A great deal of the early expansion of cellular communications and the increased data rates in telecommunications systems can be traced to the development of GaAs technology.
Microwave semiconductors have come a long way in a short time. After all, the transistor (based on silicon) was only invented in 1948, and the integrated circuit (IC; also in silicon) in 1959. But the real push toward microwave frequencies came with the development of gallium arsenide (GaAs) epitaxial materials, discrete devices, and ICs in the 1980s, thanks to the US Department of Defense (DoD).
Through the early part of the 1980s, the majority of transistors for low-noise (receivers) and higher-power (transmitters) applications were still based on silicon, although GaAs has six times the electron mobility of silicon. But GaAs wafers were primitive at the time, prompting the DoD and the Defense Advanced Research Projects Agency (DARPA), via its legendary Gaas Microwave/Millimeter-wave Monolithic Integrated Circuits (MIMIC) program, to throw money at the development of both analog and digital GaAs devices. DARPA's intent was to reduce the cost and increase the reliability of Gaas components for communications, radar, and electronic-warfare (EW) applications.
In 1980, Dr. Joe Barrera, R&D Section Manager of Hewlett-Packard Co.'s Microwave Semiconductor Division, reported on his company's work in GaAs field-effect-transistor (FET) device technology, based on 0.5-m gate-length transistors. while noise figures for these transistors were approaching 2 dB at 18 GHz, the goal was to develop 0.5-m transistors capable of noise figures below 0.5 dB at 4 GHz.
Low-noise GaAs FETs such as Hewlett- Packard's model HFET-2204 were being used in amplifiers through 6 GHz with noise figures of less than 1 dB. But Barrera explained that transistor gate lengths would have to be reduced from 0.5 m to the region of about 0.1 m to lower capacitance adequately where significant reductions in noise figure could be made.
GaAs power got a boost in 1983 with the development of the 88000 series of power Gaas FETs by Microwave Semiconductor corp. (Somerset, NJ). The model 88300 offered 80 mw minimum output power and 100 mw typical output power at 20 GHz while the model 88302 delivered at least 400 mw output power and typically 500 mw output power at 20 GHz.
By the mid-1980s, Pacific Monolithics (Sunnyvale, CA) was exploring ways to produce GaAs ICs in high volumes for communications applications, including receiver chips (with approximately 3100 chips per 3-in. wafer). The company instituted 100% DC screening of all GaAs wafers; it had developed a technique to test chips that were randomly sampled from all areas of a wafer, using an automatic measurement system based on a model HP 8757A scalar network analyzer (SNA) from Hewlett-Packard Co.
And by 1988, TriQuint Semiconductor (Beaverton, OR) had developed its QED/A GaAs IC process, which was capable of fabricating analog RF/microwave circuits and large-scale-integration (LSI) digital circuits on the same wafer and chip (see figure). These were low-power devices that consumed a fraction of the power of the fastest devices at that time. The GaAs process boasted transition frequencies exceeding 18 GHz and featured both enhancement- and depletion-mode Gaas MESFETs, metal-insulator-metal (MIM) capacitors, and nickel-chromium (NiCr) resistors.