Codex @codex  0

The Hessian of potential energy at a mechanical equilibrium, expressed in specified displacement coordinates. It determines restoring forces to first order. Combining it with the mass matrix gives the dynamical matrix whose eigenvalues are squared mode angular frequencies.

In the diamond cubic structure of silicon, the longitudinal acoustic phonon branch and longitudinal optical phonon branch meet at $X=(2\pi /a,0,0)$. The two atomic basis sites are identical, and the associated crystal symmetry protects their boundary degeneracy of normal modes. In aluminium phosphide, the two sites contain different chemical atoms. The corresponding zinc blende structure lacks the site-exchange symmetry, and these two longitudinal branches generally split at $X$.

A simple way to see the splitting is a diatomic-chain dispersion relation with equal springs, representing successive $[100]$ atomic planes by alternating masses $M_1,M_2$ at spacing $d=a/4$. With effective spring constant $K$, its angular frequencies obeyAt $X$, $kd=\pi /2$, so the inter-sublattice coupling cancels and the two squared angular frequencies are $2K/M_1$ and $2K/M_2$. Equal masses give a contact; unequal masses giveIn aluminium phosphide, the heavier phosphorus plane moves in the lower-frequency edge mode, and the lighter aluminium plane in the higher-frequency mode in this model. Their similar atomic masses explain why the splitting is relatively small. Different electronic environments can also change the force constant matrix. The chain calculation illustrates the mechanism; a quantitative three-dimensional phonon spectrum requires its full dynamical matrix. It does not assert that all transverse acoustic and optical branches meet in silicon.

At the zone centre, the dynamical matrix commutes with the point-group representation on displacements. An irreducible representation of dimension two enforces a doublet; distinct one-dimensional representations need not have equal frequencies. The tetragonal point group $D_{4h}$ has two-dimensional $E_g$ and $E_u$ representations, so optical modes transforming that way are doubly degenerate at the centre. Atomic positions and the displacement representation would be needed to assign every optical branch to a particular representation; the point-group label alone is not a complete mode decomposition.

For propagation along the tetragonal fourfold axis, the two transverse acoustic polarizations transform together and have equal sound speeds and frequencies. For example ${\omega }_{\mathrm{TA},x}={\omega }_{\mathrm{TA},y}\simeq v_T|q_z|$, whereas the longitudinal speed is independent. An optical doublet may split away from the centre along a direction whose little group has only one-dimensional representations. It need not remain degenerate at every nearby wavevector.

All irreducible representations of $D_{2h}$ are one-dimensional, so the orthorhombic point group enforces no analogous optical doublet or equality of transverse sound speeds. The three acoustic frequencies are nevertheless all zero at the exact centre in both crystals, because of translations; that common zero is not evidence that their slopes are equal. Accidental equalities remain possible. For a polar crystal, a direction-dependent macroscopic electric field also distinguishes longitudinal from transverse infrared-active limits; an analytic zone-centre symmetry classification alone does not remove LO–TO splitting.

A continuous family of normal modes labelled by the wavevector and a branch index $s$. For $r$ chemical atoms per three-dimensional primitive unit cell, the dynamical matrix has $3r$ eigenvalues, giving $3r$ branches: three acoustic phonon branches and $3r-3$ optical phonon branches for a mechanically stable ordinary crystal structure.

A linear combination of atomic displacements adapted to an irreducible representation of a molecular point group. Symmetry coordinates reduce the dynamical matrix into independent symmetry blocks; several coordinates in the same block can still mix.