Search for: All records

High-throughput combinatorial synthesis of Al_1−xRE_xN (RE = Pr, Tb) thin films with 0 <x< 0.4 was performed to assess composition-phase-property relationships in an emerging materials family. 

III-nitrides and related alloys are widely used for optoelectronics and as acoustic resonators. Ferroelectric wurtzite nitrides are of particular interest because of their potential for direct integration with Si and wide bandgap semiconductors and unique polarization switching characteristics; such interest has taken off since the first report of ferroelectric Al1−xScxN alloys. However, the coercive fields needed to switch polarization are on the order of MV/cm, which are 1–2 orders of magnitude larger than oxide perovskite ferroelectrics. Atomic-scale point defects are known to impact the dielectric properties, including breakdown fields and leakage currents, as well as ferroelectric switching. However, very little is known about the native defects and impurities in Al1−xScxN and their effect on the dielectric and ferroelectric properties. In this study, we use first-principles calculations to determine the formation energetics of native defects and unintentional oxygen incorporation and their effects on the polarization switching barriers in Al1−xScxN alloys. We find that nitrogen vacancies are the dominant native defects, and unintentional oxygen incorporation on the nitrogen site is present in high concentrations. They introduce multiple mid-gap states that can lead to premature dielectric breakdown and increased temperature-activated leakage currents in ferroelectrics. We also find that nitrogen vacancy and substitutional oxygen reduce the switching barrier in Al1−xScxN at low Sc compositions. The effect is minimal or even negative (increases barrier) at higher Sc compositions. Unintentional defects are generally considered to adversely affect ferroelectric properties, but our findings reveal that controlled introduction of point defects by tuning synthesis conditions can instead benefit polarization switching in ferroelectric Al1−xScxN at certain compositions. 

Despite recent progress in Graph Neural Networks (GNNs), explaining predictions made by GNNs remains a challenging and nascent problem. The leading method mainly considers the local explanations, i.e., important subgraph structure and node features, to interpret why a GNN model makes the prediction for a single instance, e.g. a node or a graph. As a result, the explanation generated is painstakingly customized at the instance level. The unique explanation interpreting each instance independently is not sufficient to provide a global understanding of the learned GNN model, leading to the lack of generalizability and hindering it from being used in the inductive setting. Besides, training the explanation model explaining for each instance is time-consuming for large-scale real-life datasets. In this study, we address these key challenges and propose PGExplainer, a parameterized explainer for GNNs. PGExplainer adopts a deep neural network to parameterize the generation process of explanations, which renders PGExplainer a natural approach to multi-instance explanations. Compared to the existing work, PGExplainer has better generalization ability and can be utilized in an inductive setting without training the model for new instances. Thus, PGExplainer is much more efficient than the leading method with significant speed-up. In addition, the explanation networks can also be utilized as a regularizer to improve the generalization power of existing GNNs when jointly trained with downstream tasks. Experiments on both synthetic and real-life datasets show highly competitive performance with up to 24.7% relative improvement in AUC on explaining graph classification over the leading baseline. 

Wurtzite-type ferroelectrics have drawn increasing attention due to the promise of better performance and integration than traditional oxide ferroelectrics with semiconductors such as Si, SiC, and III-V compounds. However, wurtzite-type ferroelectrics generally require enormous electric fields, approaching breakdown, to reverse their polarization. The underlying switching mechanism(s), especially for multinary compounds and alloys, remains elusive. Here, we examine the switching behaviors in Al_1−xSc_xN alloys and wurtzite-type multinary candidate compounds we recently computationally identified. We find that switching in these tetrahedrally coordinated materials proceeds via a variety of nonpolar intermediate structures and that switching barriers are dominated by the more-electronegative cations. For Al_1−xSc_xN alloys, we find that the switching pathway changes from a collective mechanism to a lower-barrier mechanism enabled by inversion of individual tetrahedra with increased Sc composition. Our findings provide insights for future engineering and realization of wurtzite-type materials and open a door to understanding domain motion.