In a full band, every electron moving one way is matched by one moving the opposite way, so the velocities cancel. A field could change that only by moving electrons into empty states, and a full band has none: it carries no current. A partly filled band has empty states just above the filled ones, so it conducts. Call the highest band full at 0 K the valence band, the empty band above it the conduction band, and the gap between them Eg, in eV.
The 4 eV border is a convention. At 300 K, kBT≈0.026 eV, so a 1 eV gap is about 38 times larger and the Boltzmann factor makes a crossing rare. Each crossing puts an electron in the conduction band and leaves a hole in the valence band, a carrier of charge +e. Write n and p for the electron and hole concentrations, now in cm−3. In a pure (intrinsic) crystal they are made in pairs, so n=p.
INTRINSIC CONCENTRATION AGAINST TEMPERATURE
ni2=np∝T3e−Eg/kBT⇒ni=CT3/2e−Eg/(2kBT)
Valid in equilibrium with Eg≫kBT, Eg taken as fixed. A pair costs Eg, so G/r carries e−Eg/kBT; its slower temperature dependence is the factor T3, stated without proof. The square root halves the exponent. C (cm−3 K−3/2) is fixed by one measured ni; T in K, kB=8.62×10−5 eV/K.
A donor such as phosphorus has five outer electrons, and silicon's bonds use four. The fifth sits in a level about 0.05 eV below the conduction band, about twice kBT at 300 K, and that band offers far more empty states than there are donors, so nearly every donor ionises: a free electron and a fixed positive ion, n-type material. An acceptor such as boron has three, so it takes a valence electron, leaving a hole and a fixed negative ion: p-type. Write ND and NA for their concentrations, in cm−3.
CHARGE NEUTRALITY AND AN N-TYPE CRYSTAL
n+NA=p+ND⇒n≈ND,p=NDni2
Valid with every dopant ionised; the left side counts negative charge, the right positive. With NA=0 and ND≫ni, thermal pairs are negligible, so n≈ND and mass action gives p. For p-type, p≈NA and n=ni2/NA. All in cm−3.