to be overcome before it could completely displace the vacuum tube.  By the early 1960s, the transistor was finally well established in the computer industry, but while it drove down the cost and size of computers like DEC’s PDP-1, a consumer product remained out of reach.  Finally, in late 1958 and early 1959 engineers working independently at two of the most important semiconductor manufacturers in the world discovered how to integrate all of the components of a circuit on one small plate, commonly called a “chip,” paving the way for cost and size reductions that would allow the creation of the first minicomputers, which remained out of reach for the individual consumer, but could at least be deployed in a public entertainment setting like an arcade.

Note:  Once again, this is a “historical interlude” post that will provide a summary of events drawn from a few secondary sources rather than the in-depth historiographic analysis of my purely game-related posts.  The majority of the information in this post is drawn from Forbes Greatest Technology Stories: Inspiring Tales of the Entrepreneurs and Inventors Who Revolutionized Modern Business by Jeffrey Young, The Man Behind the Microchip: Robert Noyce and the Invention of Silicon Valley by Leslie Berlin, The Intel Trinity: How Robert Noyce, Gordon Moore, and Andy Grove Built the World’s Most Important Company by Michael Malone, an article from the July 1982 issue of Texas Monthly called “The Texas Edison” by T.R. Reid, and The Silicon Engine, an online exhibit maintained by the Computer History Museum.

The Transistor Enters Mass Production

1951_1_2

Gordon Teal (l), whose crystal-growing techniques were crucial to mass producing the transistor

As previously discussed, on December 23, 1947, William Shockley, John Bardeen, and Walter Brattain demonstrated the transistor for the first time in front of a group of managers at Bell Labs, which is widely considered the official birthday of the device.  This transistor consisted of a lump of germanium with three wires soldered to its surface in order to introduce the electrons.  While this point-contact transistor produced the desired results, however, it was difficult to manufacture, with yield rates of only fifty percent.  Determined to create a better device — in part due to anger that Bardeen and Brattain received all the credit for the invention — William Shockley explored alternative avenues to create a less fragile transistor.

In 1940, Bell Labs researchers Russell Ohl and Jack Scaff had discovered while working on semiconductor applications for radar that semiconducting crystals could have either a positive or a negative polarity, which were classified as p-type and n-type crystals respectively.  Shockley believed that by creating a “sandwich” with a small amount of p-type material placed between n-type material on either end, he could create what he termed a junction transistor that would amplify or block a current when a charge of the appropriate polarity was applied to the p-type material in the middle.  Placing the required impurities in just the right spots in the germanium proved challenging, but by 1949, Shockley was able to demonstrate a working p-n junction transistor.  While the junction transistor was theoretically well suited for mass production, however, in reality the stringent purity and uniformity requirements of the semiconducting crystals presented great challenges.  Gordon Teal, a chemist with a Ph.D. from Brown who joined Bell Labs in 1930 and worked on radar during World War II, believed that large crystals doped with impurities at precise points would be necessary to reliably produce a working junction transistor, but he apparently garnered little support for his theories from Shockley and other managers at Bell Labs.  He finally took it upon himself to develop a suitable process for growing crystals with the help of engineer John Little and technician Ernest Buehler, which they successfully demonstrated in 1951.  That same year, another Bell Labs researcher named William Pfann developed a technique called zone refining that allowed for the creation of ultra-pure crystals with minuscule amounts of impurities, which lowered the manufacturing cost of the junction transistor significantly.  Together, the advances by Teal and Pfann provided Bell Labs with a viable fabrication process for transistors.

Part of the reason Teal could not generate much excitement about his manufacturing techniques at Bell Labs is that AT&T remained unsure about entering the transistor business.  Despite recent advances, executives remained doubtful that the transistor would ultimately replace the large and well-established vacuum tube industry.  Worse, the company was currently under investigation by the U.S Department of Justice for anti-trust violations and was therefore hesitant to enter and attempt to dominate a new field of technology.  Therefore, in 1952 the company decided to offer a royalty-free license to any company willing to research integrating the transistor into hearing aids, one of the original passions of company founder Alexander Graham Bell, and held a series of technical seminars introducing interested parties to the device.  Several large electronics companies signed up, including Raytheon, Zenith, and RCA.  They were joined by a relatively small company named Texas Instruments (TI).

74718_fig

From Left to Right, John Erik Jonsson, Henry Bates Peacock, Eugene McDermott, and Cecil Green, the men who transformed Geophysical Service, Inc. into Texas Intruments

In 1924, two physicists named Clarence Karcher and Eugene McDermott established the Geophysical Research Corporation (GRC) in Tulsa, Oklahoma, as a subsidiary of Amerada Petroleum.  The duo had been developing a reflection-seismograph process to map faults and domes beneath the earth when they realized that the same process was ideal for discovering oil deposits.  By 1930, GRC had become the leading geophysical exploration company active along the Gulf Coast, but the founders disliked working for Amerada, so they established a new laboratory in Newark, New Jersey, and with investment from geologist Everette DeGolyer formed a new independent company called Geophysical Service, Inc. (GSI).  In 1934, the company moved the laboratory to Dallas to be closer to the heart of the oil trade.

