Fundamental processes in solar cells involve charge-carrier generation, collection, and extraction. For these reasons, high quality absorber materials exhibit not only a strong absorbance but also high charge-carrier mobilities as well as long charge-carrier lifetimes. To obtain lifetimes of minority charge-carriers and to study their dynamics, time-resolved photoluminescence is a suitable method for materials with a sufficiently high luminescence quantum efficiency. Transient photoluminescence displays the time evolution of radiative recombination after pulsed excitation. In previous publications [1, 2, 3], various exponential functions or rate models have been used to describe photoluminescence decay curves of perovskite materials, but a commonly accepted theory that explains transients and offers a guideline for transient photoluminescence analysis has not been established. Because excess charge-carriers undergo complex processes like drift, diffusion, surface and bulk recombination, the resulting transient photoluminescence shapes can be very difficult to interpret.

Depending on the intensity of the laser pulse, photoluminescence transients of methylammonium lead iodide perovskite layers usually show an exponential decay at long times and an additional faster decay at short times. By comparison of experiment and simulation, we show what conclusions can and which conclusions cannot be drawn from the shape of the photoluminescence transients. The exponential decay for long times can be explained either as a Shockley-Read-Hall lifetime or it can be limited by surface recombination. Results from one-dimensional simulations indicate that low surface recombination velocities (S ≤ 10 cm/s) are necessary in first place to obtain effective lifetimes of several hundred nanoseconds. The features at short times can be either explained as a higher order recombination (radiative band-to-band and non-radiative Auger recombination) or they result from diffusion of charge-carriers due to an uneven generation profile. In order to discriminate between different effects, it is therefore advisable to vary the generation profile (by changing the illumination wavelength or direction), the distance to the surfaces (by changing the layer thickness) and the intensity of the laser pulse in order to modify the importance of linear and non-linear bulk recombination mechanisms.