Bipolar Transistor
The bipolar junction transistor replaced the point-contact transistor's two hand-placed tungsten whiskers with something a furnace could reproduce on every single unit: a diffused p-n junction whose width is a controllable process parameter instead of an assembly tolerance. It is a three-terminal device in which a small base drive controls a much larger collector current through the exact same minority-carrier injection mechanism Bardeen and Brattain's device relied on — just set by chemistry and heat instead of a steady hand under a microscope.
Structure and gain. An NPN BJT has a heavily doped emitter, thin lightly doped base, and collector engineered for voltage. Forward bias of the base-emitter junction injects electrons across the base; most diffuse to the reverse-biased collector junction rather than recombine. The fraction that survives the crossing is the base transport factor, and because the base is deliberately made far thinner than the minority-carrier diffusion length — $W_B \ll L_n$ — almost all of them make it:
That is the point-contact transistor's own $\alpha \approx e^{-s/L_p}$ relationship, just evaluated in the regime where the spacing is small compared to the diffusion length instead of comparable to it — the same physics, deliberately pushed to the opposite extreme. Collector current follows base current through
so a transport factor close to 1 turns into a current gain in the hundreds, which is why a BJT amplifies at all: a small base current controls a much larger collector current, and the transconductance $g_m = I_C/V_T$ gives high gain per unit bias current — a central analog advantage. PNP devices reverse carrier and voltage polarities but obey the same relationships.
Operating regions. Cutoff leaves both junctions effectively reverse biased and little collector current flows. Forward active operation forward-biases base-emitter and reverse-biases base-collector for amplification. Saturation forward-biases both junctions, reducing collector-emitter voltage but storing charge that slows turn-off. Reverse-active operation swaps collector and emitter with poor gain in an asymmetric device. Breakdown, quasi-saturation, Early effect, base-width modulation, self-heating, and high-current crowding bound useful operation.
BJT versus MOSFET. A MOSFET is voltage-controlled in steady state and offers enormous input resistance, easy dense digital integration, and no minority-carrier storage in normal conduction. A BJT needs base current and has lower input impedance, but achieves high transconductance, predictable exponential ratios, and strong RF performance at a given current. Bipolar and CMOS devices coexist in BiCMOS processes. SiGe HBT grading accelerates base transport and supports very high fT and fmax for low-noise amplifiers, mixers, oscillators, and wireline drivers.
Circuits and applications. Common-emitter stages provide voltage gain with phase inversion, emitter followers buffer impedance, differential pairs convert voltage differences to currents, and current mirrors distribute bias. BJTs form the VBE and ΔVBE elements of bandgap references and accurate thermal sensors. Darlington arrangements multiply current gain, while Baker clamps or Schottky clamps avoid deep saturation. RF circuits depend on noise figure, linearity, matching, stability, and package parasitics in addition to transistor speed.
| Characteristic | BJT | MOSFET | Design implication |
|---|---|---|---|
| Control variable | Base current / VBE | Gate voltage / charge | Different driver and bias networks |
| Input impedance | Moderate | Very high at DC | BJT loads its source |
| Transconductance at equal current | Typically higher | Depends on inversion level | BJT offers strong analog gain efficiency |
| Switching storage | Minority-carrier charge in saturation | Gate and node capacitance | Avoid deep BJT saturation for speed |
| Integration density | Lower for comparable digital logic | Extremely high | MOSFET dominates logic |
| RF specialization | SiGe HBT excels at gain and noise | RF CMOS integrates densely | BiCMOS combines strengths |
That reproducibility — not any difference in the underlying physics — is the entire reason the junction transistor won. Bardeen and Brattain's device and Shockley's both inject minority carriers on one side of a gap and collect whatever survives the diffusion across it; the point-contact device just measured that gap with a human hand and a microscope, while the junction transistor measures it with a thermocouple and a clock. $W_B$ can be held to the same value run after run in a way that $s$ never could, which is the one property a laboratory curiosity needed to become a manufacturable component — and later, once that same diffused-junction idea was transferred onto a flat oxidized wafer surface, the direct ancestor of the planar process that every integrated circuit since has been built on.
Read the junction transistor through a *process-parameter-not-assembly-tolerance* lens rather than a *better-amplifier* lens: $\alpha_T \approx 1-\tfrac{1}{2}(W_B/L_n)^2$ is the same minority-carrier diffusion physics the point-contact transistor already proved out, and the only thing that changed is who, or what, gets to set the critical dimension — a furnace instead of a hand, which is the difference between one working prototype and an entire industry.