The early 1930s were not a particularly auspicious time to start a new business with the Great Depression in full swing, but GSI managed to grow by aggressively expanding its oil exploration business into international markets such as Mexico, South America, and the Middle East.  Success abroad did not fully compensate for difficulties in the US, however, so in December 1938, the company reorganized in order to exploit the untapped oil fields in the American Southwest.  A new Geophysical Service, Inc. — renamed the Coronado Corporation early the next year — was established with Karcher at the helm as an oil production business, while the original GSI, now headed solely by McDermott, became a subsidiary of Coronado and continued in the exploration business.  The company failed to flourish, however, so in 1941 Karcher negotiated a $5 million sale of Coronado to Stanolind Oil & Gas.  Not particularly interested in the exploration business, Stanolind offered the employees of GSI the opportunity to buy back the company for $300,000.  McDermott, R&D head J. Erik Jonsson, field exploration head Cecil Green, and crew chief H. Bates Peaock managed to scrape together the necessary funding and purchased GSI on December 6, 1941.  The very next day, the Japanese bombed Pearl Harbor, dragging the United States into World War II.

With so much of its business tied up in international oil exploration work that would have to be abandoned during the coming global conflict, GSI would be unable to survive by concentrating solely on its primary business and now needed to find additional sources of income.  The solution to this problem came from Jonsson, a former aluminium sales engineer who had been in charge of R&D at GSI since the company’s inception in 1930, who realized that the same technology used for locating oil could also be used to locate ships and airplanes.  A fortuitous connection between McDermott and Dr. Dana Mitchell, who was part of a group working on electronic countermeasure technology, led to a contract to manufacture a device called the magnetic anomaly detection (MAD) system.  Building on this work, GSI emerged as a major supplier of military electronics by the end of the war.

During the war, Jonsson became impressed with an electrical engineer and Navy lieutenant from North Dakota working as a procurement officer for the Navy’s Bureau of Aeronautics named Patrick Haggerty.  In 1946, GSI hired Haggerty to run its new Laboratory and Manufacturing Division, which the company established to expand its wartime electronics work in both the military and private sectors.  Haggerty was determined to transform GSI into a major player in the field and convinced management to invest in a large new manufacturing plant that would require the company to tap nearly its entire $350,000 line of credit with the Republic National Bank.  By 1950, this investment had turned into annual sales of nearly $10 million a year.  With manufacturing now a far more important part of the business than oil exploration, company executives realized the name GSI no longer fit the company.  They decided to change the name to General Instruments, which conjured up visions of the great electronics concerns of the East like General Electric.  Unfortunately, there was already a defense contractor with that name, so the Pentagon asked them to pick something else.  They chose Texas Instruments.

 46-pat_haggerty

Patrick J. Haggerty, the man who brought TI into the transistor business

When Patrick Haggerty learned AT&T was offering licenses for transistor technology, he knew immediately that TI had to be involved.  AT&T, however, disagreed.  In 1952, TI had realized a profit of $900,000 on sales of just $20 million and did not appear capable of making the necessary investment to harness the full potential of the transistor.  It took a year for TI management to finally convince AT&T to grant the firm the $25,000 license, after which Haggerty made another large financial gamble, investing over $4 million in manufacturing plants, development, new hires, and other startup costs.  Before the end of 1952, TI had its first order for 100 germanium transistors from the Gruen Watch Company, and production formally began.

Haggerty had muscled TI into an important new segment of the electronics industry, but in the end it was AT&T that was proven correct:  TI really was too small to make much of an impact in the germanium transistor market.  Haggerty therefore turned to new technology to keep his company relevant in the field.  While germanium served as a perfectly fine semiconducting material at temperatures below 100 degrees Fahrenheit, the low melting point of the element inhibited its semiconducting properties at high temperatures, rendering it unsuitable for defense projects like guided missiles.  Silicon offered both better semiconducting capability and a higher temperature tolerance, but despite the best efforts of scientists at Bell Labs and elsewhere, the element had proven impossible to dope with the necessary impurities.  This did not dissuade Haggerty, who placed an ad in the New York Times for a new chief researcher who could bring TI into silicon transistors.  That ad was answered by none other than brilliant Bell Labs chemist Gordon Teal.

