MultiXC-QM9: Large dataset of molecular and reaction energies from multi-level quantum chemical methods.

Well curated extensive datasets have helped spur intense molecular machine learning (ML) method development activities over the last few years, encouraging nonchemists to be part of the effort as well. QM9 dataset is one of the benchmark databases for small molecules with molecular energies based on B3LYP functional. G4MP2 based energies of these molecules were published later. To enable a wide variety of ML tasks like transfer learning, delta learning, multitask learning, etc. with QM9 molecules, in this article, we introduce a new dataset with QM9 molecule energies estimated with 76 different DFT functionals and three different basis sets (228 energy numbers for each molecule). We additionally enumerated all possible A ↔ B monomolecular interconversions within the QM9 dataset and provided the reaction energies based on these 76 functionals, and basis sets. Lastly, we also provide the bond changes for all the 162 million reactions with the dataset to enable structure- and bond-based reaction energy prediction tools based on ML.

Equivariant Graph-Representation-Based Actor-Critic Reinforcement Learning for Nanoparticle Design.

We have developed an actor-critic-type policy-based reinforcement learning (RL) method to find low-energy nanoparticle structures and compared its effectiveness to classical basin-hopping. We took a molecule building approach where nanoalloy particles can be regarded as metallic molecules, albeit with much higher flexibility in structure. We explore the strengths of our approach by tasking an agent with the construction of stable mono- and bimetallic clusters. Following physics, an appropriate reward function and an equivariant molecular graph representation framework is used to learn the policy. The agent succeeds in finding well-known stable configuration for small clusters in both single and multicluster experiments. However, for certain use cases the agent lacks generalization to avoid overfitting. We relate this to the pitfalls of actor-critic methods for molecular design and discuss what learning properties an agent will require to achieve universality.

Neural network potentials for accelerated metadynamics of oxygen reduction kinetics at Au-water interfaces.

The application of ab initio molecular dynamics (AIMD) for the explicit modeling of reactions at solid-liquid interfaces in electrochemical energy conversion systems like batteries and fuel cells can provide new understandings towards reaction mechanisms. However, its prohibitive computational cost severely restricts the time- and length-scales of AIMD. Equivariant graph neural network (GNN) based accurate surrogate potentials can accelerate the speed of performing molecular dynamics after learning on representative structures in a data efficient manner. In this study, we combined uncertainty-aware GNN potentials and enhanced sampling to investigate the reactive process of the oxygen reduction reaction (ORR) at an Au(100)-water interface. By using a well-established active learning framework based on CUR matrix decomposition, we can evenly sample equilibrium structures from MD simulations and non-equilibrium reaction intermediates that are rarely visited during the reaction. The trained GNNs have shown exceptional performance in terms of force prediction accuracy, the ability to reproduce structural properties, and low uncertainties when performing MD and metadynamics simulations. Furthermore, the collective variables employed in this work enabled the automatic search of reaction pathways and provide a detailed understanding towards the ORR reaction mechanism on Au(100). Our simulations identified the associative reaction mechanism without the presence of *O and a low reaction barrier of 0.3 eV, which is in agreement with experimental findings. The methodology employed in this study can pave the way for modeling complex chemical reactions at electrochemical interfaces with an explicit solvent under ambient conditions.

Red Shift in Optical Excitations on Layered Copper Perovskites under Pressure: Role of the Orthorhombic Instability.

The red shift under pressure in optical transitions of layered compounds with CuCl6 4- units is explored through first-principles calculations and the analysis of available experimental data. The results on Cu2+ -doped (C2 H5 NH3 )2 CdCl4 , that is taken as a guide, show the existence of a highly anisotropic response to pressure related to a structural instability, driven by a negative force constant, that leads to an orthorhombic geometry of CuCl6 4- units but with a hole displaying a dominant 3z2 -r2 character (z being the direction perpendicular to the layer plane). As a result of such an instability, a pressure of only 3 GPa reduces by 0.21 Šthe longest Cu2+ -Cl- distance, lying in the layer plane, while leaving unmodified the two other metal-ligand distances. Owing to this fact, it is shown that the lowest d-d transition would experience a red shift of 0.34 eV while the first allowed charge transfer transition is also found to be red shifted but only by 0.11 eV that reasonably concurs with the experimental value. The parallel study on Jahn-Teller systems CdCl2 :Cu2+ and NaCl:Cu2+ involving tetragonal elongated CuCl6 4- units shows that the reduction of the long axis by a pressure of 3 GPa is three times smaller than that for the layered (C2 H5 NH3 )2 CdCl4 :Cu2+ compound. Accordingly, the optical transitions of such systems, which involve a positive force constant, are much less sensitive to pressure than in layered compounds. The origin of the red shift under pressure undergone by the lowest d-d and charge transfer transitions of (C2 H5 NH3 )2 CdCl4 :Cu2+ is discussed in detail.

