Prior to the 1950s, SiC and II-VI semiconductors had been well-known materials. Many II-VI semiconductors, e. g. ZnS and CdS, occur in nature. The very first LEDs had been made using SiC and there had been one publication by Destriau (1936) reporting LEDs made of zincblende (ZnS).

The era of III-V compound semiconductors started in the early 1950s when this class of materials was postulated and demonstrated by Heinrich Welker (1952, 1953). The class of III-V compounds had been an unknown substance prior to the 1950s and these compounds do not occur naturally. The novel manmade III-V compounds proved to be optically very active and thus instrumental to modern LED technology.

Bulk growth of the III-V compound GaAs commenced in 1954. In the mid 1950s, large single-crystal boules of GaAs were pulled from the melt. The sliced and polished wafers were used as substrates for the epitaxial growth of p-n junction diode structures, either by vapor-phase epitaxy (VPE) or liquid-phase epitaxy (LPE). Infrared (870-980 nm) LEDs and lasers based on GaAs were first reported in 1962 by groups working at RCA, GE, IBM, and MIT (Hall et al., 1962; Nathan et al., 1962; Pankove and Berkeyheiser, 1962; Pankove and Massoulie, 1962; Quist et al., 1962).

A sustained research effort on GaAs and AlGaAs/GaAs devices started in the early 1960s at the IBM Thomas J. Watson Research Center in Yorktown Heights, located about an hour’s drive north of New York City. The IBM team consisted of well-known researchers such as Jerry Woodall, Hans Rupprecht, Manfred Pilkuhn, Marshall Nathan, and others.

Woodall (2000) recalls that his work centered on the bulk crystal growth of GaAs used to fabricate semi-insulating substrates for Ge device epitaxy, and n-type substrates to fabricate injection lasers via Zn diffusion. At that time, the GaAs-based injection laser had already been demonstrated at IBM, GE, and MIT Lincoln Laboratories. Rupprecht’s interests were in impurity-diffusion theory and experiment along with experimental investigations into the newly discovered injection laser. Rupprecht was associated with a laser device physics group headed by Marshall Nathan, a co-inventor of the first injection laser (Nathan et al, 1962).

As Woodall developed a technique that lead to state-of-the-art horizontal Bridgman GaAs crystals, Rupprecht fabricated the materials into lasers and characterized them. This collaboration paid off immediately and continuous-wave (cw) operation of GaAs lasers at 77 K was attained (Rupprecht et al., 1963). They then learned of the liquid-phase epitaxy (LPE) technique pioneered by Herb Nelson at the RCA Laboratories in Princeton. The employment of LPE to grow GaAs lasers resulted in the achievement of 300 K lasers with lower threshold current densities than for Zn-diffused lasers. Stimulated by papers found in a literature search, Woodall set out to grow GaAs p-n junction diodes by using Si as an amphoteric dopant, i. e. Si atoms on Ga sites acting as donors and Si atoms on As sites acting as acceptors. This was an interesting idea, as hitherto LPE had been used to grow epilayers with only a single conductivity type.

The LPE conditions to form Si-doped p-n junctions were found very quickly. Si-doped GaAs p-n junctions were formed by cooling a Ga-As-Si melt from 900 to 850 °C to form Si donors and Si acceptors at the two temperatures, respectively. By examining the cross section of the chemically stained epitaxial layer, the lower layer, grown at 900 °C, was identified as being an n-type layer and the upper layer, grown at 850 °C, as a p-type layer. No loss in crystal quality was found in the regions of lower temperature growth. Furthermore, owing to band tailing effects caused by the highly doped, compensated region of the p-n junction, the LED emission occurred at 900-980 nm, far enough below the GaAs band edge (870 nm), so that the bulk GaAs substrate and the GaAs epilayer did not absorb much of the emitted light but acted as a transparent “window layer”. LED external quantum efficiencies as high as 6% were attained, a major breakthrough in LED technology (Rupprecht et al., 1966). Rupprecht (2000) stated: “Our demonstration of the highly efficient GaAs LED is a typical example of a discovery made by serendipity.” The quantum efficiency of the amphoterically doped GaAs LEDs was five times greater than that of GaAs p-n junctions fabricated by Zn diffusion. Si acceptor levels are deeper than Zn acceptor levels so that the emission from the compensated Si-doped active region occurs at longer wavelengths where GaAs is more transparent.

