Search

We have developed an automated and efficient scheme for the fitting of data using Curvature Constrained Splines (CCS), to construct accurate two-body potentials. The approach enabled the construction of an oscillation-free, yet flexible, potential. We show that the optimization problem is convex and that it can be reduced to a standard Quadratic Programming (QP) problem. The improvements are demonstrated by the development of a two-body potential for Ne from ab initio data. We also outline possible extensions to the method. Program summary Program Title: CCS CPC Library link to program files: http://dx.doi.org/10.17632/7dt5nzxgbs.1 Developer's repository link:gttp://github.com/aksam432/CCS Licensing provisions: GPLv3 Programming language: Python External routines/libraries: NumPy, matplotlib, ASE, CVXOPT Nature of problem: Ab initio quantum chemistry methods are often computationally very expensive. To alleviate this problem, the development of efficient empirical and semi-empirical methods is necessary. Two-body potentials are ubiquitous in empirical and semi-empirical methods. Solution method: The CCS package provides a new strategy to obtain accurate two body potentials. The potentials are described as cubic splines with curvature constraints.

The CO2 molecule is weakly bound in water. Here we analyze the influence of a dissolved CO2 molecule on the structure and OH vibrational spectra of the surrounding water. From the analysis of ab initio molecular dynamics simulations (BLYP-D3) we present static (structure, coordination, H-bonding, tetrahedrality) and dynamical (OH vibrational spectra) properties of the water molecules as a function of distance from the solute. We find a weakly oscillatory variation (“ABBA”) in the ‘solution minus bulk water’ spectrum. The origin of these features can largely be traced back to solvent–solute hard-core interactions which lead to variations in density and tetrahedrality when moving from the solute’s vicinity out to the bulk region. The high-frequency peak in the solute-affected spectra is specifically analyzed and found to originate from both water OH groups that fulfill the geometric H-bond criteria, and from those that do not (dangling ones). Effectively, neither is hydrogen-bonded.

Atomic neural networks (ANNs) constitute a class of machine learning methods for predicting potential energy surfaces and physicochemical properties of molecules and materials. Despite many successes, developing interpretable ANN architectures and implementing existing ones efficiently are still challenging. This calls for reliable, general-purpose, and open-source codes. Here, we present a python library named PiNN as a solution toward this goal. In PiNN, we designed a new interpretable and high-performing graph convolutional neural network variant, PiNet, as well as implemented the established Behler-Parrinello neural network. These implementations were tested using datasets of isolated small molecules, crystalline materials, liquid water, and an aqueous alkaline electrolyte. PiNN comes with a visualizer called PiNNBoard to extract chemical insight "learned" by ANNs. It provides analytical stress tensor calculations and interfaces to both the atomic simulation environment and a development version of the Amsterdam Modeling Suite. Moreover, PiNN is highly modularized, which makes it useful not only as a standalone package but also as a chain of tools to develop and to implement novel ANNs. The code is distributed under a permissive BSD license and is freely accessible at https://github.com/Teoroo-CMC/PiNN/with full documentation and tutorials.

We propose that crystalline ionic hydrates constitute a valuable resource for benchmarking theoretical methods for aqueous ionic systems. Many such structures are known from the experimental literature, and they contain a large variety of water–water and ion–water structural motifs. Here we have collected a data set (CRYSTALWATER50) of 50 structurally unique "in-crystal" water molecules, involved in close to 100 nonequivalent O–H···O hydrogen bonds. A dozen well-known DFT functionals were benchmarked with respect to their ability to describe these experimental structures and their OH vibrational frequencies. We find that the PBE, RPBE-D3, and optPBE-vdW methods give the best H-bond distances and that anharmonic OH frequencies generated from B3LYP//optPBE-vdW energy scans outperform the other methods, i.e., here we performed B3LYP energy scans along the OH stretching coordinate while the rest of the structure was kept fixed at the optPBE-vdW-optimized positions., Title in thesis list of papers: Fifty shades of water: Benchmarking DFT functionals against diffraction and spectroscopic data for crystalline hydrates