Transition1x - a dataset for building generalizable reactive machine learning potentials.

Machine Learning (ML) models have, in contrast to their usefulness in molecular dynamics studies, had limited success as surrogate potentials for reaction barrier search. This is primarily because available datasets for training ML models on small molecular systems almost exclusively contain configurations at or near equilibrium. In this work, we present the dataset Transition1x containing 9.6 million Density Functional Theory (DFT) calculations of forces and energies of molecular configurations on and around reaction pathways at the ωB97x/6-31 G(d) level of theory. The data was generated by running Nudged Elastic Band (NEB) with DFT on 10k organic reactions of various types while saving intermediate calculations. We train equivariant graph message-passing neural network models on Transition1x and cross-validate on the popular ANI1x and QM9 datasets. We show that ML models cannot learn features in transition state regions solely by training on hitherto popular benchmark datasets. Transition1x is a new challenging benchmark that will provide an important step towards developing next-generation ML force fields that also work far away from equilibrium configurations and reactive systems.

Cheap Turns Superior: A Linear Regression-Based Correction Method to Reaction Energy from the DFT.

Workflows to predict chemical reaction networks based on density functional theory (DFT) are prone to systematic errors in reaction energy due to the extensive use of cheap DFT exchange-correlation functionals to limit computational cost. Recently, machine learning-based models are increasingly applied to mitigate this problem. However, machine learning models require systems similar to trained data, and the models often perform poorly for out-of-distribution systems. Here, we present a simple bond-based correction method that improves the accuracy of DFT-derived reaction energies. It is based on linear regression, and the correction terms for each bond are derived from reactions among the QM9 data set. We demonstrate the effectiveness of this method with three DFT functionals in three different rungs of Jacob's ladder. The simple correction method is effective for all rungs but especially so for the cheapest PBE functional. Finally, we applied the correction method to a few reactions with molecules significantly different from those in the QM9 data set that was used to fit the linear regression model. Once corrected by this method, we found that the DFT reaction energies for such out-of-distribution reactions are within 0.05 eV of the G4MP2 method.

Artificial Intelligence Applied to Battery Research: Hype or Reality?

This is a critical review of artificial intelligence/machine learning (AI/ML) methods applied to battery research. It aims at providing a comprehensive, authoritative, and critical, yet easily understandable, review of general interest to the battery community. It addresses the concepts, approaches, tools, outcomes, and challenges of using AI/ML as an accelerator for the design and optimization of the next generation of batteries─a current hot topic. It intends to create both accessibility of these tools to the chemistry and electrochemical energy sciences communities and completeness in terms of the different battery R&D aspects covered.

Alteration of Electronic Band Structure via a Metal-Semiconductor Interfacial Effect Enables High Faradaic Efficiency for Electrochemical Nitrogen Fixation.

The interface engineering strategy has been an emerging field in terms of material improvisation that not only alters the electronic band structure of a material but also induces beneficial effects on electrochemical performances. Particularly, it is of immense importance for the environmentally benign electrochemical nitrogen reduction reaction (NRR), which is potentially impeded by the competing hydrogen evolution reaction (HER). The main problem lies in the attainment of the desired current density at a negotiable potential where the NRR would dominate over the HER, which in turn hampers the Faradaic efficiency for the NRR. To circumvent this issue, catalyst development becomes necessary, which would display a weak affinity for H-adsorption suppressing the HER at the catalyst surface. Herein, we have adopted the interfacial engineering strategy to synthesize our electrocatalyst NPG@SnS2, which not only suppressed the HER on the active site but yielded 49.3% F.E. for the NRR. Extensive experimental work and DFT calculations regarded that due to the charge redistribution, the Mott-Schottky effect, and the band bending of SnS2 across the contact layer at the interface of NPG, the d-band center for the surface Sn atoms in NPG@SnS2 lowered, which resulted in favored adsorption of N2 on the Sn active site. This phenomenon was driven even forward by the upshift of the Fermi level, and eventually, a decrease was seen in the work function of the heterostructure that increased the conductivity of the material as compared to pristine SnS2. This strategy thus provides a field to methodically suppress the HER in the realm of improving the Faradaic efficiency for the NRR.

