Removal of Chromium and Arsenic from Water Using Polyol-Functionalized Porous Aromatic Frameworks.

Chromium and arsenic are two of the most problematic water pollutants due to their high toxicity and prevalence in various water streams. While adsorption and ion-exchange processes have been applied for the efficient removal of numerous toxic contaminants, including heavy metals, from water, these technologies display relatively low overall performances and stabilities for the remediation of chromium and arsenic oxyanions. This work presents the use of polyol-functionalized porous aromatic framework (PAF) adsorbent materials that use chelation, ion-exchange, redox activity, and hydrogen-bonding interactions for the highly selective capture of chromium and arsenic from water. The chromium and arsenic binding mechanisms within these materials are probed using an array of characterization techniques, including X-ray absorption and X-ray photoelectron spectroscopies. Adsorption studies reveal that the functionalized porous aromatic frameworks (PAFs) achieve selective, near-instantaneous (reaching equilibrium capacity within 10 s), and high-capacity (2.5 mmol/g) binding performances owing to their targeted chemistries, high porosities, and high functional group loadings. Cycling tests further demonstrate that the top-performing PAF material can be recycled using mild acid and base washes without any measurable performance loss over at least ten adsorption-desorption cycles. Finally, we establish chemical design principles enabling the selective removal of chromium, arsenic, and boron from water. To achieve this, we show that PAFs appended with analogous binding groups exhibit differences in adsorption behavior, revealing the importance of binding group length and chemical identity.

PANNA 2.0: Efficient neural network interatomic potentials and new architectures.

We present the latest release of PANNA 2.0 (Properties from Artificial Neural Network Architectures), a code for the generation of neural network interatomic potentials based on local atomic descriptors and multilayer perceptrons. Built on a new back end, this new release of PANNA features improved tools for customizing and monitoring network training, better graphics processing unit support including a fast descriptor calculator, new plugins for external codes, and a new architecture for the inclusion of long-range electrostatic interactions through a variational charge equilibration scheme. We present an overview of the main features of the new code, and several benchmarks comparing the accuracy of PANNA models to the state of the art, on commonly used benchmarks as well as richer datasets.

Entropic Effects on Diamine Dynamics and CO2 Capture in Diamine-Appended Mg2(dopbdc) Metal-Organic Frameworks.

Recent measurements [Xu, J.; J. Phys. Chem. Lett. 2019, 10 (22), 7044-7049] have reported temperature-dependent rates of detachment of diamine from Mg sites in diamine-appended Mg2(dobpdc) [dobpdc4- = 4,4'-dihydroxy(1,1'-biphenyl)-3,3'-dicarboxylic] metal-organic frameworks, a process hypothesized to be a precursor for cooperative CO2 adsorption, leading to step-shaped isotherms or isobars. Here, we compute the rate of diamine exchange in this system for different diamines using metadynamics simulations based on a density functional theory-derived neural network potential. Reanalyzing recent measurements accounting for entropic effects, we find a positive correlation between the previously reported CO2 adsorption step pressure and the free energy at room temperature and show that the experiments and simulations are in good qualitative and quantitative agreement. The rates obtained here provide new insight into the chemical dynamics of CO2 adsorption in a class of materials that are promising for carbon capture and a lower bound on the time scale of cooperative adsorption.

Equivariant Neural Networks Utilizing Molecular Clusters for Accurate Molecular Crystal Lattice Energy Predictions.

Equivariant neural networks have emerged as prominent models in advancing the construction of interatomic potentials due to their remarkable data efficiency and generalization capabilities for out-of-distribution data. Here, we expand the utility of these networks to the prediction of crystal structures consisting of organic molecules. Traditional methods for computing crystal structure properties, such as plane-wave quantum chemical methods based on density functional theory (DFT), are prohibitively resource-intensive, often necessitating compromises in accuracy and the choice of exchange-correlation functional. We present an approach that leverages the efficiency, and transferability of equivariant neural networks, specifically Allegro, to predict molecular crystal structure energies at a reduced computational cost. Our neural network is trained on molecular clusters using a highly accurate Gaussian-type orbital (GTO)-based method as the target level of theory, eliminating the need for costly periodic DFT calculations, while providing access to all families of exchange-corelation functionals and post-Hartree-Fock methods. The trained model exhibits remarkable accuracy in predicting lattice energies, aligning closely with those computed by plane-wave based DFT methods, thus representing significant cost reductions. Furthermore, the Allegro network was seamlessly integrated with the USPEX framework, accelerating the discovery of low-energy crystal structures during crystal structure prediction.