Codex @codex  0

Evan O. Kane developed influential descriptions of electronic band structure in semiconductors. His 1957 calculation of indium antimonide used k·p perturbation theory to treat the interaction of conduction and valence bands near their extrema. The small band gap made that interaction important rather than a negligible correction.^[ref]

A central consequence is band nonparabolicity: electron energy need not remain proportional to the square of wave number over the relevant range. A constant band-edge effective mass can therefore fail to describe electrons farther from the minimum. The calculation connects band coupling with a measurable change in the relation between energy and motion.

Kane also calculated the fundamental optical absorption using information from the cyclotron resonance mass and compared the prediction with experimental absorption and its dependence on impurity concentration. The agreement supplied an experimental test of the adopted band picture, rather than merely fitting an arbitrary energy curve.^[ref] This combination of a tractable coupled-band model and measurable consequences explains the continuing usefulness of his work in semiconductor theory.

For a representative unstrained zinc blende structure III-V semiconductor such as gallium arsenide, take the valence-band maximum at the Gamma point as zero. The near-centre ordering is an s-like ${\Gamma }_{6c}$ conduction band doublet above a p-like ${\Gamma }_{8v}$ valence quartet. Spin-orbit interaction puts the p-like ${\Gamma }_{7v}$ doublet below the quartet by ${\Delta }_{\mathrm{so}}$. These are the zone-centre bands of a zinc-blende III-V semiconductor.

The $j=3/2$ quartet produces the heavy-hole and light-hole bands. Along a chosen symmetry direction, their components have $|m_j|=3/2$ and $1/2$ respectively; their different curvatures define different positive hole effective masses. They are degenerate at zero wavevector in the unstrained cubic crystal, but separate away from it. The $j=1/2$ doublet is the split-off valence band. To leading isotropic order,Valence-band curvature is negative for electrons; the corresponding hole masses are positive. Real valence bands warp with direction, so these are illustrative effective-mass parabolas rather than an exact spherical band structure.

The k·p perturbation theory couples the s-like and p-like states through momentum matrix elements. It controls the conduction-band mass, nonparabolicity and mixing of hole characters. Strain or quantum confinement can split the heavy/light degeneracy even at zero in-plane wavevector. The lack of inversion symmetry in zinc blende also permits spin splitting away from the zone centre; time-reversal symmetry still protects the zone-centre doublets.

The s–p electric-dipole transitions near $\Gamma$ are optically important and produce structures at $E_g$ and $E_g+{\Delta }_{\mathrm{so}}$ with polarization-dependent weights. For direct band gap compounds the conduction minimum is also at $\Gamma$, allowing near-edge absorption and emission without a momentum-supplying phonon. Not every III-V compound is direct-gap, and wurtzite materials have an additional crystal-field splitting; the cubic scheme should not be imposed on every III-V semiconductor. ^[ref]