29 May 2025
Doping, Carrier Concentration, and Mobility
Charge neutrality, donor and acceptor doping, Fermi-level displacement, mobility, and conductivity.
Let $N_D^+$ and $N_A^-$ be ionized donor and acceptor densities. Local electrical neutrality requires total positive and negative charge densities to agree:
\[\boxed{p+N_D^+=n+N_A^-}.\]When donors and acceptors are fully ionized, $N_D^+=N_D$ and $N_A^-=N_A$. Together with $np=n_i^2$, this determines the equilibrium carriers.
n-type material
For $N_D>N_A$, neutrality becomes $n-p=N_D-N_A\equiv N$. Since $p=n_i^2/n$,
\[n-\frac{n_i^2}{n}=N,\]so
\[n^2-Nn-n_i^2=0, \qquad \boxed{n=\frac{N+\sqrt{N^2+4n_i^2}}2}, \qquad p=\frac{n_i^2}{n}.\]If $N\gg n_i$, then $n\simeq N_D-N_A$ and $p\simeq n_i^2/n$. Using $n=n_i\exp[(E_F-E_i)/(k_BT)]$,
\[\boxed{E_F-E_i=k_BT\ln(n/n_i)}.\]Thus donor doping raises the Fermi level.
p-type material
For $N_A>N_D$, put $P=N_A-N_D$. Then $p-n=P$ and
\[\boxed{p=\frac{P+\sqrt{P^2+4n_i^2}}2}, \qquad n=\frac{n_i^2}{p},\]with
\[\boxed{E_i-E_F=k_BT\ln(p/n_i)}.\]Acceptor doping therefore lowers the Fermi level. In heavily doped material $E_F$ approaches or enters a band, and these non-degenerate exponential formulas must be replaced by Fermi-Dirac integrals.
Temperature dependence of carrier concentration
For shallow dopants there are three useful temperature ranges. At low temperature, dopants are only partly ionized (freeze-out), so the majority-carrier density rises rapidly as dopant atoms ionize. In the intermediate extrinsic range, ionization is nearly complete and the majority density is approximately the net dopant density; for example, $n\simeq N_D-N_A$ in compensated n-type material. At sufficiently high temperature, thermal pair generation dominates the dopants and
\[n\simeq p\simeq n_i =\sqrt{N_cN_v}\exp\!\left[-\frac{E_g}{2k_BT}\right].\]Because $N_c,N_v\propto T^{3/2}$ for temperature-independent effective masses, $n_i\propto T^{3/2}e^{-E_g/(2k_BT)}$. The minority density in the extrinsic range follows $p=n_i^2/n$ or $n=n_i^2/p$, so it remains strongly temperature dependent even when the majority density is nearly fixed.
Mobility and its dependence
In the relaxation-time model, a carrier with effective mass $m^*$ obeys
\[m^*\frac{d\mathbf v}{dt}=q_c\mathbf E-\frac{m^*\mathbf v}{\tau}.\]In steady state,
\[\mathbf v_d=\frac{q_c\tau}{m^*}\mathbf E,\]so the positive mobility magnitude is
\[\boxed{\mu=\frac{\lvert q_c\rvert\tau}{m^*}}.\]Lattice vibrations shorten $\tau$ as temperature rises, while ionized impurities shorten it as doping rises. If the two scattering mechanisms are independent, their rates add:
\[\boxed{\frac1\mu=\frac1{\mu_{\rm lattice}}+\frac1{\mu_{\rm impurity}}}.\]Carrier concentration and mobility therefore affect conductivity separately:
\[\sigma=q(n\mu_n+p\mu_p).\]Doping normally increases $n$ or $p$ by many orders of magnitude even though impurity scattering reduces mobility; the net conductivity consequently rises.
The mobility laws are material- and temperature-range dependent. In simple three-dimensional models, lattice-limited mobility often varies approximately as $T^{-3/2}$, whereas ionized-impurity-limited mobility varies approximately as $T^{3/2}/N_{\rm ion}$. These powers are approximations, not universal constants.
Discussion