Being in the LED research business, the IBM group wondered if this doping effect could be extended to a crystal host with visible emission. There were two candidates, GaAsP and AlGaAs.

Whereas Rupprecht tried to do GaAsP epitaxy via LPE, Woodall set up an apparatus for AlGaAs. It was difficult to form good quality GaAsP epilayers by LPE due to the 3.6% lattice mismatch between GaP and GaAs. AlGaAs had problems of its own: “AlGaAs is lousy material” was the pervasive opinion at that time, because, as Woodall (2000) stated, “aluminum loves oxygen”. This results in the incorporation of the “luminescence killer” oxygen in AlGaAs; in particular, in the vapor-phase epitaxy (VPE) process, but less so in the LPE process.

Without the support of IBM management, Rupprecht and Woodall “went underground” with their research, conducting the LPE AlGaAs epigrowth experiments after regular working hours and on the weekends. Woodall designed and built a “vertical dipping”-type LPE apparatus, using graphite and alumina melt containers. As an undergraduate student Woodall had majored in metallurgy at MIT and he remembered something about phase diagrams. He made an “intelligent guess” to select the Al concentrations for the LPE melts. He added Si to the melt for the first experiment, saturated the melt and then “dipped” the GaAs substrate while cooling the melt from about 925 to 850 °C. Finally, the substrate and epilayer were withdrawn from the melt, and the apparatus was returned to 300 K. Although no Si-doped p-n junction was observed, a 100 thick high-quality layer of AlGaAs had been grown with a bandgap in the red portion of the visible spectrum (Rupprecht et al., 1967, 1968).

Visible-spectrum AlGaAs LEDs were also grown on GaP, a lattice mismatched but transparent substrate. Micrographs of the structure are shown in Fig. 1.3. When AlGaAs was grown on GaP substrates, the thermodynamics of LPE made the initially grown material Al - richer due to the Al distribution coefficient in the melt. As a result, the high-Al-content AlGaAs acts as a transparent window layer for the light emitted from the low-Al-content AlGaAs active region (Woodall et al., 1972).

Fig. 1.3. (a) Cross section micrograph of AlGaAs LED grown on transparent GaP substrate, (b) Electro­luminescence originating from current-injected region located under stripe-shaped contact viewed through transparent GaP substrate (after Woodall et al., 1972).

Pilkuhn, also an “IBM’er” who had worked with Rupprecht on GaAsP LEDs and lasers (Pilkuhn and Rupprecht, 1965), had built a small battery-powered circuit with an LED emitting
visible red light, which he showed to his colleagues and management at IBM (Pilkuhn, 2000). The reactions ranged from “nice but useless” to “great and useful”. However, it was soon realized that the latter was true, i. e. that LEDs were extremely useful devices. The first application of the GaAsP LEDs was as indicator lights on circuit boards, where the LEDs indicated the status and proper function of the circuit board. LEDs were also used to show the status of the data processing unit of the classic IBM System 360 mainframe computer shown in Fig. 1.4.

Fig. 1.4. This classic 1964 mainframe computer IBM System 360 used high-voltage gas-discharge lamps to indicate the status of the arithmetic unit. In later models, the lamps were replaced by LEDs. The cabinet-sized 360 had a performance comparable to a current low-end laptop computer.

According to Rostky (1997), the first commercial GaAs LED was offered by the Texas Instruments Corporation in the early 1960s. The LED emitted infrared radiation near 870 nm. The manufacturing quantities of the product were low, probably caused by the high price for one LED, which reportedly was 130 US$.

The resonant-cavity light-emitting diode (RCLED) was first demonstrated in the AlGaAs/GaAs materials system (Schubert et al., 1992, 1994). RCLEDs represented a new class of LEDs making use of spontaneous emission enhancement occurring in microscopic optical resonators or microcavities. Enhancement is greatest for wavelengths in resonance with the fundamental mode of the cavity. The emission enhancement is mediated by changes in the optical mode density within the cavity. RCLEDs have higher emission intensities along the optical axis of the cavity, which allows for higher coupling efficiencies to optical fibers.

At the present time, infrared GaAs/AlGaAs LEDs are widely used in video and audio remote controls and as sources for local-area communication networks. In addition, red AlGaAs/AlGaAs LEDs are used as high-brightness visible-spectrum LEDs with efficiencies higher than the

GaAsP/GaAs red LEDs but lower than the AlGaInP/GaAs red LEDs.