Ab initio Molecular Dynamics Investigations of the Speciation and Reactivity of Deep Eutectic Electrolytes in Aluminum Batteries.

Invited for this month's cover is the Section for Atomic Scale Materials Modelling led by Prof. Tejs Vegge at the Department of Energy Conversion and Storage, Technical University of Denmark. The central image of the cover picture illustrates one of the chemical reaction mechanisms observed in a deep eutectic electrolyte formed by AlCl3 and urea. This is a promising electrolyte for inexpensive and environmentally friendly next-generation batteries based on aluminum. We have developed the computational techniques needed to identify chemical species and track reaction mechanisms across an ab initio molecular dynamics trajectory. The reaction mechanisms and speciation observed help to gain more insight in the development of such batteries. The Full Paper itself is available at 10.1002/cssc.202100163.

Ab initio Molecular Dynamics Investigations of the Speciation and Reactivity of Deep Eutectic Electrolytes in Aluminum Batteries.

Deep eutectic solvents (DESs) have emerged as an alternative for conventional ionic liquids in aluminum batteries. Elucidating DESs composition is fundamental to understand aluminum electrodeposition in the battery anode. Despite numerous experimental efforts, the speciation of these DESs remains elusive. This work shows how ab initio molecular dynamics (AIMD) simulations can shed light on the molecular composition of DESs. For the particular example of AlCl3 :urea, one of the most popular DESs, we carried out a systematic AIMD study, showing how an excess of AlCl3 in the AlCl3 :urea mixture promotes the stability of ionic species vs neutral ones and also favors the reactivity in the system. These two facts explain the experimentally observed enhanced electrochemical activity in salt-rich DESs. We also observe the transfer of simple [AlClx (urea)y ] clusters between different species in the liquid, giving rise to free [AlCl4 ]- units. The small size of these [AlCl4 ]- units favors the transport of ionic species towards the anode, facilitating the electrodeposition of aluminum.

Understanding the Molecular Structure of the Elastic and Thermoreversible AlCl3 : Urea/Polyethylene Oxide Gel Electrolyte.

It is possible to prepare elastic and thermoreversible gel electrolytes with significant electroactivity by dissolving minimal weight fractions of ultra-high molecular weight polyethylene oxide (UHMW PEO) in an aluminum deep eutectic solvent (DES) electrolyte composed of AlCl3 and urea at a molar ratio of 1.5 : 1 (AlCl3 /urea). The experimental vibrational spectra (FTIR and Raman) provide valuable information on the structure and composition of the gel electrolyte. However, the complexity of this system requires computational simulations to help interpretation of the experimental results. This combined approach allows us to elucidate the speciation of the DES liquid electrolyte in the gel and how it interacts with the polymer chains to give rise to an elastic network that retains the electroactivity of the liquid electrolyte to a very great extent. The observed reactions occur between the ether in the polymer and both the amine groups in urea and the aluminum species. Thus, similar elastomeric gels may likely be prepared with other aluminum liquid electrolytes, making this procedure an effective way to produce families of gel aluminum electrolytes with tunable rheology and electroactivity.

Surface Pyroelectricity in Cubic SrTiO3.

