Opening the book…
When an electron in a semiconductor is excited into the conduction band, it leaves behind an empty state in the valence band, called a hole. A hole behaves like a positive charge. When a field is applied, an electron in a neighboring bond can hop into the hole, which leaves a hole where that electron used to be; as the process repeats, the hole moves through the crystal in the direction a positive charge would move. A pure, or intrinsic, semiconductor has equal numbers of conduction electrons and holes, and both carry current. In pure silicon at room temperature there are only about \(1\times 10^{16}\,\mathrm{}\) of each per cubic meter, one for every five trillion atoms.
Doping. The properties of a semiconductor can be transformed by doping: adding a tiny, controlled amount of an impurity (Figure 42.5). Silicon has four valence electrons, each used in a bond to one of four neighbors. Replace one silicon atom with phosphorus, which has five. Four of phosphorus’s electrons form the four bonds, and the fifth is left over, bound only weakly to the phosphorus ion. It takes only about \(0.045\,\mathrm{eV}\) to free it into the conduction band, less than twice \(k_BT\) at room temperature, so nearly every phosphorus atom donates an electron. Phosphorus is called a donor, and silicon doped this way is n-type, because its current is carried mostly by negative electrons. Replace a silicon atom with boron instead, which has only three valence electrons, and one bond is left an electron short: a hole. Boron is an acceptor, and silicon doped this way is p-type, with its current carried mostly by positive holes.
Example 42.6: A pinch of phosphorus
Silicon has \(5.0\times 10^{28}\,\mathrm{atoms/m^3}\). A sample is doped with one phosphorus atom for every \(10^{7}\) silicon atoms, and each phosphorus atom donates one conduction electron. Compare the density of conduction electrons with that of pure silicon, \(1.0\times 10^{16}\,\mathrm{m^{-3}}\).
Solution
The donor density is
\[n_d = \frac{5.0\times10^{28}}{10^{7}} = 5.0\times 10^{21}\,\mathrm{m^{-3}},\]
so the conduction-electron density rises by a factor of
\[\frac{5.0\times10^{21}}{1.0\times10^{16}} = \boxed{5\times10^{5}}.\]
A doping level of one part in ten million, far below the impurity level of any “pure” chemical sold in a store, increases the conductivity half a million times. That is why semiconductor manufacturing demands silicon purified to better than one part in a billion before the dopants are added deliberately, and why the dopants themselves are placed with such care.
Checkpoint 42.5
Germanium, like silicon, has four valence electrons. Is germanium doped with arsenic (five valence electrons) n-type or p-type? What about germanium doped with gallium (three)?
Diodes
The real power of doping appears when p-type and n-type material meet in a single crystal, forming a p–n junction (Figure 42.6a). Electrons from the n side diffuse across the boundary and fill holes on the p side. This leaves a thin layer on the n side with positive donor ions and no free electrons, and a layer on the p side with negative acceptor ions and no holes. This layer, depleted of mobile carriers, is the depletion zone. Its fixed charges create an electric field pointing from the n side to the p side, and a potential difference of about \(0.7\,\mathrm{V}\) in silicon, that stops further diffusion.
Now connect the junction to a battery. If the p side is made positive, called forward bias, the applied voltage opposes the built-in field and shrinks the depletion zone. Electrons pour across from the n side and holes from the p side, and a large current flows. If the p side is made negative, called reverse bias, the applied voltage adds to the built-in field, the depletion zone widens, and almost no current flows. The junction is a diode: a one-way valve for current. The current through an ideal diode at voltage \(V\) is
Key Equation: The diode equation
\[I = I_0\left(e^{eV/k_BT} - 1\right), \tag{42.6}\]
where \(I_0\) is a small saturation current, typically \(10^{-12}\) amperes or less for a silicon diode. For reverse bias (\(V < 0\)) the exponential quickly vanishes and \(I \to -I_0\), a negligible trickle. For forward bias the current grows by a factor of \(e\) for every \(k_BT/e = 25.9\,\mathrm{mV}\) at room temperature (Figure 42.6b). Diodes convert the alternating current of Chapter 26 to direct current in every phone charger.
Example 42.7: Turning on a diode
A silicon diode at \(300\,\mathrm{K}\) has \(I_0 = 1.0\times 10^{-12}\,\mathrm{A}\). Find the current at (a) \(V = +0.60\,\mathrm{V}\), (b) \(V = +0.66\,\mathrm{V}\), and (c) \(V = -5.0\,\mathrm{V}\).
Solution
At \(300\,\mathrm{K}\), \(k_BT/e = 0.02585\,\mathrm{V}\).
(a) \(eV/k_BT = 0.60/0.02585 = 23.2\), so
\[I = (1.0\times10^{-12})(e^{23.2} - 1) = (1.0\times10^{-12})(1.2\times10^{10}) = \boxed{12\,\mathrm{mA}}.\]
(b) \(eV/k_BT = 0.66/0.02585 = 25.5\), so
\[I = (1.0\times10^{-12})(1.2\times10^{11}) = \boxed{0.12\,\mathrm{A}}.\]
An increase of only \(60\,\mathrm{mV}\) multiplies the current by ten. Over a wide range of currents, a forward-biased silicon diode therefore holds a nearly constant voltage of about \(0.6\,\mathrm{}\) to \(0.7\,\mathrm{V}\).
(c) \(e^{-5.0/0.02585}\) is utterly negligible, so \(I = \boxed{-1.0\times 10^{-12}\,\mathrm{A}}\): a picoampere, eleven orders of magnitude smaller than in (a).
