Energy Bands in Solids

Band theory is the working language of the electronics industry, and every choice a chip designer makes traces back to a band gap. Silicon's $1.1\ \text{eV}$ is the workhorse of logic and memory because it is small enough to conduct on demand yet large enough to switch cleanly off. Wider-gap materials — gallium nitride ($3.4\ \text{eV}$) and silicon carbide ($3.3\ \text{eV}$) — now run high-voltage chargers and electric-car inverters, precisely because their bigger gaps withstand fields that would break silicon down. LEDs glow at a colour set by their gap ($hc/E_g$ again); solar cells choose a gap near $1.4\ \text{eV}$ to match sunlight. Everything in the rest of this chapter, from the intrinsic semiconductor to the transistor and the logic gate, is built on the single diagram you now own: two bands and the forbidden gap between them.

Start with a single silicon atom. Its electrons occupy sharp energy levels — think of the Bohr picture: discrete rungs, each holding a fixed number of electrons, with wide forbidden gaps between them where no electron is allowed. This is the atom you already know. Now bring a second silicon atom close. The outer electrons begin to feel each other, and a rule of quantum mechanics — the Pauli exclusion principle — forbids two electrons in the joined system from occupying exactly the same state. Nature's fix is to split each shared level into two slightly different ones. Bring up a third atom and it splits into three; a fourth, into four. Each atom you add forces every level to split once more. In a real crystal the number of atoms is staggering: a crystal you can see holds of order $10^{23}$ atoms, so each atomic level splits into about $10^{23}$ sub-levels. Spread across an energy range of only a few electronvolts, these are separated by unimaginably tiny gaps — as we will compute, around $10^{-23}\ \text{eV}$ apart, far smaller than any energy the electron ever exchanges with its surroundings. To every physical purpose the $10^{23}$ rungs merge into a smooth, continuous ramp of allowed energies: an energy band.

Two of these bands do all the work. The valence band is the highest band that the crystal's bonding electrons fill at absolute zero — in silicon it holds the electrons shared in the covalent bonds. Just above it, separated by the forbidden gap, sits the conduction band: the next allowed band, normally empty. The vertical distance between the top of the valence band and the bottom of the conduction band is the band gap (or forbidden energy gap) $E_g$, in electronvolts. Almost nothing about the crystal changes except this one number as we walk from copper to silicon to diamond — and that one number is the whole story.

Here is the subtlety a lot of students miss: having electrons is not enough to conduct; the electrons must be able to change their state. An applied electric field nudges electrons to move preferentially one way, which means giving them slightly more energy and momentum — pushing them into nearby empty states. In a completely filled band there are none: every allowed level is already occupied, so for every electron the field speeds up, another is already moving the opposite way, and the two cancel exactly. A full band carries no net current, however many electrons it holds. So conduction demands one of two situations. Either the highest occupied band is only partly filled, leaving empty states right alongside the occupied ones for electrons to slide into — a metal. Or the valence band is full but a few electrons have been lifted across the gap into the conduction band, where they sit in an almost-empty band with endless room to move. The forbidden gap is precisely what stands between the full valence band and the empty conduction band, so its size controls how easily electrons reach a band where they can conduct.

Read the three cases straight off the band picture. In a conductor — copper, silver, aluminium — the highest occupied band is either only partly filled, or the valence and conduction bands physically overlap so filled and empty states merge with no gap ($E_g \approx 0$). Either way a vast reservoir of empty states is immediately available, so even the faintest field sets a current flowing. This is why metals conduct at any temperature. In an insulator — diamond, glass, quartz — the valence band is full, the conduction band empty, and the gap between them large: $\gtrsim 3\ \text{eV}$, and for diamond about $5.5\ \text{eV}$. Crossing so wide a gap needs an energy kick that ordinary heat or a modest field cannot supply, so essentially no electron ever reaches the conduction band. No mobile carriers, no current. A semiconductor — silicon ($E_g \approx 1.1\ \text{eV}$), germanium ($E_g \approx 0.7\ \text{eV}$) — has the same full-valence, empty-conduction arrangement as an insulator, but with a small gap. At absolute zero it too is a perfect insulator; but the gap is narrow enough that at room temperature a thermal trickle of electrons is kicked across into the conduction band, leaving mobile vacancies behind, and the crystal conducts feebly. A semiconductor is, quite literally, an insulator with a gap small enough for heat to breach — which is what makes it tunable, and useful.

