Room-temperature photoluminescence (PL) measurements are performed on GaInN/GaN multiple-quantum-well heterostructures grown on GaN-on-sapphire templates with different threading-dislocation densities. The selective optical excitation of quantum wells and the dependence of integrated PL intensity on excitation power allow us to determine the internal quantum efficiency (IQE) as a function of carrier concentration. The measured IQE of the sample with the lowest dislocation density is as high as 64%. The measured nonradiative coefficient varies from to as the dislocation density increases from to , respectively.
We model the carrier recombination mechanisms in GaInN/GaN light-emitting diodes as , where represents carrier leakage out of the active region. The term is expanded into a power series and shown to have higher-than-third-order contributions to the recombination. The total third-order nonradiative coefficient (which may include an leakage contribution and an Auger contribution) is found to be . Comparison of the theoretical model with experimental data shows that a good fit requires the inclusion of the term.
The reverse leakage current of a GaInN light-emitting diode (LED) is analyzed by temperature dependent current–voltage measurements. At low temperature, the leakage current is attributed to variable-range-hopping conduction. At high temperature, the leakage current is explained by a thermally assisted multi-step tunneling model. The thermal activation energies (95–162 meV), extracted from the Arrhenius plot in the high-temperature range, indicate a thermally activated tunneling process. Additional room temperature capacitance–voltage measurements are performed to obtain information on the depletion width and doping concentration of the LED.
The internal quantum efficiency (IQE)-versus-carrier-concentration curves of GaN-based light-emitting diodes have been frequently described by the model: . We show that this model predicts IQE-versus- curves that have even symmetry. Phase-space filling makes the and coefficients concentration-dependent. We also show that IQE-versus- curves that take into account phase-space filling possess even symmetry. In contrast, experimental IQE-versus- curves exhibit asymmetry. The asymmetry requires a fourth-power or higher-power contribution to the recombination rate and provides insight into the mathematical form of the droop-causing mechanisms.