The adsorption of SO2 on single crystalline TiO2(110) has been investigated by means of polarized infrared reflection-absorption spectroscopy (IRRAS) experiments and density functional theory (DFT) calculations. IR absorption bands were detected at 1324 cm-1 and 985 cm-1 with p-polarized light incident along both the [110] and [001] crystallographic directions at 123 K. When the temperature was increased to 153 K, the peak at 1324 cm-1 disappears, while a new, weak band appears at 995 cm-1. Simultaneously, a band at 995 cm-1 also emerges with s-polarized light along the [110] direction. Based on the symmetry properties of the IRRAS spectra and accompanying ab initio simulations of the spectra employing a three layer model (vacuum-adsorbate-substrate), it is shown that the low temperature absorption IRRAS bands can be attributed to an SO3-like adsorbate structure. This is also the most stable adsorption structure (Ead = - 0.58 eV) on the stoichiometric surface. The combined IRRAS and DFT results show that the band appearing at 995 cm-1 is associated with a surface sulfite specie which is stabilized by residual surface water. The DFT calculations also revealed that a stable adsorption structure exists on a reduced TiO2 surface, where SO2 binds strongly to an oxygen vacancy site. It is suggested that this is an intermediate that form surface sulfate upon further reactions with water, although it was not observed on the stoichiometric surface studied in this work.

Unraveling the atomistic details of solid/liquid interfaces, e.g., by means of vibrational spectroscopy, is of vital importance in numerous applications, from electrochemistry to heterogeneous catalysis. Water-oxide interfaces represent a formidable challenge because a large variety of molecular and dissociated water species are present at the surface. Here, we present a comprehensive theoretical analysis of the anharmonic OH stretching vibrations at the water/ZnO(10 (1) over bar0) interface as a prototypical case. Molecular dynamics simulations employing a reactive high-dimensional neural network potential based on density functional theory calculations have been used to sample the interfacial structures. In the second step, one-dimensional potential energy curves have been generated for a large number of configurations to solve the nuclear Schrodinger equation. We find that (i) the ZnO surface gives rise to OH frequency shifts up to a distance of about 4 angstrom from the surface; (ii) the spectrum contains a number of overlapping signals arising from different chemical species, with the frequencies decreasing in the order v (adsorbed hydroxide) > v (non-adsorbed water) > v (surface hydroxide) > v (adsorbed water); (iii) stretching frequencies are strongly influenced by the hydrogen bond pattern of these interfacial species. Finally, we have been able to identify substantial correlations between the stretching frequencies and hydrogen bond lengths for all species.

This paper concerns thin water films and their hydrogen-bond patterns on ionic surfaces. As far as we are aware, this is the first time H-bond correlations for surface water and hydroxide species are presented in the literature while hydrogen-bond relations in the solid state have been scrutinized for at least five decades. Our data set, which was derived using density functional theory, consists of 116 unique surface OH groups–intact water molecules as well as hydroxides–on MgO(001), CaO(001) and NaCl(001), covering the whole range from strong to weak to no H-bonds. The intact surface water molecules are found to always be redshifted with respect to the gas-phase water OH vibrational frequency, whereas the surface hydroxide groups are either redshifted (OsH) or blueshifted (OHf) compared to the gas-phase OH– frequency. The surface H-bond relations are compared with the traditional relations for bulk crystals. We find that the “ν(OH) vs R(H···O)” correlation curve for surface water does not coincide with the solid state curve: it is redshifted by about 200 cm–1 or more. The intact water molecules and hydroxide groups on the ionic surfaces essentially follow the same H-bond correlation curve.

