Alpha Current Gain
Alpha (α) is the common-base current gain of a bipolar transistor: the ratio of collector current to emitter current, written α = IC / IE. Because the base steals a small slice of the emitter current on its way across, the collector always receives slightly less than the emitter sends, so alpha in a junction transistor sits just below 1 — typically 0.98 to 0.998 for ordinary small-signal silicon devices. It is the older of the two standard gain figures, the one in which the original transistor theory was written, and it describes the device as a nearly lossless current conveyor rather than a current multiplier.
Alpha is a product of two small losses, and understanding them explains why it can never reach exactly 1. The first is emitter injection efficiency: some of the emitter current is carried by carriers flowing backward from base into emitter, and these never head for the collector. Heavy doping of the emitter relative to the base keeps this fraction small. The second is the base transport factor: carriers that do enter the base must diffuse across it, and a few recombine on the way. A thin base (narrow compared with the diffusion length of minority carriers) keeps recombination low. Multiply the two, add a collector multiplication factor that is essentially 1 at normal voltages, and the result is alpha. Every refinement in transistor fabrication — thinner bases, graded doping, cleaner surfaces — is in effect an effort to push alpha a few thousandths closer to 1.
The tiny gap between alpha and 1 is exactly what produces the larger gain figure, beta. Since the emitter current splits into collector and base currents (IE = IC + IB), the base current is IE × (1 − α), and dividing collector current by base current gives β = α / (1 − α). The relation is steeply nonlinear near 1: α = 0.98 gives β = 49, α = 0.99 gives β = 99, and α = 0.998 gives β = 499. The reverse, α = β / (β + 1), shows why alpha barely moves while beta swings widely — a device whose beta doubles from 100 to 200 has changed its alpha only from about 0.990 to 0.995. Alpha is therefore a poor number for describing manufacturing spread, which is why datasheets quote beta instead.
Common-base gain of just under 1 is not a failure to amplify. In the common-base connection the input is applied at the emitter and the output taken from the collector, with the base held at AC ground. The current gain is α, but the emitter looks into a very low impedance (a few tens of ohms) while the collector looks like a high-impedance current source, so a modest current can develop a large output voltage across a load resistor. Voltage gain and power gain well above 1 are therefore available from a stage whose current gain is below 1. The point-contact transistor of December 1947 was different: its collector current could exceed the emitter current, giving an alpha greater than 1, an effect tied to the way carriers released at the point contact altered the collector region. That behaviour made early point-contact devices touchy and prone to instability, and the lower, better-behaved alpha of the junction transistor was part of what made it the practical winner.
Alpha also changes with operating conditions, though more gently than beta. It falls at very low currents, where recombination in the emitter-base depletion region becomes a larger share of the total, and again at high currents, where high-level injection reduces injection efficiency. It creeps up with collector voltage because the widening collector depletion region narrows the effective base — the Early effect, which is why a measured collector characteristic is never perfectly flat. Alpha also dips at high frequency: the cutoff frequency of alpha, f-alpha, is the point where it has fallen to about 0.707 of its low-frequency value, and was the figure of merit quoted for early germanium and alloy-junction transistors.
The main pitfall is mixing up the two gains. A schematic or datasheet that says "current gain of 0.99" is describing alpha, not a stage that attenuates a signal, while a beta of 100 is not an alpha of 100. Another common error is assuming alpha is a fixed property of a transistor type; it varies with current, voltage, temperature and part. Modern design rarely uses alpha directly, but it survives in compact models such as the Ebers-Moll equations, in the analysis of current mirrors (where the base-current error is a factor of 1 / (1 + 2/β)), and in the treatment of RF and cascode stages that use the common-base connection to avoid Miller capacitance.
Read alpha as the transistor's yield on a current hand-off: nearly every carrier the emitter launches reaches the collector, and the few that do not — lost to the base terminal — are precisely the small leakage that beta magnifies into a useful amplification factor.