Why does room temperature lift many electrons across germanium's gap, a handful across silicon's, and none across diamond's? Because the average thermal energy available to an electron is fixed by temperature, and it is small. The natural yardstick is $kT$:

where: $k = 8.617\times10^{-5}\ \text{eV/K}$ is the Boltzmann constant in convenient units, $T = 300\ \text{K}$ is room temperature, and $kT$ is the characteristic thermal energy per particle. The physical reading: at room temperature nature hands each electron thermal jostles of only about $0.026\ \text{eV}$ on average — roughly a fortieth of an electronvolt. Compare that with the gaps: silicon's $1.1\ \text{eV}$ is about forty times $kT$, and diamond's $5.5\ \text{eV}$ is over two hundred times $kT$. Ordinary heat is nowhere near enough, on average, to launch an electron across. But "on average" is not "never." Thermal energy is shared randomly, and a small fraction of electrons momentarily carry far more than the average. The probability that one is lifted across a gap $E_g$ follows an exponential (Boltzmann) suppression — a result we quote, since the exact carrier count needs Fermi–Dirac statistics from a later, university-level treatment:

where: $f$ is roughly the fraction of valence electrons thermally promoted to the conduction band, $E_g$ is the gap, and $kT$ the thermal energy of equation (1); the factor of $2$ appears because the Fermi level sits near mid-gap, so a carrier need only be lifted about half the gap on average. The reading is dramatic. Because $E_g$ sits in an exponent, small changes in the gap produce enormous changes in conduction. Put the three materials through it and the sorting becomes overwhelming — germanium leaks about a part in a million, silicon about a part in a billion, diamond a part in $10^{47}$. That $E_g$ appears exponentially, not linearly, is the deepest reason a $1\ \text{eV}$ gap and a $5\ \text{eV}$ gap describe utterly different worlds.

There is a second, independent way to read a band gap — with light instead of heat. A photon carries energy $E = hf = hc/\lambda$. If that energy at least matches the gap, the photon can be absorbed by lifting a valence electron into the conduction band; if it falls short, the photon sails through untouched. So the longest wavelength a material can absorb across its gap is

where: $\lambda_{\text{max}}$ is the threshold wavelength, $E_g$ the band gap, and $hc = 1240\ \text{eV}\,\cdot\,\text{nm}$ the handy product that turns a gap in electronvolts straight into a wavelength in nanometres. For silicon, $\lambda_{\text{max}} = 1240/1.1 = 1127\ \text{nm}$ — deep in the infrared, beyond the red edge of vision. Silicon absorbs all visible light (every visible photon, $1.8$ to $3.1\ \text{eV}$, easily clears the $1.1\ \text{eV}$ gap) and looks an opaque metallic grey, yet is transparent to far-infrared light. Diamond's threshold is $1240/5.5 = 225\ \text{nm}$, in the ultraviolet: no visible photon carries the $5.5\ \text{eV}$ to jump its gap, so visible light passes straight through and pure diamond is water-clear. The colour of a solid is, in part, a direct picture of its band gap.

Board level — classifying three solids by their band gap (CBSE and ICSE pattern)

Three crystals have measured band gaps of $E_g = 0\ \text{eV}$ (bands overlapping), $E_g = 0.7\ \text{eV}$, and $E_g = 5.5\ \text{eV}$. Classify each as a conductor, semiconductor, or insulator, and state which will conduct better at room temperature — and why.

Step 1. The zero-gap crystal has its valence and conduction bands overlapping, so filled and empty states merge with no forbidden region. Empty states are always available: this is a conductor.

Step 2. The $0.7\ \text{eV}$ crystal has a full valence band and empty conduction band separated by a small gap. Compare with $kT \approx 0.026\ \text{eV}$: the gap is about $27$ times $kT$, small enough that a thermal trickle crosses at room temperature. This is a semiconductor (in fact germanium).

Step 3. The $5.5\ \text{eV}$ crystal has the same arrangement but a gap over $200$ times $kT$ — far too wide for heat to breach. Essentially no carriers reach the conduction band: an insulator (diamond).

Step 4. Among the two non-metals the smaller gap wins: the $0.7\ \text{eV}$ crystal promotes vastly more carriers than the $5.5\ \text{eV}$ one, because the promoted fraction falls off exponentially with the gap. Order of conductivity: conductor $\gg$ semiconductor $\gg$ insulator.