Feeling unappreciated after facing such resistance to his research at Bell Labs, Teal was ready to move on, but despite answering the TI ad, he was not certain the Texas company was the right fit.  Solving the problems with silicon would require a great deal of time and money, and TI remained a relatively small concern.  Haggerty reassured him, however, by revealing that TI was preparing to merge with Intercontinental Rubber, a cash-rich firm listed on the New York Stock Exchange with a faltering tire and rubber business.  This merger, completed in October 1953, made TI a public company and guaranteed that Teal would have the funding he needed.  Haggerty promised Teal anything and anyone he needed with only one stipulation: after one year, Teal would need to have a product TI could bring to market.  Teal accepted the challenge.

1954 proved to be a trying year for TI.  While the transistor business failed to gain traction against larger competitors, the defense contracts the company depended upon as its primary source of revenue began to dry up with the end of the Korean War and a subsequent cut in military spending.  Revenues that had risen to $27 million in 1953 declined to $24 million, profits fell slightly from $1.27 million to $1.2 million, and the stock began trading in single digits.  That same year, however, Teal succeeded in developing a complicated high-temperature doping and zone refining process that yielded a viable silicon transistor.  At a conference on airborne aeronautics held in Dayton, Ohio, that spring, Teal not only proudly announced to the assembled that TI had a working silicon transistor in production, he also provided a dramatic demonstration.  A record player was produced, specially modified so that a transistor could be snapped in and out to complete a circuit.  First, Teal snapped in a germanium transistor and then dropped it into a beaker of hot oil, which destroyed the transistor and stopped the player.  Then, he performed the same action with a silicon transistor.  The music played on.  TI quickly found itself swamped with orders.

New Players

Traitorous-8-Fairchild

The “Traitorous Eight,” who left Shockley Semiconductor to establish Fairchild Semiconductor.

From left: Gordon Moore, C. Sheldon Roberts, Eugene Kleiner, Robert Noyce, Victor Grinich, Julius Blank, Jean Hoerni, and Jay Last

In 1954 Bell Labs chemist Calvin Fuller developed a new technique called the diffusion process in which silicon could be doped at high temperatures using gasses containing the desired impurities.  By the next March, Bell Labs chemist Morris Tanenbaum had succeeded in harnessing the diffusion process to create semiconducting material so thin that a silicon wafer could be created in which each layer of the n-p-n sandwich was only a millimeter thick.  The resulting diffusion-base transistor operated at much higher frequencies than previous junction transistors and therefore performed much faster.  With Gordon Teal’s crystal-growing expertise and Patrick Haggerty’s salesmanship, TI kept pace with these advancements and enjoyed a virtual monopoly on the emerging field of silicon transistors during the next few years, with company revenues soaring to $45.7 million in 1956.  The transistor business, however, remained a relatively small part of the overall electronics industry.  Between 1954 and 1956, 17 million germanium transistors and 11 million silicon transistors were sold in the United States.  During the same period, 1.3 billion vacuum tubes were sold.

Practically speaking, the vacuum tube companies appeared to hold a distinct advantage, as they could theoretically use the enormous resources at their disposal from their vacuum tube sales to support R&D in transistors and gradually transition to the new technology.  In reality, however, while most of the major tube companies established small transistor operations, they were so accustomed to the relatively static technologies and processes associated with the tube industry that they were unable to cope with the volatile pricing and ever-changing manufacturing techniques that defined the transistor industry.  The Philco Corporation is a poster child for these difficulties.  Established in Philadelphia in 1892 as the Helios Electric Company to produce lamps, Philco became a major player in the emerging field of consumer radios in the mid-1920s and by the end of World War II was one of the largest producers of vacuum tubes in the United States.  The company seriously pursued transistor technology, creating in 1953 the high-speed surface-barrier transistor discussed in a previous post that powered the TX-0.  In 1956, Philco improved the surface-barrier transistor by employing the diffusion process, but the company soon grew leery of attempting to keep up with new transistor technologies.  The original surface-barrier transistor had been fast, but expensive, and the diffusion-based model cost even more, retailing for around $100.  As technology continued to progress, however, the price fell to $50 within six months, and then to $19 a year after that.  By the next year, lots of 1,000 Philco transistors could be had for a mere $6.75.  Spooked, the company ultimately decided to remain focused on vacuum tubes.  By 1960, Philco had entered bankruptcy, and Ford subsequently purchased the firm in 1961.

