Electrical Engineering › Section P · Semiconductor theory, diodes, and transistors › Topic P.3
Inside BJTs, JFETs, MOSFETs, and thyristors
Add a third doped layer to a diode, or a control electrode beside a channel, and the junction stops being a one-way gate and becomes a valve that one terminal can open and close. This topic looks inside the four great device families to see how each one controls current.
From one junction to a controlled valve
"Electrons, holes, the P-N junction, and junction diodes" showed that a junction conducts one way and that its depletion region widens or narrows with bias. Transistors and thyristors arrange those effects so a small input controls a large current. Bipolar devices, the BJT and the thyristor, inject carriers across forward-biased junctions and use both electrons and holes. Field-effect devices, the JFET and MOSFET, squeeze or create a channel with an electric field and use one kind of carrier, so they are called unipolar.
The bipolar junction transistor
Bell Labs demonstrated the first transistor, a germanium point-contact device built by John Bardeen and Walter Brattain, in December 1947; William Shockley worked out the more practical junction transistor within weeks, and the three shared the 1956 Nobel Prize in Physics. An NPN BJT has an n-type emitter, a p-type base and an n-type collector. Two features make it more than back-to-back diodes: the base is extremely thin, and the emitter is far more heavily doped than the base. The collector is lightly doped near the base so it can withstand tens of volts, while the emitter-base junction breaks down at only about 5 to 7 V in reverse.
How the BJT controls current
In normal, active operation the base-emitter junction is forward biased, about 0.7 V for silicon, and the collector-base junction is reverse biased. The emitter injects a flood of electrons into the base, where they are minority carriers. Because the base is so thin, almost all diffuse across before they can recombine, and the collector junction's strong field sweeps them into the collector. Only the few that recombine in the base, plus holes the base sends back into the emitter, form the small base current. Change the base current slightly and the large collector current changes in proportion.
Alpha and beta
Two ratios describe the split of currents. Since all the current enters or leaves through the emitter, KCL gives \(I_E = I_C + I_B\). The fraction of emitter current reaching the collector is \(\alpha = I_C/I_E\), typically 0.99 or more in a small-signal transistor. The current gain is \(\beta = I_C/I_B\), often written \(h_{FE}\) on datasheets, typically 50 to 300 for small-signal parts. The two are linked:
Because \(1 - \alpha\) is tiny, small changes in \(\alpha\) make large changes in \(\beta\), so β varies widely between transistors of one type.
Worked example: currents in an NPN
An NPN transistor in active mode has an emitter current of 10.0 mA and a base current of 0.100 mA. Find \(I_C\), α and β. 1. KCL: \(I_C = I_E - I_B = 10.0 - 0.100 = 9.90\) mA. 2. Alpha: \(\alpha = 9.90/10.0 = 0.990\). 3. Beta: \(\beta = 9.90/0.100 = 99.0\). 4. Check with the link: \(0.990/(1 - 0.990) = 99.0\). 5. Sensitivity: if a better-made transistor reaches \(\alpha = 0.995\), then \(\beta = 0.995/0.005 = 199\). Half a percent more emitter current reaching the collector doubles the gain, which is why circuits are designed not to depend on an exact β.
NPN and PNP
A PNP transistor swaps every layer: p-type emitter, n-type base, p-type collector, with holes as the injected carriers. All its bias polarities reverse: the base sits about 0.7 V below the emitter, and the collector is negative relative to the emitter. The emitter arrow on the symbol marks the base-emitter junction and points the way conventional current flows: out of the emitter of an NPN, into the emitter of a PNP. So current flows into an NPN's collector and base and out of its emitter. Discrete PNPs perform almost as well as NPNs, but integrated PNPs are poorer, so chip designers favour NPNs.
The junction field-effect transistor
Julius Lilienfeld patented field-effect ideas in the 1920s, decades before they could be built reliably. An n-channel JFET is a lightly doped n-type channel with a source at one end and a drain at the other, and heavily doped p-type gate regions on its sides. With no gate voltage the channel conducts, so the JFET is normally on. Reverse-biasing the gate-channel junction widens the depletion regions into the channel, narrowing it and raising its resistance. At the pinch-off voltage, typically a few volts, the channel closes. Because the gate junction is reverse biased, the gate draws almost no current. A p-channel JFET reverses the doping and every polarity.
