The name permeable-base transistor was initiated by Bozler et al. when they presented results on GaAs material with tungsten grids in 1979. Structures on Si with CoSi2 grids were reported by Rosencher et al. in 1986. These two structures have well-defined metal grids, patterned by lithography, that are completely embedded in the semiconductors. Slightly different structures can be found in literature earlier. Lindmayer in 1960 used a thin metal film of molybdenum and utilized the small, natural pinholes as channel openings. He called this device a metal-gate transistor. This technique was also examined more recently, using epitaxial CoSi2 for Si devices and tungsten for GaAs devices. To avoid epitaxial overgrowth of semiconductor on metal, Wright in 1960 mentioned a trenched semiconductor structure where the subsequently deposited metal grids were not completely embedded in the semiconductor. This structure has been reported and analyzed more vigorously by Rathman. Up to now, the permeable-base transistor remains a subject for research and is not produced commercially.
The permeable-base transistor is basically a vertical MESFET with a very short channel length. The structure shown has completely embedded metal grids that are defined by lithography and etching. The line and space currently achievable are in the order of 0.2 nm. The channel doping is around 1016 cm-3 and is usually designed for an enhancement device such that the depletion width surrounding the metal grids at zero base voltage is larger than half the spacing between the metal strips.
The metal thickness becomes the channel length and thicknesses of a few hundred Angstroms have been used. The metal should be chosen to yield high Schottky-barrier height to minimize forward-bias leakage, and it should be a refractory metal or silicide that can withstand high temperature during subsequent epitaxial growth. For p-channel GaAs devices, tungsten has been used, and for Si devices, both CoSi2 and molybdenum have been used. The technology for epitaxial overgrowth of semiconductor on metal is not trivial. The mechanism is first by vertical growth of the semiconductor over the window, followed by lateral growth over the metal. The control of the doping level in the area directly over the metal is sometimes a practical problem. The epitaxial film thicknesses for the p-type emitter and collector layers are in the order of 0.2 nm and 1.0 nm, respectively. In the structure the metal grids are not defined, but are natural random pinholes (~ 100 A) that exist in a very thin metal or silicide film whose thickness is less than 100 A. The channel openings, in this case, are much narrower and the device operation regime will be different, as discussed later. The third structure shown eliminates the need for epitaxial overgrowth of a semiconductor on metal. The semiconductor trench is created by lithography and anisotropic etching. The depth of the trench is typically around 0.5 .
The terminals of a permeable-base transistor have been given the nomenclatures of a potential-effect transistor (emitter, base, collector) because of its close structural resemblance to a metal-base transistor. Under normal operation, the transistor behaves like a MESFET with very short channel length. The emitter, base, and collector terminals can be described more accurately as the source, gate, and drain, respectively, of an FET. The channels are the vertical gaps between the metal grids, and the effective channel width is controlled by the depletion width around the metal base.When the depletion regions are merged together, the channel is cut off. When the depletion width is reduced by a forward base bias, a neutral region or channel opening exists and the transistor is considered to be on. The off-state is caused by the potential barrier near the emitter. The threshold voltage for the base (with the emitter grounded) is applied to eliminate this barrier potential.
.