12ShockleySemiConLabs-1384982690614

The Shockley Semiconductor Laboratory in the Heart of the region that would become Silicon Valley

While the old guard in the electronics industry ultimately exerted little influence on the transistor business, TI soon faced competition from more formidable opponents.  In 1950, William Shockley paid a visit to Georges Doriot, the pioneering venture capitalist who later funded the Digital Equipment Corporation.  Surprisingly, their discussion did not focus on the transistor, but rather on another invention Shockley patented in 1948, a “Radiant Energy Control System,” essentially a feedback system using a visual sensor.  Shockley had worked on improving bomb sights during World War II and saw this system as the next step, potentially allowing a self-guided bomb to compare photographs of targets with visual data from the sensor for increased accuracy.  The same technology could also be used for facial recognition, or for automated sorting of components in manufacturing.  Since the publication of mathematician Norbert Wiener’s groundbreaking book, Cybernetics, in 1948, the Cambridge academic community had been excited by the prospect of using artificial systems to replace human labor for more mundane tasks.  Indeed, in 1952 this concept would gain the name “automation,” a term first coined by Delmar Harder at Ford and popularized by Harvard Business School Professor John Diebold in his book Automation: The Advent of the Automatic Factory.  When Doriot learned of Shockley’s control system, he urged the eminent physicist to waste no time in starting his own company.

By 1951, Shockley had refined his “Radiant Energy Control System” into an optoelectronic eye he felt could form the core of an automated robot that could replace humans on the manufacturing line.  After negotiating an exemption with Bell Labs allowing him to maintain the rights to any patents he filed related to automation for the period of one year, Shockley filed a patent for an “Electrooptical Control System” and wrote a memo to Bell Labs president Mervin Kelly urging the organization to build an “automatic trainable robot.”  When Kelly refused to consider such a project, Shockley, already stripped of most of his responsibilities regarding transistor development due to incessant conflicts with his team, took a leave of absence from Bell Labs in late 1952.  After a year as a visiting professor at CalTech, Shockley became director of the Pentagon’s Weapons Systems Evaluation Group and spent the next year or so studying methods for the U.S. to fight a nuclear war while periodically turning down offers to teach at prestigious universities or establish his own semiconductor operation.

In February 1955, Shockley met renowned chemist Arnold Beckman at a gala in Los Angeles honoring Shockley and amplifier inventor Lee DeForest.  The two bonded over their shared interest in automation and kept in touch over the following months.  Finally, in June 1955, Shockley decided he needed to radically change his life, so he resigned from both Bell Labs and his Pentagon job, divorced his wife, and began to seriously consider offers to start his own company.  The next month, he contacted Beckman to propose forming a company together to bring the new diffusion transistor to market and develop methods to automate the production of transistors.  After a period of negotiation, the Shockley Semiconductor Laboratory was established in September 1955 as a subsidiary of Beckman Instruments.  Even though Beckman was headquartered in Southern California, Shockley convinced his new partner to locate Shockley Semiconductor further north in Palo Alto, California, so he could once again remain close to his mother.

Unable to recruit personnel from Bell Labs, where his reputation as a horrible boss proceeded him, Shockley scoured technical conferences, college physics departments, and research laboratories for bright young scientists and engineers.  One of his first hires also proved to be his most important, a young physicist named Bob Noyce.  Born in 1927 in Burlington, Iowa, Robert Norton Noyce was the son of a Congregationalist minister who moved his family all over the state of Iowa as he migrated from one congregation to the next.  This itinerant life, made even more difficult by the Depression, finally ended in 1940 when Ralph Noyce took a job in the college town of Grinnell, Iowa.  Bob Noyce thrived in Grinnell, where his natural charisma and sense of adventure soon made him the leader among the neighborhood children.  A brilliant student despite a penchant for mischief and goofing off, Noyce took a college physics course at Grinnell College during his senior year of high school and graduated class valedictorian.  The Miami University Department of Physics offered to give him a job as a lab assistant if he attended the school — an honor usually reserved for graduate students — but worrying he could just be another face in the crowd at such a large institution, Noyce chose to study at Grinnell College instead.

At Grinnell, Noyce nearly lost his way at the end of his junior year.  Eager to maintain his social standing among older students returning from World War II, Noyce agreed to “procure” a pig to roast at a Hawaiian Luau dorm party.  Soon after, he learned his girlfriend was pregnant and would need an abortion.  Depressed, Noyce got drunk and with the help of a friend stole a pig from a local farmer’s field.  Feeling remorseful, they returned the next day to apologize to the farmer and pay for the pig only to learn that he was the mayor of Grinnell and did not take the prank lightly.  Noyce was almost expelled as a result, but he was saved by his physics professor, Grant Gale, who saw Noyce as a once-in-a-generation talent that should not be squandered over an ill-advised prank.  The university relented and merely suspended him for a semester.

