High-Frequency Transistor
A high-frequency transistor is one built so that it still amplifies at radio, VHF, UHF and microwave frequencies, where ordinary devices lose their gain. Every transistor has a frequency above which its current gain falls toward one, and the high-frequency types are those engineered to push that limit far beyond the audio range: narrow active regions, tiny junction capacitances, low parasitic resistances, and layouts that keep signal paths short. They sit inside radios, television tuners, mobile phones, radar and satellite links.
The standard figure of merit is the transit frequency, fT, the frequency at which a transistor's small-signal current gain has fallen to unity. Below that point the gain drops by roughly six decibels per octave, so a device with an fT of 1 gigahertz has a current gain of about 10 at 100 megahertz, a rule of thumb that lets designers estimate usable gain at a given frequency. For a bipolar transistor fT is set by the time carriers need to cross the base plus the time to charge the junction capacitances, and is often written as 1 / (2π × total delay). The base transit time is proportional to the square of the base width divided by the carrier diffusion coefficient, so halving the base width roughly quarters that delay. Related figures include fmax, the maximum frequency of oscillation, which depends also on base resistance and feedback capacitance and indicates where power gain disappears.
Early high-frequency germanium devices were physically thinned or sculpted to shorten the base. The alloy-junction transistors of the early 1950s were limited to a few megahertz by their thick bases. Philco's surface-barrier transistor, introduced in 1953, used jets of electrolyte to etch and then plate a very thin base region from both sides, reaching tens of megahertz and finding use in early transistor computers. Diffusion changed the game: by diffusing dopants into a wafer, the base width could be controlled to a fraction of a micrometre. The mesa transistor, with its etched pedestal that cuts junction area, was commercialized in the late 1950s, and the planar process of 1959 allowed the same fine diffused bases with protected junctions. Diffused-base silicon types such as Fairchild's 2N697 family, though mainly made for switching, show how this technique carried speed into mainstream parts.
Capacitance and layout matter as much as the base thickness. The collector-base capacitance feeds the output back to the input and, through the Miller effect, can multiply into a large effective input capacitance, so RF transistors are built with small junction areas, often as many narrow emitter stripes in parallel (interdigitated geometries) to keep base resistance low while keeping area small. Package parasitics also count: bond wire inductance and lead capacitance limit performance, which is why RF parts use short leads, special flanged ceramic packages, or are used bare in modules. Lower-noise devices add very low base resistance, because thermal noise in that resistance is amplified along with the signal.
Heterostructure devices extended the range from hundreds of megahertz to hundreds of gigahertz. In a silicon-germanium heterojunction bipolar transistor, a thin base of silicon alloyed with germanium has a narrower band gap than the silicon emitter, so the base can be doped very heavily for low resistance and still keep good gain, allowing fT beyond 100 gigahertz in production processes developed in the 1990s and 2000s while remaining compatible with silicon manufacturing. The high electron mobility transistor, or HEMT, invented in Japan around 1980 (credited to Takashi Mimura at Fujitsu), forms a two-dimensional electron gas at a gallium arsenide and aluminium gallium arsenide interface, where electrons travel with little impurity scattering; HEMTs give very low noise and became staples of satellite receivers and radio astronomy front ends.
A frequent misconception is that fT equals the highest frequency at which the device is usable. The transit frequency is a measure of current gain and is extrapolated; usable power gain extends up to fmax, which may be higher or lower, and practical circuits operate well below either, since matching networks, stability and noise limit the design. Another pitfall is thinking that a higher-fT part is always better: very fast devices have lower breakdown voltage, may oscillate unexpectedly in a circuit built for audio, and need careful bypassing. Choosing a transistor therefore means balancing gain, noise, power, voltage and cost at the intended frequency.
Modern relevance is hard to overstate, since the wireless world is built on such devices. Silicon CMOS now handles much of the radio at gigahertz frequencies in phones, with SiGe BiCMOS used for demanding front ends and automotive radar around 77 gigahertz, while gallium arsenide and gallium nitride devices power base stations and satellites. Each step followed the same logic that began with thinning the base of a germanium crystal: shorten the path and shrink the capacitance, and the transistor can follow a faster signal.
Read high-frequency transistor as a race against carrier delay and capacitance: fT summarizes how fast charge can cross and recharge the structure, and each generation, from surface-barrier germanium to SiGe and HEMT devices, is a new way to win that race.