Board level — the longest wavelength silicon can absorb (CBSE and ICSE pattern)

Silicon has a band gap $E_g = 1.1\ \text{eV}$. Using $hc = 1240\ \text{eV}\,\cdot\,\text{nm}$, find the longest-wavelength photon that can lift an electron across the gap, and state whether silicon should look transparent or opaque to visible light.

Step 1. A photon is absorbed across the gap only if its energy $E = hc/\lambda$ is at least $E_g$. The threshold (longest wavelength, lowest energy that still works) is where $E = E_g$:

Step 2. $1127\ \text{nm}$ lies in the near-infrared, past the red end of the visible band (which stops around $700\ \text{nm}$). Photons longer than $1127\ \text{nm}$ cannot cross the gap and pass through; photons shorter — including all visible light — carry more than $1.1\ \text{eV}$ and are absorbed.

Step 3. Every visible photon ($400$–$700\ \text{nm}$, i.e. $1.8$–$3.1\ \text{eV}$) exceeds the $1.1\ \text{eV}$ gap, so all visible light is absorbed. Silicon looks opaque and metallic-grey to the eye, yet is transparent to far-infrared light — the reason infrared cameras and fibre-optic detectors are built on it.

Competitive level — how many electrons cross the gap, and how fast it changes with temperature (JEE and NEET pattern)

For silicon ($E_g = 1.1\ \text{eV}$), estimate the fraction of valence electrons thermally promoted to the conduction band at $300\ \text{K}$ using $f \sim e^{-E_g/2kT}$ with $k = 8.617\times10^{-5}\ \text{eV/K}$. Then find by what factor that fraction increases when the crystal is heated to $400\ \text{K}$.

Step 1. Thermal energy at $300\ \text{K}$: $kT = 8.617\times10^{-5}\times300 = 0.0259\ \text{eV}$, so $2kT = 0.0517\ \text{eV}$.

Step 2. Exponent at $300\ \text{K}$: $E_g/2kT = 1.1/0.0517 = 21.3$. Fraction $f_{300} \sim e^{-21.3} = \boxed{6\times10^{-10}}$ — roughly one valence electron in $1.7$ billion.

Step 3. At $400\ \text{K}$: $kT = 8.617\times10^{-5}\times400 = 0.0345\ \text{eV}$, $2kT = 0.0689\ \text{eV}$, exponent $= 1.1/0.0689 = 15.96$. So $f_{400} \sim e^{-15.96}$.

Step 4. The factor of increase is the ratio of the two exponentials: $f_{400}/f_{300} = e^{-15.96}/e^{-21.3} = e^{\,21.28-15.96} = e^{5.32} \approx \boxed{200}$. A mere $100\ \text{K}$ rise multiplies the carrier count about two-hundredfold — which is why a semiconductor's conductivity climbs with temperature (opposite to a metal), and why devices must be kept cool to behave predictably. Timing note: about two minutes. The whole item is two evaluations of $E_g/2kT$ and a subtraction in the exponent — never compute the two tiny exponentials separately and divide, that invites calculator overflow; instead take the ratio as $e^{\Delta(\text{exponent})}$. Memorise $2kT = 0.052\ \text{eV}$ at room temperature and the first exponent is a one-step division.

Competitive level — reading a gap off an absorption threshold (JEE Advanced pattern)

A semiconductor stops absorbing light for wavelengths longer than $1771\ \text{nm}$; below that it is opaque. Find its band gap in electronvolts and identify the likely material. Take $hc = 1240\ \text{eV}\,\cdot\,\text{nm}$.

Step 1. The absorption edge marks the photon energy that just equals the gap: $E_g = hc/\lambda_{\text{max}}$. Rearranging equation (4) and inserting the threshold:

Step 2. A gap of $0.70\ \text{eV}$ is the textbook value for germanium. The logic runs both ways: measure the longest wavelength a crystal absorbs and you have measured its band gap directly, no thermometer required.

Step 3. Sanity check: silicon's larger $1.1\ \text{eV}$ gap gives a shorter threshold, $1127\ \text{nm}$ — a wider-gap material absorbs only shorter, higher-energy wavelengths, exactly as $E_g = hc/\lambda$ demands. The ordering is consistent, so $0.70\ \text{eV}$ for germanium is secure. Timing note: under a minute. One division, $1240/\lambda$, converts any absorption edge in nanometres straight to a gap in electronvolts. Keep $hc = 1240\ \text{eV}\,\cdot\,\text{nm}$ memorised and this class of question is instant.

