Dark excitons get a parity map: mirror symmetry splits bright from dark and shuts polar phonons out of GaN monolayers
Synopsis
In planar monolayers, mirror parity is shown to hold exactly for every electron, phonon, and exciton state across the whole Brillouin zone, so exciton-phonon channels are not merely forbidden but selected by parity. In monolayer GaN as the test case, the lowest exciton is dark on two counts, finite momentum and odd parity, and only one odd phonon can scatter it into bright states and emit circularly polarized light, while polar optical phonons with matrix elements an order of magnitude larger contribute nothing to the bright-state linewidth because parity does not match. Real-time Boltzmann transport shows about 45% of the population reaches the K and Q valleys within 30 fs, and about 90% accumulates there after thermalization.
Interpretation
Every electronic state, phonon mode, and exciton band in a planar monolayer carries a mirror parity that is exact at every wavevector, reducing exciton-phonon coupling to one rule: even phonons connect states of equal parity, odd phonons connect states of opposite parity. Group theory previously sorted finite-momentum excitons by irreducible representation into allowed and forbidden channels, but a selection rule states only what may happen, not which channel carries the relaxation. The lowest exciton band is even inside a closed contour around Γ and odd throughout the rest of the zone including K and M; the contour marks where an even and an odd band cross, and mirror symmetry forbids their mixing, so the crossing is protected.
Parity determines which channels exist while coupling strength determines how much each allowed channel carries, and the two roles separate in relaxation. Linewidth and relaxation are mediated by different phonon branches: polar optical branches carry exciton-phonon matrix elements an order of magnitude larger than any other mode yet contribute nothing to the bright-state linewidth because they are even. At 300 K the lowest bright exciton at Γ has a linewidth of 151 meV, a lifetime of about 4 fs, with absorption contributing 43% and the lowest acoustic branch 44%; at 5 K absorption is thermally frozen out and the roughly 37 meV linewidth comes almost entirely from spontaneous phonon emission.
The real-time Boltzmann transport equation follows the full exciton population across the Brillouin zone rather than the out-scattering rate of a single state. The linewidth gives the out-scattering rate of a state, while real-time transport follows the population itself, and the two agree to within about 1% under identical meshes, couplings, and delta-function integration. After in-plane optical injection at 300 K, about 45% of the population reaches the K and Q valleys within 30 fs and about 90% accumulates there by thermalization; the state at K acts as a bottleneck because it cannot emit a phonon, and the equilibrium flux is nearly three orders of magnitude smaller than during the initial cascade.
Perspective
The analysis targets planar monolayers with a horizontal mirror plane, using monolayer GaN as the test case and monolayer hBN for contrast; buckling or Janus asymmetry removes the mirror plane and with it these constraints. GaN grown so far by graphene encapsulation on SiC is buckled, and only the planar phase keeps exact parity, which is the phase studied here.
The parity switch of the lowest exciton band across the closed contour around Γ relies on that crossing being protected by mirror symmetry; the linewidth at K primarily measures dephasing rather than population depletion, and population remains trapped in the valley much longer than the lifetime inferred from the linewidth.
