In short: An NPN BJT works because a forward-biased base-emitter junction injects electrons into a very thin p-type base, where most of them diffuse across and are swept into the collector by the reverse-biased base-collector junction, so a small base current Ib controls a collector current Ic = beta x Ib.
The physical structure: three layers, three doping levels
An NPN transistor is an n-p-n sandwich: an n-type emitter, a thin p-type base and an n-type collector. The layers are deliberately unequal, and the inequality is the whole design. The emitter is heavily doped (around 1019 to 1020 cm-3) so it has an enormous supply of electrons. The base is moderately doped (roughly 1017 cm-3) and extremely thin: about 1 µm in an old discrete part, and well under 100 nm in a modern SiGe HBT. The collector is lightly doped (around 1015 to 1016 cm-3) and physically larger, so it can tolerate high reverse voltage and dissipate heat.
Because of these ratios the device is not symmetric. If you swap emitter and collector leads, you still get a transistor, but beta collapses to a value like 1 to 5. This is why datasheet pin order matters and why the "reverse-active" mode is almost never used.
Carrier flow in the active region
"Active" means the base-emitter junction is forward biased (Vbe near 0.65 V for silicon at room temperature) while the base-collector junction is reverse biased (Vce clearly above about 0.3 V). Follow the electrons step by step:
- Injection: forward bias lowers the emitter-base barrier, and electrons from the n+ emitter flood into the base. Holes flow the other way too, but there are far fewer of them because the base is lightly doped compared with the emitter.
- Diffusion: inside the base these electrons are minority carriers. They move by diffusion toward the collector because the concentration at the collector edge is almost zero.
- Collection: at the collector depletion edge, the strong field of the reverse-biased junction sweeps them into the collector. Nothing else is needed; the collector simply "catches" them.
- Loss: a small fraction (typically 0.5 to 2 percent) recombines with holes in the base. Those holes must be resupplied through the base terminal, and that resupply is the base current Ib.
So the base terminal does not carry the main current; it only pays the recombination "tax" and the small hole injection into the emitter. The collector current is set by how many electrons the emitter injects, which is set exponentially by Vbe.
Where Ic = beta x Ib comes from
The collector current follows the Ebers-Moll relation Ic = Is · exp(Vbe / Vt), where Vt = kT/q is about 25.85 mV at 300 K and Is is a saturation current of roughly 10-14 to 10-15 A for a small-signal device. Every 60 mV increase in Vbe multiplies Ic by about ten, and Vbe falls about 2 mV per degree Celsius at fixed current.
Base current is a fixed small fraction of that: Ib = Ic / beta. Beta (hFE) is therefore the ratio of "electrons delivered to the collector" over "holes consumed at the base". Two things set it: emitter injection efficiency (doping ratio emitter versus base) and base transport factor (how few electrons recombine, which depends on base width relative to the electron diffusion length). A thinner base and a heavier emitter both raise beta. Typical values: 2N3904 around 100 to 300, a power transistor like TIP31 around 25 to 100, and SiGe RF devices higher still.
Key fact: the emitter current is Ie = Ic + Ib = (beta + 1) Ib, and alpha = Ic/Ie = beta/(beta + 1). For beta = 100, alpha is 0.990, which means 99 percent of emitter electrons reach the collector.
A worked example
Take a 2N3904-style NPN, Vcc = 10 V, Rc = 3.3 kΩ, and assume beta = 100. Drive the base with Ib = 20 µA.
- Ic = 100 x 20 µA = 2 mA
- Drop across Rc = 2 mA x 3.3 kΩ = 6.6 V
- Vce = 10 - 6.6 = 3.4 V, which is above Vce(sat), so the device is active and the relationship holds
- Ie = 2.02 mA
Now raise Ib to 100 µA. The formula predicts 10 mA, which would drop 33 V across Rc, impossible with a 10 V supply. Instead Vce falls to about 0.2 V, the collector junction becomes forward biased, and the transistor saturates with Ic limited to roughly (10 - 0.2)/3.3 k = 2.97 mA. Beta no longer applies; the effective "forced beta" is only about 30. This is the practical boundary: Ic = beta x Ib is valid only in the active region.
Operating regions compared
| Region | B-E junction | B-C junction | Behaviour |
|---|---|---|---|
| Cutoff | Reverse / below 0.5 V | Reverse | Ic is about 0, switch open |
| Active | Forward | Reverse | Ic = beta x Ib, amplifier |
| Saturation | Forward | Forward | Vce about 0.2 V, switch closed |
| Reverse active | Reverse | Forward | Beta about 1 to 5, rarely used |
Key fact: beta is not a constant. It varies with Ic (it peaks mid-range and drops at high current), with temperature (rising roughly 0.5 to 1 percent per degree C) and from part to part by 3:1 or more. Robust circuits are biased so the result barely depends on beta.
Second-order effects interviewers like
Early effect: as Vce rises, the B-C depletion region widens, shrinking the effective base width. Ic then increases slightly with Vce, modelled as Ic = Is exp(Vbe/Vt)(1 + Vce/Va) with Early voltage Va of 50 to 150 V. This gives finite output resistance ro = Va/Ic, which limits single-transistor gain to about Va/Vt, the intrinsic gain limit.
Transconductance: gm = Ic/Vt, so 1 mA gives about 39 mS, and the small-signal input resistance is rπ = beta/gm. Dividing both sides tells you why bias stability matters: gm depends only on current, but input resistance scales with beta.
Design practice: in SPICE-class tools such as Cadence Spectre or Siemens Eldo/Analog FastSPICE, BJTs use the Gummel-Poon or VBIC models; the parameters BF (forward beta), IS, VAF and IKF capture exactly the effects above. Use a DC operating-point analysis to confirm Vce stays above saturation before trusting any gain number. For a deeper look at the curves, see our article on BJT transistor characteristics, and for biasing methods read BJT biasing circuits.
BJTs are still chosen for high-speed analog and RF where low noise and high gm per unit current matter. Teams building such mixed-signal blocks, for example the design services group at Avecas, often still use bipolar devices in bandgap references and RF front-ends.
Want to go from device physics to full amplifier design? Explore the hands-on analog modules in the CourseTron courses and practise biasing, small-signal analysis and simulation with guided labs.
Frequently asked questions
Why is the base made so thin?
A thin base means electrons cross it before they can recombine, which pushes the base transport factor toward 1 and makes beta high. A thick base would behave like two back-to-back diodes and most injected electrons would be lost.
Is beta the same as gain?
No. Beta is the DC current ratio Ic/Ib. Voltage gain of an amplifier depends on gm and the load, for example Av = -gm x Rc in a simple common-emitter stage, and beta only enters through input resistance.
Can a BJT be used as a switch?
Yes. Driven into saturation with a base current of about Ic/10, it gives Vce(sat) near 0.1 to 0.3 V. The trade-off is slower turn-off due to stored base charge.
Does the base current flow into or out of the transistor?
For an NPN, conventional base current flows into the base terminal, and Ie flows out of the emitter. Electron motion is opposite to conventional current, which often confuses beginners.