They worked at RCA. So it wasn't surprising that developers there thought in terms of television. Paul Weimer was one of the key figures. He had started with RCA in 1942, right after he won his PhD in physics at Ohio State. He did an assignment for RCA, studying solid-state physics in Paris in 1959-60. When he returned, the company offered him an opportunity to switch from tubes to semiconductors. Weimer had already cut quite a path in tubes. He participated in developing historic TV-camera tubes, including the image orthicon and the vidicon.

When he returned from Paris, he noticed that people working on advances in transistors were using coplanar structures similar to those he had used in developing the tricolor vidicon; they had fine red, green and blue conducting stripes in contact with the photoconductor. Who knows? Maybe this work in solid state could lead to a tubeless TV camera.

A film of transistors

In the fall of 1960, Weimer started making thin-film transistors on glass substrates. In one pumpdown of his vacuum system, he would deposit a gold source and drain, then deposit polycrystalline semiconductor material over that and, finally, place a gate on top. This was a coplanar process that was similar to what he used in the tricolor vidicon. (It should be noted that Weimer's work depended on depositions, not photoresists, which were used in developing integrated circuits, just then getting under way.)

At first, these deposited transistors weren't very good. But then he placed an insulator between the gate and the semiconductor material and got what he called beautiful characteristics. His 1962 paper, "The TFT — A New Thin-Film Transistor", in the Proceedings of the IEEE drew worldwide attention.

At first, Weimer used cadmium sulphide as the semiconductor material because it was a high-resistivity semiconductor with which he was somewhat familiar. He later used cadmium selenide, which was suggested by a group member, Frank Shallcross. This made for an even better field-effect transistor, not as good as silicon, Weimer recalls, but quite respectable.

In 1963, when Weimer first proposed building a tubeless self-scanned solid-state sensor for cameras, he quickly obtained substantial support from the Research and Technology Division of the Air Force System Command at Wright Patterson Air Force Base, which, combined with continuing RCA support, lasted until the end of the decade, when it became clear that other approaches were more promising.

In 1962, Weimer's work on thin-film transistors attracted the attention of T. Peter Brody at Westinghouse's Research Lab in Pittsburgh. Brody, with a doctorate in theoretical physics from the University of London, encountered resistance, if not outright hostility, from the people at Westinghouse's IC division (which the company called the Molecular Electronics Division) in Baltimore. They felt that the future belonged to bipolar transistors and bipolar ICs. His article, "The Birth and Early Childhood of Active Matrix — A Personal Memoir," in the October 1996 issue of the Journal of the Society for Information Display tells how Brody succeeded in keeping the TFT alive at Westinghouse, thanks to some government contracts.

Following some of Weimer's work with tellurium films, Brody and a colleague, Derrick Page, tried something ridiculous. They deposited tellurium TFTs on a strip of paper instead of the usual glass substrate. It worked. That encouraged them to try thin-film transistors on a wide range of flexible substrates. They even created a half-watt audio amplifier on a small strip of aluminum foil.

Out to kill

Brody's work, like Weimer's work, won international attention. Nevertheless, many people at Westinghouse's IC division kept urging management to kill his money-wasting project. Despite these death threats, which Brody asserts in his article were a valuable spur to invention, his small group devised a continuous fabrication process that used reels of anodized aluminum foil as substrates.

What helped his project survive was the demise of the Westinghouse IC division in 1968, so the people there could no longer campaign against his work on TFTs. Brody's work continued, financed by the Tube Division, and, indeed, expanded to become the Thin Film Devices Department.

By early 1968, Brody demonstrated flexible transistors that could withstand 100 V. They were used with Westinghouse's 14-segment electroluminescent display. This application led Brody to realize that there was lots of activity aimed at developing flat-panel displays, but rather little devoted to addressing picture elements on the displays.

He felt that if the addressing problem were solved, the materials problem would fall into place. He coined the expression "large scale display integration," which was replaced by "active matrix" in his 1975 IEEE Transactions paper, "A 6 X 6 inch 20 lpi Electroluminescent Display Panel."

A 6-inch IC?

Brody's work not only drew lots of attention, it also drew lots of derision. Semiconductor experts talked of their difficulty in making a 1/8-inch chip, yet Brody dreamed of a 6-inch integrated circuit. That was crazy. And the CRT people just laughed. "Who needs a flat panel?"

In the late 1960s and early 1970s, Brody's focus was on an active-matrix electroluminescent display (while Weimer's had been on a solid-state camera). By 1971, liquid-crystal technology emerged as the new target for active-matrix addressing. His work did win some Army contracts despite opposition by George Heilmeier, then deputy director of research and engineering, who worked on early liquid-crystal displays at RCA.