The MOSFET
In the MOSFET, first made at Bell Labs by Mohamed Atalla and Dawon Kahng in about 1959, the gate is a conductor separated from the silicon by a thin layer of silicon dioxide. In an n-channel enhancement MOSFET, two n-type regions, source and drain, sit in a p-type body and are isolated with the gate at 0 V, so the device is normally off. A positive gate repels holes and attracts electrons under the oxide; above the threshold voltage they form an inversion layer, an n-type channel from source to drain. The insulated gate draws no steady current, only surges to charge its capacitance. MOSFETs are now the most numerous devices made, billions to a chip, and power MOSFETs dominate low-voltage switching.
Worked example: charging a MOSFET gate
A power MOSFET's datasheet gives a total gate charge of 40 nC to drive the gate from 0 to 10 V, and it switches at 100 kHz. 1. Charge: the gate takes 40 nC at each turn-on and returns it at each turn-off. 2. Average gate-drive current: \(I = Q_g f = (40 \times 10^{-9})(100 \times 10^{3}) = 4.0\) mA. 3. Driver power: \(P = Q_g V f = (40\,\text{nC})(10\,\text{V})(100\,\text{kHz}) = 0.040\) W. 4. Switching speed: a driver that can source 1.0 A charges the gate in about \(Q_g/I = 40\) ns. 5. Interpret: no steady gate current, but fast switching needs a driver that delivers brief, large pulses.
Thyristors
A thyristor has four alternating layers, P-N-P-N, between anode and cathode. Shockley proposed the four-layer diode in about 1950, and General Electric introduced the silicon-controlled rectifier (SCR) in the late 1950s. The SCR acts like a PNP and an NPN transistor with each collector feeding the other's base. A gate pulse turns on the NPN, which turns on the PNP, which drives the NPN harder: regeneration latches both on, and removing the gate signal changes nothing. The SCR stops only when its anode current falls below the holding current, as happens every half-cycle on ac. A gate turn-off thyristor (GTO) can also be switched off by a strong negative gate current. Wafer-sized SCRs handle kilovolts and thousands of amperes.
Common mistakes to avoid
First, treating a BJT as back-to-back diodes; without the thin base no carriers reach the collector. Second, treating β as fixed for a part number; it varies widely because it depends on \(1 - \alpha\). Third, drawing the emitter arrow the electron way; it shows conventional current. Fourth, thinking a JFET needs gate voltage to conduct; gate bias turns it off. Fifth, saying an insulated gate never draws current; it draws charging pulses. Sixth, expecting an SCR to turn off when the gate signal ends.
Key takeaways
A BJT has a heavily doped emitter, a very thin base and a lightly doped collector; with base-emitter forward biased and collector-base reverse biased, carriers injected by the emitter cross the base into the collector. I_E = I_C + I_B, α = I_C/I_E ≈ 0.99, β = I_C/I_B = α/(1 − α). JFETs are normally on and pinched off by reverse gate bias; enhancement MOSFETs are normally off and turned on by gate voltage above threshold, drawing only charging current. Thyristors are P-N-P-N latches: a gate pulse turns them on, and they stay on until current falls below the holding current.
Flashcards
Flashcard
BJTs and thyristors use electrons and holes; FETs use one carrier
Test your understanding
Question 1 of 5
An NPN transistor in active mode has β = 150 and a base current of 40 µA. What are its collector and emitter currents?
Question 2 of 5
Why must a BJT's base region be made very thin?
Question 3 of 5
Which device conducts between its main terminals with zero gate-source voltage?
Question 4 of 5
A power MOSFET needs 25 nC of gate charge and switches at 200 kHz. What average current must the gate driver supply?
Question 5 of 5
An SCR in a dc circuit is triggered on, and then its gate signal is removed. What happens?
