Accelerating the Discovery of Metastable IrO2 for the Oxygen Evolution Reaction by the Self-Learning-Input Graph Neural Network

Rewc-GNN Algorithm for the Property Prediction of Large-Scale Crystals

The space group of a crystal describes the symmetry and periodic arrangement of its structure. As the fundamental element in the structure, it plays a vital role in determining the physical and chemical properties of crystals. The investigation of crystal space group information allows for the prediction of material properties, thereby providing guidance for material design and synthesis to enhance their performance or functionality. Currently prevalent first-principles-based computational methods exhibit good accuracy, but they rely heavily on computing resources, greatly limiting the efficiency of material screening. In this paper, our study is oriented toward the prediction the spatial group of crystals, and an algorithm named Rewc, based on graph neural networks (GNNs) is proposed. This algorithm encodes all atoms and the interactions between atoms in the crystal as features by combining Floyd algorithm and k-hop message passing and employs multilayer convolutional networks to extract connections between k layers. This allows for the automatic learning of more representative atomic vector representations through iterations of feature information for each atom and its neighbors. Experimental results demonstrate that the Rewc framework exhibits reliable accuracy and good generalization capabilities in predicting the crystal structure compared to previous GNN methods.

Theoretical Prediction and Experimental Verification of IrOx Supported on Titanium Nitride for Acidic Oxygen Evolution Reaction

Reducing iridium (Ir) catalyst loading for acidic oxygen evolution reaction (OER) is a critical strategy for large-scale hydrogen production via proton exchange membrane (PEM) water electrolysis. However, simultaneously achieving high activity, long-term stability, and reduced material cost remains challenging. To address this challenge, we develop a framework by combining density functional theory (DFT) prediction using model surfaces and proof-of-concept experimental verification using thin films and nanoparticles. DFT results predict that oxidized Ir monolayers over titanium nitride (IrOx/TiN) should display higher OER activity than IrOx while reducing Ir loading. This prediction is verified by depositing Ir monolayers over TiN thin films via physical vapor deposition. The promising thin film results are then extended to commercially viable powder IrOx/TiN catalysts, which demonstrate a lower overpotential and higher mass activity than commercial IrO2 and long-term stability of 250 h to maintain a current density of 10 mA cm-2. The superior OER performance of IrOx/TiN is further confirmed using a proton exchange membrane water electrolyzer (PEMWE), which shows a lower cell voltage than commercial IrO2 to achieve a current density of 1 A cm-2. Both DFT and in situ X-ray absorption spectroscopy reveal that the high OER performance of IrOx/TiN strongly depends on the IrOx-TiN interaction via direct Ir-Ti bonding. This study highlights the importance of close interaction between theoretical prediction based on mechanistic understanding and experimental verification based on thin film model catalysts to facilitate the development of more practical powder IrOx/TiN catalysts with high activity and stability for acidic OER.

Exploring an accurate machine learning model to quickly estimate stability of diverse energetic materials

High energy and low sensitivity have been the focus of developing new energetic materials (EMs). However, there has been a lack of a quick and accurate method for evaluating the stability of diverse EMs. Here, we develop a machine learning prediction model with high accuracy for bond dissociation energy (BDE) of EMs. A reliable and representative BDE dataset of EMs is constructed by collecting 778 experimental energetic compounds and quantum mechanics calculation. To sufficiently characterize the BDE of EMs, a hybrid feature representation is proposed by coupling the local target bond into the global structure characteristics. To alleviate the limitation of the low dataset, pairwise difference regression is utilized as a data augmentation with the advantage of reducing systematic errors and improving diversity. Benefiting from these improvements, the XGBoost model achieves the best prediction accuracy with R2 of 0.98 and MAE of 8.8 kJ mol-1, significantly outperforming competitive models.

Optimizing Edge Active Sites via Intrinsic In-Plane Iridium Deficiency in Layered Iridium Oxides for Oxygen Evolution Electrocatalysis

Improving catalytic activity of surface iridium sites without compromising catalytic stability is the core task of designing more efficient electrocatalysts for oxygen evolution reaction (OER) in acid. This work presents phase transition of a bulk layered iridate Na2IrO3 in acid solution at room temperature, and subsequent exfoliation to produce 2D iridium oxide nanosheets with around 4 nm thickness. The nanosheets consist of OH-terminated, honeycomb-type layers of edge-sharing IrO6 octahedral framework with intrinsic in-plane iridium deficiency. The nanosheet material is among the most active Ir-based catalysts reported for acidic OER and gives an iridium mass activity improvement up to a factor of 16.5 over rutile IrO2 nanoparticles. The material also exhibits good catalytic and structural stability and retains the catalytic activity for more than 1300 h. The combined experimental and theoretical results demonstrate that edge Ir sites of the layer are active centers for OER, with structural hydroxyl groups participating in the catalytic cycle of OER via a non-traditional adsorbate evolution mechanism. The existence of intrinsic in-plane iridium deficiency is the key to building a unique local environment of edge active sites that have optimal surface oxygen adsorption properties and thereby high catalytic activity.