Giant In-Plane Optical Anisotropy of Semiconductor Heterostructures with No-Common-Atom

The envelope function theory (EFT) originally proposed by G. Bastard [1] for the calculation of quantum well electronic properties has become very popular because, in spite of its remarkable simplicity, no quantitative or qualitative breakdown of its predictions has been reported until very recently. The EFT has been widely used not only to evaluate bare ≪ first order ≫ electronic properties, which was its original motivation, but also to predict many refined effects, such as QW polaritons [2], second order optical nonlinearities [3] or spin-relaxation phenomena [4,5]. The classical EFT is based on a somewhat simplified version of the 8x8 k.p hamiltonian describing the electronic properties of bulk III-V or II-VI semiconductors having the zinc-blend structure [6], to which scalar potentials describing the shifts of the band extrema at interfaces, and an eventual external potential are added [1]. Here, we consider exclusively the situation of heterostructures grown along the (001) axis. A characteristic feature of the EFT is that the projection Jz of the angular momentum on the quantification axis z is a good quantum number at the zone center, i.e. when the in-plane wavevector kt= 0. In other words, the heavy hole states | 3/2 ± 3/2〉 do not couple to the Jz = ±1/2 light particle states at kt= 0. This remains true even if the quantum well potential is asymmetric, as for instance when an axial electric field is present. Biaxial strain due to a possible lattice mismatch does not change this result [7], which contrasts with the classical group theoretical result stating that (neglecting effects associated with the integer or half-integer character of the layer thicknesses [8]) the point groups of symmetric and asymmetric quantum wells are respectively D2d and C2v [9–11].