In time, Brody developed an active-matrix electroluminescent video panel and an active-matrix LCD panel. In 1978, Westinghouse terminated the program. In 1981, following several years of money-search, Brody founded Panelvision.

By 1983, the company was first on the market with a working active-matrix LCD display. In the following year, it had a 640 X 400-pixel 9.5-inch display. Brody was told that customers would welcome a bid on 100,000 panels. But he lacked the financial muscle to go into production and the computer companies did not want to invest in his company. So he couldn't build up capacity. In 1985, he sold Panelvision's assets to Litton Industries.

Many experts advised Brody that it was too late for such displays. These experts, however, failed to proffer such advice to companies in Europe and the Far East, many of whom may have decided to build active-matrix displays after Brody's visit on a quest for money.

While Brody's efforts were advancing, so, too, were the efforts at RCA.

Weimer made TFT circuits. He was pleased because he was making transistor circuits with cadmium selenide before RCA made them with silicon. But his focus remained on TV cameras. He could see that his technology would lead to solid-state TV sensors, with TFTs used for scanning. He might have a solid-state replacement for the vidicon.

Weimer and his group reached the point of building photoconductors with 500 X 500 elements, using thin-film transistors on the periphery for X-Y scanning. In one array, they had 1,000 transistors on the periphery of a one-inch square of photoconducting elements. This array was still limited by the stability problems of the early TFTs. That problem was finally solved by a deposited hydrogenated amorphous silicon on transparent glass substrates. A number of workers at RCA and organizations in England contributed to this development, which served very well for driving single pixels in a liquid-crystal display.

But silicon was becoming increasingly useful. Workers at Bell Labs found it possible to reduce the leakage of reverse-biased silicon diodes so much that a silicon diode could store charge for the 1/30-second frame time required for television. This development permitted silicon to be used for both camera tubes and solid-state sensors such as bucket brigades and charge-coupled devices.

But silicon was taking over. The whole industry, Weimer recalls, was working on silicon in tubeless sensors, though it appeared, at first, to be too conducting. Bell Labs found a way to insulate the silicon enough so that you could make a silicon vidicon. Eventually, Weimer's group started using silicon, but it was too late. In retrospect, says Weimer, "We kept at the TFT using early semiconductors longer than we should have."

Work on the self-scanned solid-state sensor for nearly seven years had been remarkably rewarding in spite of its instability problem with early semiconductors. This project yielded many patents for Weimer and members of his group as well as technical recognition and many award-winning papers.

Maybe for large-area displays

While the concentration was still on camera-tube replacements, Weimer recognized that, if he could build transistors on glass, this might make for large-area displays. But, he reports, his group was too small to pursue this avenue. And perhaps, by the time he realized that the ideal application for TFTs was displays, not cameras, RCA management became disenchanted and terminated the joint effort with Wright Patterson, just as Westinghouse terminated Brody's efforts. But RCA continued to support Weimer's work on bucket-brigade ICs and CCDs. (Brody is now president of a consulting group, Active Matrix Associates.)

One of the key members of Weimer's group, Harold Borkan, recalls that the thin-film-transistor work may have been a bit too early in history. It would likely have been much more successful at a later time when better semiconductor materials were available.

In 1980, after 30 years at RCA, Borkan was appointed director of microelectronics at the Army's Electronic Technology and Devices Laboratory at Fort Monmouth, N.J. His responsibility included flat-panel displays for military applications, with emphasis on electroluminescence rather than liquid-crystal technology because of brightness considerations. He was later promoted to the position of deputy director of the entire laboratory, responsible for practically all of the Army's electronic-device R&D.

The thin-film transistor for scanning solid-state sensors died. What about displays? The concentration was in other areas. So thin-film transistors for displays also died.

Well, not quite. In the 1990s, it became apparent that the brightness of liquid-crystal displays could be greatly enhanced by incorporating a thin-film transistor at each pixel element in the display. It became practical using amorphous silicon as the semiconductor.

Liquid crystals grew up from displaying a few characters in digital instruments, wristwatches and calculators. They moved into large-area displays for laptop computers, a market far beyond the camera-tube-replacement dreams of Weimer and Borkan. Laptop displays were a natural for thin-film transistors, now in silicon. Liquid-crystal displays brought thin-film transistors back to life, a very vibrant life. It was a perfect marriage.

The Century of the Engineer: Misunderstood Milestones