Drift Transistor

A drift transistor is a bipolar junction transistor whose base is doped unevenly, heavily near the emitter and lightly near the collector, so that a built-in electric field pushes minority carriers across the base instead of leaving them to wander by diffusion alone. In a uniform base the carriers injected by the emitter move only because there are more of them at one side than the other, which is slow and random; in a graded base the doping gradient itself acts like a constant downhill slope. The result is a much shorter base transit time and therefore a much higher cutoff frequency from the same physical base width. The idea is associated with Herbert Kroemer, whose theory of diffusion and drift transistors appeared in 1953, and it quietly became a feature of nearly every high-speed bipolar transistor made afterward.

The core physics is that a doping gradient creates an electric field without any external voltage. In an n-p-n transistor with the acceptor concentration falling from emitter to collector, the majority holes tend to diffuse toward the lightly doped side, but the ionized acceptors they leave behind hold them back, and the equilibrium that results is a small electric field of roughly (kT/q) times the logarithmic slope of the doping, pointing in the direction that sweeps electrons toward the collector. At room temperature kT/q is about 26 millivolts, so a base whose doping varies by a factor of 100 from one side to the other produces, across a base a micrometre thick, a field on the order of a thousand volts per centimetre. Electrons in that field cross the base noticeably faster than diffusion alone would carry them, with the field adding a steady directed velocity to their random thermal motion.

The quantity that improves is base transit time, the delay between a carrier entering the base from the emitter and reaching the collector. For a uniform base of width W and a minority-carrier diffusion coefficient D, the transit time is approximately W² divided by 2D, which is why thin bases are so valuable: halving the width cuts the delay by four. With an exponentially graded base the transit time becomes roughly W² divided by η × D for large field factors, where η is the natural logarithm of the doping ratio across the base, and the exact expression gives about a third of the uniform value for a hundred-fold ratio. Since the cutoff frequency fT is set largely by the transit time together with the junction charging times, a drift-field design can reach frequencies that a uniform-base design would only reach with a much more difficult, thinner layer.

Kroemer proposed the structure in theory, and companies including Bell Labs and RCA built working versions during the mid-1950s. Kroemer was a young German physicist whose 1953 analysis was published in German before he joined RCA Laboratories in Princeton in 1954, and he later shared the 2000 Nobel Prize in Physics for work on semiconductor heterostructures. What made the drift idea practical was the arrival of diffusion technology: a dopant diffused into a wafer from the surface naturally produces a concentration that falls steadily with depth, so a diffused base is graded by default. RCA's germanium drift-field transistors and the diffused-base devices described by Bell Labs researchers belong to this lineage, and the name drift transistor was commonly used for these devices in the 1950s.

A frequent misconception is that the drift field accelerates carriers throughout the transistor, when it acts only within the base. The emitter-base and collector-base depletion regions have their own fields, which inject and collect carriers, but those regions are not part of the drift-transistor idea; what the graded doping adds is a field in the neutral base itself, where an ordinary transistor has essentially none. A second pitfall is assuming that the graded base is free. Heavier doping at the emitter side lowers the base resistance in a helpful way, but the lightly doped collector side makes the base more vulnerable to punch-through, where the collector depletion region reaches the emitter junction at high collector voltage, so designers trade speed against voltage rating.

The term has faded but the principle has not. Few modern data sheets call any transistor a drift transistor because nearly all bipolar transistors made by diffusion or ion implantation have a graded base, so the label stopped distinguishing one product from another. The same reasoning appears again in silicon-germanium heterojunction bipolar transistors, where grading the germanium content across the base creates a built-in field through the band structure rather than through doping, a direct descendant of Kroemer's idea that earned his heterostructure work its later recognition. It also appears in power devices, where graded doping shapes electric fields to raise breakdown voltage.

Read drift transistor as the realization that you can build a ramp into the base itself, so carriers slide across it instead of having to diffuse.