Light-emitting diodes and solar cells. In a forward-biased junction, electrons crossing into the p side drop into holes, falling across the band gap. In silicon the energy goes mostly into lattice vibrations. But in some semiconductors, such as gallium arsenide and gallium nitride, each such recombination usually emits a photon with energy close to \(E_g\). A junction made of such a material is a light-emitting diode, or LED, and its color is set by its band gap:
\[\lambda \approx \frac{hc}{E_g}. \tag{42.7}\]
Gallium arsenide (\(E_g = 1.42\,\mathrm{eV}\)) emits infrared at \(870\,\mathrm{nm}\), used in television remote controls. Blending in phosphorus raises the gap to give red LEDs, and the gallium nitride family yields blue and violet. The blue LED, whose invention earned the 2014 Nobel Prize in Physics, made white LED lighting possible; a white LED is a blue LED coated with a phosphor that converts part of its light to yellow. Put a pair of mirrors on the ends of a suitable junction and the stimulated emission of Section 41.7 turns the LED into a diode laser, the laser in every barcode scanner and fiber-optic link.
Run the process backward and the diode becomes a solar cell. A photon with \(hf \ge E_g\) absorbed near the junction creates an electron–hole pair, and the built-in field sweeps the electron to the n side and the hole to the p side. Charge builds up on each side and drives a current through an external circuit. Any photon energy beyond \(E_g\) is lost as heat, and photons with less than \(E_g\) pass through unused, which is why there is an optimum band gap for converting sunlight, close to silicon’s.
The Transistor
A diode lets current flow one way. A transistor lets a small signal control a larger current, and that makes it both an amplifier and a switch. The first transistor was built by John Bardeen, Walter Brattain, and William Shockley at Bell Laboratories in 1947. The kind used in nearly all modern electronics is the metal–oxide–semiconductor field-effect transistor, or MOSFET (Figure 42.7).
Two n-type regions, the source and the drain, are set into a block of p-type silicon. Between them, separated from the silicon by a layer of insulating oxide only a few nanometers thick, is a metal electrode called the gate. With no voltage on the gate, current cannot pass from source to drain: the path crosses p-type silicon, and whichever way the voltage is applied, one of the two p–n junctions along the path is reverse biased. Now make the gate positive. The gate and the silicon beneath it form a capacitor (Chapter 19), and the positive gate attracts electrons to the surface just under the oxide. Once the gate voltage is large enough, these electrons outnumber the holes in a thin layer, which becomes effectively n-type: a conducting channel connects source and drain, and current flows. The gate voltage controls the current, but because the oxide insulates it, the gate itself draws almost no current, so a tiny amount of power controls a much larger one. Used with a range of gate voltages, the MOSFET is an amplifier; switched between “off” and “on,” it is a switch, and a switch is all that is needed to represent the \(0\)s and \(1\)s of binary logic.
Checkpoint 42.6
A p-channel MOSFET is built the other way around: p-type source and drain set into n-type silicon. Should its gate be made positive or negative to turn it on, and what kind of carrier then fills the channel?
The Integrated Circuit
Early transistors were wired together one at a time, like the resistors and capacitors of a circuit board. In 1958 and 1959, Jack Kilby at Texas Instruments and Robert Noyce at Fairchild Semiconductor independently showed that transistors, resistors, capacitors, and the connections between them could all be built at once into a single chip of silicon: the integrated circuit. The patterns are printed by photolithography: a light-sensitive coating on the wafer is exposed to ultraviolet light through a mask, the exposed regions are etched away, dopants are implanted, and the cycle is repeated dozens of times to build up the layers of a circuit.
Because every transistor on a wafer is made in the same steps, shrinking the patterns puts more transistors on a chip at almost no extra cost. In 1965 Gordon Moore observed that the number of transistors on a chip was doubling at a steady rate; the doubling time settled at about two years and held for half a century, a trend known as Moore’s law.
Example 42.8: Fifty years of Moore’s law
The first commercial microprocessor, the Intel 4004 of 1971, held about \(2300\) transistors. If the number doubled every \(2.0\) years, how many transistors would a chip hold in 2023?
Solution
From 1971 to 2023 is \(52\) years, or \(26\) doublings, so
\[N = 2300\times2^{26} = 2300\times(6.7\times10^{7}) = \boxed{1.5\times10^{11}}.\]
The largest processors of 2023 did indeed contain somewhat more than \(10^{11}\) transistors. Fifty years of exponential growth multiplied the count by nearly a hundred million.
The growth cannot go on forever. The smallest features of today’s chips are only a few tens of atoms across, and the gate oxide is only a few atoms thick. At that scale the quantum effects that made the transistor possible begin to work against it: electrons tunnel straight through the thin oxide (Chapter 40), leaking current and wasting power. Much of modern chip design is a struggle against tunneling, waged with new materials and three-dimensional transistor shapes. The phones, computers, and networks built from these chips are the most tangible products of the quantum mechanics of the last four chapters.
Answers to Checkpoints
Checkpoint 42.5. Arsenic has one more valence electron than germanium needs for its four bonds, so it donates an extra electron: arsenic-doped germanium is n-type. Gallium has one fewer, leaving a hole: gallium-doped germanium is p-type.
Checkpoint 42.6. Negative. Everything is reversed: a negative gate repels electrons from the n-type surface and attracts holes, and once enough holes gather under the oxide, they form a p-type channel joining the p-type source and drain. The current in the channel is carried by holes. (Modern chips pair n-channel and p-channel transistors in “complementary” (CMOS) circuits, so that in either logic state one of each pair is off and almost no current flows.)
This page titled 42.4: Semiconductors was last modified on Thu, 12 Nov 2026 04:27:08 GMT and is shared under a CC BY 4.0 license and was authored, remixed, and/or curated by Michael Ragsdale.
The book could not be loaded just now.