We analyse the OH vibrational signatures of water molecules and hydroxide ions on thin water films on MgO(001) and CaO(001), using DFT-generated anharmonic potential energy surfaces. We find that the OH stretching frequencies of intact water molecules on the surface are always downshifted with respect to the gas-phase species while the OH– groups are either upshifted or downshifted. Despite these differences, the main characteristics of the frequency shifts for all three types of surface OH groups (OHw, OsH and OHf) can be accounted for by one unified expression involving the in situ electric field from the surrounding environment, and the molecular properties of the vibrating species (H2O or OH–). The origin behind the different red- and blueshift behaviour can be traced back to the fact that the molecular dipole moment of a gas-phase water molecule increases when an OH bond is stretched, but the opposite is true for the hydroxide ion. We propose that familiarity with the relations presented here will help surface scientists in the interpretation of vibrational OH spectra for thin water films on ionic crystal surfaces.

STM experiments on CeO2(111) reveal depressions in the surface oxygen sub-lattice which are observed to form clusters of various shapes and sizes [1].While these depressions were assumed to be oxygen vacancies, subsequent DFTcalculations have indicated that clusters of oxygen vacancies are energeticallyunstable [2-4]. Recently, we showed theoretically that fluorine impurities shouldappear almost identical to oxygen vacancies in STM experiments, but that theirproperties are more in line with those of the defects observed in experiments [5].Here, I will present the results of a further investigation into the distribution ofF impurity clusters at CeO2(111), using a combination of DFT+U calculations,and Monte Carlo sampling based on a simple but accurate pair potential whichwas fitted to the DFT results. The distribution is characterised in terms of thenumber of clusters of a certain size, and also on their topology, i.e. whetherthey are compact or open/linea r. Our results compare favourably with theexperiments, and also exhibit some interesting physics in their own right. [1] F. Esch et al., Science 309, 752 (2005).[2] J. Conesa, Cat. Today 143, 315 (2009).[3] C. Zhang et al., Phys. Rev. B 79, 075433 (2009).[4] X.-P. Wu & X.-Q. Gong, Phys. Rev. Lett. 116, 086102 (2016).[5] J. Kullgren, M. J. Wolf et al., Phys. Rev. Lett. 112, 156102 (2014).

The interplay between energetics and entropy in determining defect distributions at ceria(111) is studied using a combination of DFT+U and lattice Monte Carlo simulations. Our main example is fluorine impurities, although we also present preliminary results for surface hydroxyl groups. A simple classical force-field model was constructed from a training set of DFT+U data for all symmetrically inequivalent (F-)(n)(Ce3+)(n) nearest-neighbor clusters with n = 2 or 3. Our fitted model reproduces the DFT energies well. We find that for an impurity concentration of 15% at 600 K, straight and hooked linear fluorine clusters are surprisingly abundant, with similarities to experimental STM images from the literature. We also find that with increasing temperature the fluorine cluster sizes show a transition from being governed by an attractive potential to being governed by a repulsive potential as a consequence of the increasing importance of the entropy of the Ce3+ ions. The distributions of surface hydroxyl groups are noticeably different.

The hydration-shell of CO2 is characterized using Raman multivariate curve resolution (Raman-MCR) spectroscopy combined with ab initio molecular dynamics (AIMD) vibrational density of states simulations, to validate our assignment of the experimentally observed high-frequency OH band to a weak hydrogen bond between water and CO2. Our results reveal that while the hydration-shell of CO2 is highly tetrahedral, it is also occasionally disrupted by the presence of entropically stabilized defects associated with the CO2-water hydrogen bond. Moreover, we find that the hydration-shell of CO2 undergoes a temperature-dependent structural transformation to a highly disordered (less tetrahedral) structure, reminiscent of the transformation that takes place at higher temperatures around much larger oily molecules. The biological significance of the CO2 hydration shell structural transformation is suggested by the fact that it takes place near physiological temperatures.