Thinking "more electrons" means "better conductor." Insulators are full of electrons; what they lack is empty states nearby. A completely filled band carries no current. Always frame conduction as the availability of accessible empty states, not the mere presence of electrons — this single idea answers most band-theory conceptual questions.

Confusing the valence band with the conduction band. The valence band is the highest filled band (the bonding electrons); the conduction band is the next one up, normally empty, where electrons are free to move. Students routinely swap the two or say current flows "in the valence band" of a semiconductor — the mobile electrons are in the conduction band (the holes they leave behind are the next article's story).

Forgetting that a semiconductor is an insulator at absolute zero. At $0\ \text{K}$ silicon has a full valence band, an empty conduction band, and no thermal energy to bridge them — it does not conduct at all. Semiconduction is a thermal effect. The corollary trips people up too: a semiconductor's conductivity rises with temperature (more carriers promoted), whereas a metal's falls (more lattice vibrations scatter the already-free electrons). Getting this temperature dependence backwards is a classic lost mark.

Misusing the gap values. Quote them cleanly: conductor $E_g \approx 0$; semiconductor small (Si $1.1\ \text{eV}$, Ge $0.7\ \text{eV}$); insulator large ($\gtrsim 3\ \text{eV}$, diamond $\approx 5.5\ \text{eV}$). Do not write silicon's gap as $0.7$ (that is germanium) or claim insulators have "no bands" — they have bands and a wide gap.

Board scope. Energy bands, the valence/conduction/gap picture, and the conductor–semiconductor–insulator sorting are core on CBSE (Ch. 14), ICSE/ISC, JEE Main, JEE Advanced, and NEET, and are useful background for the A-Level 9702 electronics and diode topics, though the band picture is not itself a 9702 exam topic. They are not examined on AP Physics 2 (no semiconductor electronics at all), on IGCSE 0625 (which starts at the diode, skipping bands), or in the current IB Physics core (electronics has largely been dropped). On those last three, learn the band picture for understanding but do not expect it on the paper.

Texts. NCERT Physics Class 12, Chapter 14 ("Semiconductor Electronics"), the energy-bands section · H. C. Verma, Concepts of Physics Vol. 2, "Semiconductors and Semiconductor Devices" · Kittel, Introduction to Solid State Physics, for the full band-structure treatment · Ashcroft & Mermin, Solid State Physics, for Bloch's theorem done properly.

Provenance. LANE_SPEC / anchor constants. Recompute: $kT = 8.617\times10^{-5}\times300 = 0.0259\ \text{eV} \approx 0.026\ \text{eV}$; $2kT = 0.0517\ \text{eV}$; band-formation spacing $\Delta E = 5/(6.0\times10^{23}) = 8\times10^{-24}\ \text{eV}$, ratio $kT/\Delta E = 3\times10^{21}$; Boltzmann exponents $E_g/2kT$ = $13.5$ (Ge), $21.3$ (Si), $106$ (diamond); fractions $1.3\times10^{-6}$, $6\times10^{-10}$, $6\times10^{-47}$; Ge/Si carrier ratio $\approx 2.3\times10^{3}$ (from exponent $21.28-13.54=7.74$) versus measured $n_i$ ratio $2.4\times10^{19}/1.5\times10^{16} = 1600$; Si $300\to400\ \text{K}$ boost $e^{5.32} = 204$; absorption thresholds $\lambda = hc/E_g$ = $1127\ \text{nm}$ (Si), $1771\ \text{nm}$ (Ge), $225\ \text{nm}$ (diamond); visible band $1.8$–$3.1\ \text{eV}$ from $1240/700$ and $1240/400$.

Prerequisites. The Bohr Model of the Atom — the source of the sharp, discrete atomic energy levels that broaden into bands here; without quantized levels there is nothing to split. Leads to. Intrinsic Semiconductors (the next article), where the thermal electron–hole pairs sketched here get their proper treatment ($n_e = n_h = n_i$); and, for university enrichment, the seams on the Fermi–Dirac distribution and Fermi energy, momentum and temperature, which supply the exact carrier statistics we only quoted (go-deeper only, never an exam dependency).