When Noyce returned to Grinnell after working for a life insurance company in New York during his forced exile, he was introduced to the technology that would change his life.  His mentor Gale was an old friend of transistor co-inventor John Bardeen, with whom he had attended the University of Wisconsin, while the head of research at Bell Labs, Oliver Buckley, was a Grinnell graduate.  Gale therefore learned of the transistor’s invention early and was able to secure a wide array of documentation on the new device from Bell.  When Noyce saw his professor enraptured by these documents, he dove right in himself and soon resolved to learn everything he could about transistors.  After graduating from Grinnell with degrees in mathematics and physics, Noyce matriculated to the physics department at MIT, where he planned to focus his studies on solid-state physics.  As transistors were so new, most of Noyce’s classwork revolved around vacuum tubes, but his dissertation, completed in mid 1953, dealt with matters related to transistor development.  Upon earning his doctorate in physics, Noyce took a job at Philco, where in 1950 R&D executive Bill Bradley had established the 25-man research group that developed the surface-barrier transistor.  Noyce rose through the ranks quickly at Philco, but he soon became disillusioned with the layers of bureaucracy and paperwork inherent in working for a large defense contractor, especially after the company was forced to significantly curtail R&D activities due to losses.  Just as Noyce was looking for a way out, Shockley called in January 1956 after reading a paper Noyce had presented on surface-barrier transistors several months earlier at a conference.  In March, Noyce headed west to join Shockley Semiconductor.

Before long, Shockley had succeeded in recruiting a team of about twenty with expertise in a variety of fields related to transistor creation. These individuals included a Ph.D. candidate in the solid state physics program at MIT named Jay Last, a chemist at the Johns Hopkins Applied Physics Lab named Gordon Moore, a mechanical engineer at Western Electric named Julius Blank, Viennese World War II refugee and expert tool builder Eugene Kleiner, metallurgist Sheldon Roberts, Swiss theoretical physicist Jean Hoerni, and Stanford Research Institute physicist Vic Grinich.  Shockley hoped these bright young scientists would secure his company’s dominance in the semiconductor industry.

LE281L4

Sherman Fairchild, the inventor and businessman who financed Fairchild Semiconductor

On November 1, 1956, William Shockley learned that he had been awarded the Nobel Prize for Physics — shared with Walter Brattain and John Bardeen — for the invention of the transistor.  Theoretically at the height of his fame and powers, Shockley soon found his entire operation falling apart.  Always a difficult man to work for, his autocratic tendencies grew even worse now that he was a Nobel laureate in charge of his own company.  He micromanaged employees, even in areas outside of his expertise, and viciously attacked them when their work was not up to his standards.  Feeling threatened by Jean Hoerni and his pair of doctorates, he once exiled the physicist to an apartment to work alone, though he later relented.  He discouraged his employees from pursuing their own projects and insisted on adding his name to any paper they presented, whether he had any involvement in the subject or not.  Once, when a secretary cut her hand on a piece of metal protruding from a door, he insisted it must have been an act of sabotage and threatened to hire a private investigator and subject the staff to lie detector tests.  He was finally dissuaded by Roberts, who convinced him with the aid of a microscope that the piece of metal was merely a tack that had lost its plastic head.

The final straw was Shockley’s insistence on pulling staff and resources from improving upon the diffusion-base silicon transistor to work on a new four-layer diode project he believed could act as both a transistor and a resistor and was theoretically faster and cheaper than a germanium transistor.   In reality, this device proved impossible to create, and R&D costs began to spiral out of control with no sellable product to show for it.  This caused Beckman to become more involved with company operations, which in turn led several of Shockley’s disgruntled employees to feel they could effect real change.  They nominated Robert Noyce as their spokesman, both because he maintained a cordial relationship with Shockley and because he was possessed of an impressive charisma that made him both a natural team leader and an easy person to talk to.  With Beckman’s blessing, Noyce, Moore, Kleiner, Last, Hoerni, Roberts, Blank, and Grinich confronted Shockley and attempted to force him out of day-to-day operations at the company.  The octet wanted Noyce to serve as their new manager, but Shockley refused, arguing that Noyce did not have what it took to be an aggressive and decisive leader, criticisms that later events would show were completely justified.  Beckman therefore appointed an interim management committee and began an external search for an experienced manager.  Less than a month later, he reversed course and declared Shockley to be in charge, most likely influenced by colleagues at either Bell Labs or Stanford who pointed out that undermining Shockley would unduly tarnish the reputation of the Nobel laureate.  As a compromise, Noyce was placed in charge of R&D and a manager from another division of Beckman named Maurice Hanafin was installed as a buffer between Shockley and the rest of the staff.

Noyce was satisfied with this turn of events, but his seven compatriots were not, especially when it became clear that Shockley remained in complete control despite the appointment of Hanafin.  Led by Last, Hoerni, and Roberts, the seven scientists decided to leave the company.  Feeling they were more valuable as a group, however, they resolved to continue working together rather than going their separate ways, meaning they would need to convince an established company to hire them together and form a semiconductor research group around them.  To facilitate this process, Kleiner decided to write to a New York investment firm where his father had an account called Hayden, Stone, and Company, which had recently arranged financing for the first publicly held transistor firm, General Transistor.  Kleiner’s letter was addressed to the man in charge of his father’s account and asked for $750,000 in funding to start a new semiconductor group.  As it turned out, the account man was no longer there, so the letter ended up on the desk of a recent hire and Harvard MBA named Arthur Rock.  Rock liked what he saw and met with the seven along with his boss, Arthur “Bud” Coyle.  The two bankers strongly believed in the potential of the scientists and urged them to reach beyond their original plan and ask for a million dollars or more to fund an entire division.  In order to entice a company to form a semiconductor division, however, the seven scientists would need a leader, and none of them felt up to the task.  They realized they would have to recruit their former ringleader in their fight against Shockley, Bob Noyce.  It took some convincing, but Noyce ultimately came on board.  The seven were now eight.