We present periodic plane-wave density functional theory (DFT) Perdew-Burke-Ernzerhof (PBE-D2) calculations for four highly hydrated crystals, Na2CO3 center dot 10H(2)O, MgSO4 center dot 7H(2)O, MgSO4 center dot 11H(2)O, and Al(NO3)(3)center dot 9H(2)O, containing 37 structurally unique water molecules and 74 unique hydrogen bonds. The calculated R(H center dot center dot center dot O) distances lie in the range 1.60-2.05 angstrom, the anharmonic OH frequencies in the range 2570-3425 cm(-1), and the water dipole moments lie in the range 2.9-4.3 Debye, as calculated from the Wannier function centers and the nuclei. We present the following findings. (i) Our optimized intramolecular r(OH) distances are always larger than the gasphase value and thus more accurate than those derived from neutron diffraction experiments; (ii) The local in situ electric field over the molecule, calculated from the positions of the nuclei and the Wannier centers in the surrounding crystal, appears to be a good descriptor of the pertturbation from the water molecule's surroundings as the internal molecular properties (r(e), nu, mu) are found to correlate well with the crystal-generated electric field; (iii) We have added DFT-calculated data points to the well-known experimental 'OH frequency versus R(H center dot center dot center dot O) correlation curve in a region where the experimental data points are scarce; (iv) For all 37 water molecules, the Wannier centers located in the lone-pair region, and those located in the OH bonds, displace about equally much due to the polarizing environment. Finally, we propose that our resulting 'OH frequency versus Wannier-type electric field' correlation curve may constitute a useful tool for predicting OH vibrational frequencies from snapshots from PBE-D2-based ab initio molecular dynamics simulations of water-containing systems.

We present experimental and calculated IR spectra of the water molecules in crystalline aluminium nitrate nonahydrate and a method to generate a realistic and well resolved isotope-isolated spectrum from periodic DFT calculations. Our sample crystal contains 18 structurally different OH groups and is a perfect benchmark compound to validate vibrational models and the structure-property relationship of bound water molecules. FTIR spectra (ATR technique) were recorded for the Al(NO3)(3)center dot 9H(2)O crystal at 138 and 298 K, and due to a multitude of OH contributions and couplings, they are naturally poorly resolved and yield a broad OH band in the range 3500 to 2700 cm(-1) at both temperatures. Isotope-isolated IR spectra have the clear advantage over non-deuterated spectra that they are better resolved and easier to interpret - here we have extended the experimental study by simulating the isotope-isolated IR spectrum, using PBE-D2 and auxiliary B3LYP calculations and an anharmonic OH vibrational model. We find excellent agreement between the shapes and frequency ranges of the experimental and calculated OH spectral bands. We make use of four different vibrational models: (i) a harmonic lattice-dynamical model for the isotope-isolated crystal with 1 H among 71 D, (ii) a harmonic lattice-dynamical model for the normal undeuterated crystal involving all the vibrational couplings, (iii) a harmonic 1-dimensional uncoupled OH vibrational model, and (iv) the anharmonic variant of the previous model, which yields the final spectrum. We also use the individual frequencies, resolved by the calculations, to quantify new or extended relationships involving OH frequencies versus local electric fields and H-bond distances. We explore the correlation between OH frequency and molecular dipole moment for bound water molecules.

Precise molecular-level information on the water molecule is precious, since it affects our interpretation of the role of water in a range of important applications of aqueous media. Here we propose that electronic structure calculations for highly hydrated crystals yield such information. Properties of nine structurally different water molecules (19 independent O center dot center dot center dot O hydrogen bonds) in the Al(NO3)(3) center dot 9H(2)O crystal have been calculated from DFT calculations. We combine the advantage of studying different water environments using one and the same compound and method (instead of comparing a set of independent experiments, each with its own set of errors) with the advantage of knowing the exact atomic positions, and the advantage of calculating properties that are difficult to extract from experiment. We find very large Wannier dipole moments for H2O molecules surrounding the cations: 4.0-4.3 D (compared to our calculated value of 1.83 D in the gas phase). These are induced by the ions and the H-bonds, while other water interactions and the relaxation of the internal water geometry in fact decrease the dipole moments. We find a good correlation between the water dipole moment and the O center dot center dot center dot O distances, and an even better (non-linear) correlation with the average electric field over the molecule. Literature simulation data for ionic aqueous solutions fit quite well with our crystalline `dipole moment vs. O center dot center dot center dot O distance' curve. The progression of the water and cation charges from ` small clusters double right arrow large clusters double right arrow the crystal' helps explain why the net charges on all the water molecules are so small in the crystal.