Ultrafast decay of photoluminescence from high-density excitons inAlxGa1−xNmixed crystals: Diffusive propagation of exciton-polaritons

Abstract

We report on the ultrafast decay dynamics of photoluminescence ͑PL͒ in highly excited Al x Ga 1−x N mixed crystals under exciton resonant excitation at low temperatures. When the excitation intensity is increased, the P-band emission appears due to exciton-exciton inelastic scattering processes. The PL intensity of the P band decays rapidly, with a much shorter decay time than the radiative recombination time of the excitons. We show that the ultrafast PL decay dynamics can be understood as due to the disorder-induced diffusive propagation of photonlike exciton-polaritons in Al x Ga 1−x N mixed crystals.

Theory of the lifetime of an exciton incoherently created below its resonance frequency by inelastic scattering

“…22,26,27 Although these lifetimes basically depend on materials of samples, they are generally much shorter than the emission lifetime τ emit of bottleneck excitons at quasi-equilibrium (in the order of nanoseconds). The time-resolved measurements revealed also that the P-emission lifetime is an increasing function of the emission frequency, 20,21,[23][24][25]27 and the lifetime at each emission frequency is almost independent of the pumping power. 24,25 Note that the emissionfrequency-dependence of the P-emission lifetime can be scaled phenomenologically by that of inverse of the group velocity of the photon-like polariton.…”

“…In this interpretation, when polaritons are stabilized by a large enough transition dipole, they are created in the time scale of the exciton-photon interchange time τ Rabi = 2π/g µ of the polariton state, and it is certainly negligible (τ Rabi ∼ 0.06 ps in ZnO) compared to the other time scales except the escape time τ escape of polariton (then there is a crossover around the material size comparable to the radiation wavelength 33 ). Then, if the P-emission lifetimes do not originate from the lifetime of excitons at the bottleneck, obeying the conventional interpretation, we need the interpretations of the polariton diffusion 20 or of the polariton escape from a sample with an incredibly large effective thickness. 21,23 Let us examine whether this conventional interpretation is really justified or not from a fundamental viewpoint.…”

“…high-power pumping exceeding a threshold, and we have also an optical gain at the P-emission frequency. [11][12][13][14][15][16] The relaxation and scattering processes toward the P emission has been investigated in time-resolved measurements performed by optical Kerr gating method [17][18][19][20][21][22][23][24][25] and by streak camera, 26,27 and then the following facts have been revealed. 1) The onset time reflects the time of energy relaxation of excitons toward the bottleneck region on the lower exciton-polariton branch.…”

“…The typical P-emission lifetimes are observed as a few ps [17][18][19][20][21][23][24][25] or a few tens of ps. 22,26,27 Although these lifetimes basically depend on materials of samples, they are generally much shorter than the emission lifetime τ emit of bottleneck excitons at quasi-equilibrium (in the order of nanoseconds).…”

“…21,[23][24][25] The lifetime of the spectrally-integrated P-emission signal was shortened through the lowering of the peak frequency with an increase in pumping power (effective temperature) for InGaN. 26 However, it was almost unchanged for CuI 17,19 and AlGaN, 20 because the peak frequencies were not strongly changed. Further, the lifetime of the spectrallyintegrated P-emission signal was also independent of the pumping frequency for CuI.…”