Finding a company to shelter the eight co-conspirators proved harder than Rock and Coyle initially hoped.  The duo drew up a list of thirty companies they believed could handle the investment they were looking for, but were turned down by all of them.  Simply put, no one was interested in giving a group of scientists between the ages of 28 and 32 that had never developed a salable product yet felt they could run a division better than a Nobel Prize winner $1 million to pursue new advances in a volatile field of technology.  Running out of options, Coyle mentioned the plan to an acquaintance possessed of both a large fortune and a reputation for risk-taking:  Sherman Fairchild.  Sherman was the son of George Fairchild, a businessman and six-term Congressman who played a crucial role in the formation of the International Time Recording Company — one of the companies that merged to form C-T-R — and was the chairman and largest shareholder of C-T-R/IBM from its inception until his death in 1924.  A prolific inventor, Sherman developed a camera suitable for aerial photography for the United States Army during World War I and then established the Fairchild Aerial Camera Corporation in 1920.  Subsequently, Fairchild established several more companies based around his own inventions in fields ranging from aerial surveying to aircraft design.  In 1927, he consolidated seven of these organizations under the holding company Fairchild Aviation, which he renamed Fairchild Camera and Instrument (FCI) in 1944 after spinning back out his aviation business.  By 1957, Fairchild was no longer involved in the day-to-day running of any of his companies, but he was intrigued by the opportunity represented by Noyce and his compatriots and encouraged FCI to take a closer look.

Based in Syosset, New York, Fairchild Camera and Instrument had recently been placed under the care of John Carter, a former vice president of Corning Glass who felt that FCI had become too reliant on defense work for its profits, which had become scarcer and scarcer since the end of the Korean War.  Carter believed acquisitions would be the best way to secure a new course for FCI, so he proved extremely amenable to Noyce and company’s request for funding.  After a period of negotiation, Fairchild Semiconductor Corporation was formally established on September 19, 1957.  Officially, FCI loaned Fairchild Semiconductor $1.3 million in startup funding and in return was granted control of the company through a voting trust.  Ownership of Fairchild Semiconductor remained with the eight founding members and Hayden, Stone, but FCI had the right to purchase all outstanding shares of the company on favorable terms any time before it achieved three successive years of earnings of $300,000 or more.  When the scientists finally broke the news of their imminent departure to Shockley, the Nobel laureate was devastated, and though he never actually dubbed them the “Traitorous Eight,” a phrase invented by a reporter some years later, the phrase came to be associated with his feelings on the matter.  Shockley continued to pursue his dream of a four-layer diode until Beckman finally sold Shockley Semiconductor, which had never turned a profit, in 1960.  Shockley himself ultimately left the industry to teach at Stanford.

The Process

1959_1_2

A transistor built using the “planar process,” which revolutionized the nascent semiconductor industry

 In October 1957, Fairchild Semiconductor moved into its new facilities on Charleston Road near the southern border of Palo Alto, not far from the building that housed Shockley Semiconductor.  The Fairchild executive responsible for negotiating the final deal between FCI and the Traitorous Eight, Richard Hodgson, took on the role of chairman of the semiconductor company to look after FCI’s interests and began a search for a general manager.  Hodgson’s first choice was the charismatic Noyce, but the physicist hated confrontation and felt unready to run a whole company besides and contented himself with leading R&D.  Hodgson therefore brought in an old friend, a former physics professor that had worked as a sales manager for FCI in the 1950s named Tom Bay, to head up sales and marketing and a former paratrooper who managed the diode operation at Hughes Aircraft named Ed Baldwin as general manager.

Fairchild Semiconductor came into being at just the right time.  On October 4, 1957, the Soviet Union launched Sputnik into orbit, inaugurating a space race with the United States that greatly increased the Federal Government’s demand for transistors for use in rockets and satellites, technologies particularly unsuited to vacuum tubes due to the need for small, durable components.  At the same time, the rise of affordable silicon transistors had government agencies reevaluating the use of vacuum tubes across all their projects, particularly in computers.  This led directly to Fairchild’s first major contract.