Symmetry-imposed restrictions on the number of available pyroelectric and piezoelectric materials remain a major limitation as 22 out of 32 crystallographic material classes exhibit neither pyroelectricity nor piezoelectricity. Yet, by breaking the lattice symmetry it is possible to circumvent this limitation. Here, using a unique technique for measuring transient currents upon rapid heating, direct experimental evidence is provided that despite the fact that bulk SrTiO3 is not pyroelectric, the (100) surface of TiO2 -terminated SrTiO3 is intrinsically pyroelectric at room temperature. The pyroelectric layer is found to be ≈1 nm thick and, surprisingly, its polarization is comparable with that of strongly polar materials such as BaTiO3 . The pyroelectric effect can be tuned ON/OFF by the formation or removal of a nanometric SiO2 layer. Using density functional theory, the pyroelectricity is found to be a result of polar surface relaxation, which can be suppressed by varying the lattice symmetry breaking using a SiO2 capping layer. The observation of pyroelectricity emerging at the SrTiO3 surface also implies that it is intrinsically piezoelectric. These findings may pave the way for observing and tailoring piezo- and pyroelectricity in any material through appropriate breaking of symmetry at surfaces and artificial nanostructures such as heterointerfaces and superlattices.

Descriptors and Thermodynamic Limitations of Electrocatalytic Carbon Dioxide Reduction on Rutile Oxide Surfaces.

A detailed understanding of the electrochemical reduction of CO2 into liquid fuels on rutile metal oxide surfaces is developed by using DFT calculations. We consider oxide overlayer structures on RuO2 (1 1 0) surfaces as model catalysts to elucidate the trends and limitations in the CO2 reduction reaction (CO2RR) based on thermodynamic analysis. We aim to specify the requirements for CO2RR catalysts to establish adsorbate scaling relations and use these to derive activity volcanoes. Computational results show that the OH* binding free energy is a good descriptor of the thermodynamic limitations and it defines the left leg of the activity volcano for CO2RR. HCOOH* is a key intermediate for products formed through further reduction, for example, methanediol, methanol, and methane. The surfaces that do not bind HCOOH* are selective towards formic acid (HCOOH) production, but hydrogen evolution limits their suitability. We determine the ideal binding free energy for H* and OH* to facilitate selective CO2RR over H2 /CO evolution to be ΔGB [H]>0.5 eV and -0.5 eV<ΔGB [OH]<0.1 eV. The Re-containing overlayers considered in this work display excellent promise for selectivity, although they are active at a highly reducing potential.

Quantization of Hall Resistance at the Metallic Interface between an Oxide Insulator and SrTiO_{3}.

The two-dimensional metal forming at the interface between an oxide insulator and SrTiO_{3} provides new opportunities for oxide electronics. However, the quantum Hall effect, one of the most fascinating effects of electrons confined in two dimensions, remains underexplored at these complex oxide heterointerfaces. Here, we report the experimental observation of quantized Hall resistance in a SrTiO_{3} heterointerface based on the modulation-doped amorphous-LaAlO_{3}/SrTiO_{3} heterostructure, which exhibits both high electron mobility exceeding 10,000 cm^{2}/V s and low carrier density on the order of ∼10^{12} cm^{-2}. Along with unambiguous Shubnikov-de Haas oscillations, the spacing of the quantized Hall resistance suggests that the interface is comprised of a single quantum well with ten parallel conducting two-dimensional sub-bands. This provides new insight into the electronic structure of conducting oxide interfaces and represents an important step towards designing and understanding advanced oxide devices.

Temperature dependent reversible p-n-p type conduction switching with colossal change in thermopower of semiconducting AgCuS.

Semiconductors have been fundamental to various devices that are typically operated with electric field, such as transistors, memories, sensors, and resistive switches. There is growing interest in the development of novel inorganic materials for use in transistors and semiconductor switches, which can be operated with a temperature gradient. Here, we show that a crystalline semiconducting noble metal sulfide, AgCuS, exhibits a sharp temperature dependent reversible p-n-p type conduction switching, along with a colossal change in the thermopower (ΔS of ~1757 µV K(-1)) at the superionic phase transition (T of ~364 K). In addition, its thermal conductivity is ultralow in 300-550 K range giving AgCuS the ability to maintain temperature gradients. We have developed fundamental understanding of the phase transition and p-n-p type conduction switching in AgCuS through temperature dependent synchrotron powder X-ray diffraction, heat capacity, Raman spectroscopy, and positron annihilation spectroscopy measurements. Using first-principles calculations, we show that this rare combination of properties originates from an effective decoupling of electrical conduction and phonon transport associated with electronic states of the rigid sulfur sublattice and soft vibrations of the disordered cation sublattices, respectively. Temperature dependent p-n-p type conduction switching makes AgCuS an ideal material for diode or transistor devices that operate reversibly on temperature or voltage changes near room temperature.