Strain within grains and at grain boundaries (GBs) in polycrystalline thin-film absorber layers limits the overall performance because of higher defect concentrations and band fluctuations. However, the nanoscale strain distribution in operational devices is not easily accessible using standard methods. X-ray nanodiffraction offers the unique possibility to evaluate the strain or lattice spacing at nanoscale resolution. Furthermore, the combination of nanodiffraction with additional techniques in the framework of multimodal scanning X-ray microscopy enables the direct correlation of the strain with material and device parameters such as the elemental distribution or local performance. This approach is applied for the investigation of the strain distribution in CdTe grains in fully operational photovoltaic solar cells. It is found that the lattice spacing in the (111) direction remains fairly constant in the grain cores but systematically decreases at the GBs. The lower strain at GBs is accompanied by an increase of the total tilt. These observations are both compatible with the inhomogeneous incorporation of smaller atoms into the lattice, and local stress induced by neighboring grains.
Bragg coherent diffraction imaging (BCDI) is a powerful technique to explore the local strain state and morphology of microscale crystals. The method can potentially reach nanometer-scale spatial resolution thanks to the advances in synchrotron design that dramatically increase coherent flux. However, there are experimental bottlenecks that may limit the image reconstruction quality from future high signal-to-noise ratio measurements. In this work we show that angular uncertainty of the sample orientation with respect to a fixed incoming beam is one example of such a factor, and we present a method to mitigate the resulting artifacts. On the basis of an alternative formulation of the forward problem, we design a phase retrieval algorithm which enables the simultaneous reconstruction of the object and determination of the exact angular position corresponding to each diffraction pattern in the data set. We have tested the algorithm performance on simulated data for different degrees of angular uncertainty and signal-to-noise ratio.