In early 1958, Tom Bay learned that the IBM Federal Systems Division was having difficulty sourcing the parts it needed to create a navigational computer for the United States Air Force’s experimental B-70 long-range bomber.  The Air Force required particularly fast and durable silicon transistors for the project and TI, still the only major force in silicon, had been unable to provide a working model up to their specifications.  Through inheritance from his father, Sherman Fairchild was the largest shareholder at IBM and wielded some influence at the company, so Bay and Hodgson convinced him to secure a meeting with the project engineers.  IBM remained skeptical even after Noyce stated Fairchild’s engineers were up to the task, but Sherman Fairchild leaned hard on Tom Watson Jr., basically saying that if he trusted the engineers enough to invest over $1 million in their work, then Watson should trust them too.  With Sherman’s help, Fairchild Semiconductor secured a contract for 100 silicon transistors in February 1958.

Noyce knew that the project would require a type of transistor known as a mesa transistor that had been developed by Bell Labs and briefly worked on at Shockley Semiconductor, but had yet to be mass produced by any company.  Unlike previous transistors, the mesa transistor could be diffused on only one side of the wafer by taking advantage of new techniques in doping and etching.  Basically, dopants were diffused beneath a layer of silicon, after which a drop of wax was placed over the wafer.  The entire surface would then be doused in a strong acid that etched away the entire top layer except at the point protected by the wax.  This created a distinctive bump that resembled the mesas of the American Southwest, hence its name.  Fairchild decided to develop the first commercial double-diffused silicon mesa transistor, but were unsure whether an n-p-n or p-n-p configuration would perform better.  They therefore split into two teams led by Moore and Hoerni to develop both, ultimately settling on the n-p-n configuration.  Putting the transistor into production was a complete team effort.  Roberts took charge of growing the silicon crystals, Moore and Hoerni oversaw the diffusion process, Noyce and Last handled the photolithographic process to define the individual transistors on the wafer, Grinich took charge of testing, and Blank and Kleiner designed the manufacturing facility.  By May, the team had completed the design of the transistor, which they delivered to IBM in the early summer.  In August, the team presented their transistor at Wescon, an important trade show established six years before by the West Coast Electronics Manufacturers Association, and learned that their double-diffusion transistor was the only one on the market.  They maintained a monopoly on the device for about a year.

Orders soon began pouring in for double-diffused mesa transistors, most notably from defense contractor Autonetics, which wanted to use them in the Minuteman guided missile program, then the largest and most important defense project under development.  Late in 1958, however, Fairchild realized there was a serious problem with the transistor: it was exceedingly fragile.  So fragile, in fact, that even a tap from a pencil could cause one to stop working.  After testing, the team determined that when the transistor was sealed, a piece of metal would often flake off the outer can and bounce around inside, ultimately causing a short.  Fairchild would need to solve this problem quickly or risk losing its lucrative defense contracts.

During the transistor creation process, an oxide layer naturally builds up on the surface of the silicon wafer.  While this oxide layer does not interfere with the operation of the transistor, it would nevertheless be removed to prevent impurities from becoming trapped under its surface.  As early as 1957, Jean Hoerni speculated that the impurity problem was entirely imaginary and that the oxide layer could, in fact, provide a service by protecting the otherwise exposed junctions of the transistor and thus prevent just the kind of short Fairchild was now grappling with.  Hoerni did not pursue the concept at the time because Faircihld was so focused on bringing its first products to market, but in January 1959, he attacked the problem in earnest and within weeks had figured out a way to introduce an oxide mask at proper points during the diffusion process while still leaving spaces for the necessary impurities to be introduced.  On March 12, 1959, Hoerni proudly demonstrated a working transistor protected by an oxide layer, spitting on it to demonstrate it would continue working even when subjected to abuse.  Unlike the mesa transistor, a transistor created using Hoerni’s new technique resembled a bullseye with an outer layer shaped like a teardrop and was flat and smooth.  He therefore named his new technique the “planar process.”

The planar process instantly rendered all previous methods of creating transistors obsolete.  Consequently, Fairchild would not only be able to corner the market in the short term by bringing the first planar transistor to market, but it would also be able to generate income in the long term by licensing the planar process to all the other companies in the transistor business.  Complete dominance of the semiconductor industry appeared to be within Fairchild’s grasp, but then in mid-March 1959, TI announced a new product that would change the entire course of the electronics industry and, indeed, the modern world.

