Compositional engineering of perovskites enables the precise control of key material properties such as the bandgap [1]. This possibility makes perovskites a promising material for multi-junction/tandem solar cells, where the combination of two different bandgaps allows to easily break the Shockley-Queisser limit and thus improve efficiency [2]. Another critical factor is the material environmental stability under operating conditions. Two-dimensional perovskites have shown promising stability when subjected to light, humidity and heat [3]. The remaining challenge is to find structures with the target bandgap which are both stable in the environment as well as non-toxic (i.e. lead-free). To this end, computer simulations allow rapid screening of a large array of compositions for a given structure and subsequent data modeling. However, typical high-throughput calculations fall short in capturing the effect of different geometries, thus severely constraining the applicability of the result in predicting “new” structures. This becomes especially problematic for larger systems, such as the two-dimensional perovskites, where different organic spacer-cations might have varying relaxed geometries and sampling the whole configurational space becomes infeasible. Moreover, many machine learning models make difficult to determine a simple connection between particular structural features of the material and the consequent property like the bandgap. In our approach geometrical information is incorporated into the data-modeling by variations of the RDF based “property density distribution function” (PDDF) [4]. The resulting large feature vector is reduced via an Autoencoder (AE) to a lower dimensional latent space to prevent the Neural Network from overfitting. Autoencoders can be used also to address the second issue. Beta-VariationalAE are able to disentagle the generative factors of the structure. Varying these factors and observing the response of the predicted bandgap allows an understanding of the structure-property-relationship and facilitate to select the structural properties which produce a target bandgap.      «

Compositional engineering of perovskites enables the precise control of key material properties such as the bandgap [1]. This possibility makes perovskites a promising material for multi-junction/tandem solar cells, where the combination of two different bandgaps allows to easily break the Shockley-Queisser limit and thus improve efficiency [2]. Another critical factor is the material environmental stability under operating conditions. Two-dimensional perovskites have shown promising stability w...     »