The Texas Edison

co1043

Jack Kilby, the inventor of the first integrated circuit

As Fairchild was just starting its transistor business in 1958, Texas Instruments continued to extend its dominance as company revenues reached $90 million and profits soared, but the company was not content to rest on its laurels.  With the space race beginning, the military, to which TI still devoted a large portion of its electronic components business, required ever more sophisticated rockets and computers that would require millions of components to function properly.  Clearly, as long as an electronic circuit continued to require discrete transistors, resistors, capacitors, diodes, etc. all connected by wires, it would be impossible to build the next generation of electronic devices.  The solution to this problem was first proposed by a British scientist named Geoffrey Dunmer in 1952, who spoke of a solid block of material without any connecting wires that would integrate all the functionality of the discrete components of a circuit.  Dunmer was never able to complete a working block circuit based on his theories, but other organizations were soon following in his footsteps, including a physical chemist at Texas Instruments named Willas Adock.  Working under an Army contract, Adcock assembled a small task force to build a simpler circuit, which included an electrical engineer named Jack Kilby.

Born in Jefferson City, Missouri, Jack St. Clair Kilby grew up in Great Bend, Kansas, where his father worked as an electrical engineer and ultimately rose to the presidency of the Kansas Power Company.  Kilby became hooked on electrical engineering during summers spent travelling across western Kansas with his father in the 1930s as the elder Kilby visited power plants and substations inspecting and fixing equipment.  A good student, Kilby planned to continue his education at MIT, but his high school did not offer all the required math courses.  Kilby was forced to travel to Cambridge to take a special entrance exam, but did not pass.  He attended the University of Illinois instead, but his education was interrupted by service during World War II.  Kilby finally graduated in 1947 with an unremarkable academic record and took a job at a Milwaukee firm called Centralab, the only company that offered him a job.

Centralab was not a particularly important company in the electronics industry, but it did experiment with an early form of integrated circuit in which company engineers attempted to place resistors, vacuum tubes, and wiring on a single ceramic base, exposing Kilby to the concept for the first time.  In May 1958, Kilby joined Adcock’s team at TI.  Adcock was attempting to create something called a “micromodule,” in which all the components of a circuit are manufactured in one size with the wiring built into each part so they could simply be snapped together, thus obviating the need for individual wiring connections.  While a circuit built in this manner would still be composed of discrete components, it would theoretically be much smaller, more durable, and easier to manufacture.  Having already tried something similar at Centralab, however, Kilby was convinced this approach would not work.

In the 1950s, Texas Instruments followed a mass vacation policy in which all employees took time off during the same few weeks in the summer.  Too new to have accrued any vacation time, Kilby therefore found himself alone in the lab in July 1958 and decided to tinker with alternate solutions to the micromodule.  Examining the problem through a wide lens, Kilby reasoned that TI was strongest in silicon and should therefore focus on working with that element.  At the time, capacitors were created using metal and ceramics and resistors were made of carbon, but there was nothing stopping a company from creating both of those components in silicon.  While the performance of these parts would suffer significantly over their traditional counterparts, by crafting everything out of silicon, it would be possible to place the circuit on a single block of material and eliminate wires entirely.  Kilby jotted down some preliminary plans in a notebook on July 24, 1958, and then received approval from Adcock to explore the concept further when everyone returned from vacation.

On September 12, 1958, Kilby successfully demonstrated a working integrated circuit to a group of executives at TI.  While Kilby’s intent had been to craft the device out of silicon, TI did not have any blocks of the element suitable for Kilby’s project on hand, so he was forced to craft his first circuit out of germanium.  Furthermore, Kilby had not yet figured out how to eliminate wiring completely, so his original hand-crafted design could not be reliably mass produced.  Therefore, while TI brought the first integrated circuit into the world, it would be Fairchild Semiconductor that actually made them practical.

In January 1959, as Hoerni was perfecting his planar process, Robert Noyce took inspiration from his colleague’s work and began theorizing how P-N junctions and oxide layers could be used to isolate and protect all the components of a circuit on a single piece of silicon, but just as Hoerni initially sat on his planar process while Fairchild focused on delivering finished products, so too did Noyce decide not to pursue his integrated circuit concept any further.  After Kilby debuted his circuit in March, however, Noyce returned to his initial notes.  While the TI announcement may have partially inspired his work, Fairchild’s patent attorney had previously asked every member of the Fairchild team to brainstorm as many applications for the new planar process as possible for the patent filing, which appears to have been Noyce’s primary motivator.  Regardless of the impetus, Noyce polished up his integrated circuit theories and tasked Jay Last with turning them into a working product.

By May 1960, Fairchild had succeeded in creating a practical and producible integrated circuit in which all of the components were etched on a single sliver of silicon with aluminum traces resting atop a protective oxide layer replacing the wiring.  Both the Minuteman missile and the Apollo moon landing projects quickly embraced the new device as the entire transistor industry became obsolete overnight.  While discrete transistors would power several important computer projects in the 1960s — and even the first home video game system in the early 1970s — the integrated circuit ultimately ushered in a new era of small yet powerful electronic devices that could sit on a small desk or, eventually, be held in the palm of one’s hand yet perform calculations that had once required equipment filling an entire room.  In short, without the integrated circuit, the video game industry as it exists today would not be possible.

Advertisements

One comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s