Computational prediction of Lee retention indices of polycyclic aromatic hydrocarbons by using machine learning

Given the difficult of experimental determination, quantitative structure-property relationship (QSPR) and deep learning (DL) provide an important tool to predict physicochemical property of chemical compounds. In this paper, partial least squares (PLS), genetic function approximation (GFA), and deep neural network (DNN) were used to predict the Lee retention index (Lee-RI) of PAHs in SE-52 and DB-5 stationary phases. Four molecular descriptors, molecular weight (MW), quantitative estimate of drug-likeness (QED), atomic charge weighted negative surface area (Jurs_PNSA_3), and relative negative charge (Jurs_RNCG) were selected to construct regression models based on genetic algorithm. For SE-52, PLS model showed best prediction power, followed by DNN and GFA. The relative error (RE), root mean square error (RMSE), and regression coefficient (R² ) of best PLS regression model are 1.228%, 5.407, and 0.980. For DB-5, DNN model showed best prediction power, followed by GFA and PLS. The RE, RMSE and R² of best DNN regression model for DB-5-1 and DB-5-2 are 1.058%, 4.325%, 0.976%, 0.821%, 3.795%, and 0.970%, respectively. The three regression models not only show good predictive ability, but also highlight the stability and ductility of the models.

Understanding fast-ion conduction in solid electrolytes

The ability of some solid materials to exhibit exceptionally high ionic conductivities has been known since the observations of Michael Faraday in the nineteenth century (Faraday M. 1838 Phil. Trans. R. Soc. 90), yet a detailed understanding of the atomic-scale physics that gives rise to this behaviour remains an open scientific question. This theme issue collects articles from researchers working on this question of understanding fast-ion conduction in solid electrolytes. The issue opens with two perspectives, both of which discuss concepts that have been proposed as schema for understanding fast-ion conduction. The first perspective presents an overview of a series of experimental NMR studies, and uses this to frame discussion of the roles of ion-ion interactions, crystallographic disorder, low-dimensionality of crystal structures, and fast interfacial diffusion in nanocomposite materials. The second perspective reviews computational studies of halides, oxides, sulfides and hydroborates, focussing on the concept of frustration and how this can manifest in different forms in various fast-ion conductors. The issue also includes five primary research articles, each of which presents a detailed analysis of the factors that affect microscopic ion-diffusion in specific fast-ion conducting solid electrolytes, including oxide-ion conductors [Formula: see text] and [Formula: see text], lithium-ion conductors [Formula: see text] and [Formula: see text], and the prototypical fluoride-ion conductor [Formula: see text]-[Formula: see text]. This article is part of the Theo Murphy meeting issue 'Understanding fast-ion conduction in solid electrolytes'.

Analysis of Diffusion in Solid-State Electrolytes through MD Simulations, Improvement of the Li-Ion Conductivity in β-Li3PS4 as an Example

Molecular dynamics simulations are a powerful tool to study diffusion processes in battery electrolyte and electrode materials. From molecular dynamics simulations, many properties relevant to diffusion can be obtained, including the diffusion path, amplitude of vibrations, jump rates, radial distribution functions, and collective diffusion processes. Here it is shown how the activation energies of different jumps and the attempt frequency can be obtained from a single molecular dynamics simulation. These detailed diffusion properties provide a thorough understanding of diffusion in solid electrolytes, and provide direction for the design of improved solid electrolyte materials. The presently developed analysis methodology is applied to DFT MD simulations of Li-ion diffusion in β-Li₃PS₄. The methodology presented is generally applicable to diffusion in crystalline materials and facilitates the analysis of molecular dynamics simulations. The code used for the analysis is freely available at: https://bitbucket.org/niekdeklerk/md-analysis-with-matlab. The results on β-Li₃PS₄ demonstrate that jumps between bc planes limit the conductivity of this important class of solid electrolyte materials. The simulations indicate that the rate-limiting jump process can be accelerated significantly by adding Li interstitials or Li vacancies, promoting three-dimensional diffusion, which results in increased macroscopic Li-ion diffusivity. Li vacancies can be introduced through Br doping, which is predicted to result in an order of magnitude larger Li-ion conductivity in β-Li₃PS₄. Furthermore, the present simulations rationalize the improved Li-ion diffusivity upon O doping through the change in Li distribution in the crystal. Thus, it is demonstrated how a thorough understanding of diffusion, based on thorough analysis of MD simulations, helps to gain insight and develop strategies to improve the ionic conductivity of solid electrolytes.

Design of fast ion conducting cathode materials for grid-scale sodium-ion batteries

The obvious cost advantage as well as attractive electrochemical properties, including excellent cycling stability and the potential of high rate performance, make sodium-ion batteries prime candidates in the race to technically and commercially enable large-scale electrochemical energy storage. In this work, we apply our bond valence site energy modelling method to further the understanding of rate capabilities of a wide range of potential insertion-type sodium-ion battery cathode materials. We demonstrate how a stretched exponential function permits us to systematically quantify the rate performance, which in turn reveals guidelines for the design of novel sodium-ion battery chemistries suitable for high power, grid-scale applications. Starting from a diffusion relaxation model, we establish a semi-quantitative prediction of the rate-performance of half-cells from the structure of the cathode material that factors in dimensionality of Na⁺ ion migration pathways, the height of the migration barriers and the crystallite size of the active material. With the help of selected examples, we also illustrate the respective roles of unoccupied low energy sites within the pathway and temperature towards the overall rate capability of insertion-type cathode materials.

Vibrational spectroscopic and bond valence study of structure and bonding in Al2O3-containing AgI–AgPO3 glasses

We present a detailed investigation of the effects of synthesis conditions on glasses xAgI–(1 − x)AgPO₃ with 0 ≤ x ≤ 0.4.

First and Second Universalities: Expeditions Towards and Beyond

Understanding the mechanisms of translational and localised ionic movements in disordered materials has seen intense activity spanning several decades. This article attempts to convey a concise overview of our contribution to this field over the period from 2005 to 2010 and to place it in its broad context.

Many-ion Dynamics: The Common View of CM and MC

For solution of the problem of conductivity relaxation and diffusion of ions in ionic conductors with high density of ions, it is essential not to neglect treatment of the effects of many-ion dynamics. This view is shared by the Coupling Model (CM) and the MIGRATION CONCEPT (MC), although the treatment, emphasis and some predictions of the two models are different. Notwithstanding, a basic element is common to both models, namely the primitive relaxation, which performs two important functions. It terminates the caged ion dynamics at short times and initiates the many-ion dynamics at longer times. We demonstrate by experiments and molecular dynamics simulations the existence of the primitive relaxation, and the two functions it performs. The relation of the primitive relaxation time to the conductivity relaxation time predicted by the CM is shown to hold in all cases considered.

Nearly Constant Loss in Lithium and LiCl-doped Borate Glasses

Measurements of the complex dielectric permittivity in (Li2O)x(B2O3)100-x and halide-doped (Li2O)x (LiCl)y(B2O3)100-x-y glasses conducted in the nearly constant loss (NCL) regime at low temperatures are reported. Scaling properties reveal that the NCL in (Li2O)x(B2O3)100-x is largely ionic in origin for these materials and increases in proportion to the Li2O content. However, at low Li2O content, the NCL becomes dominated by a non-ionic loss arising from the borate network or other impurity polarizations. Halide doping is seen to produce an additional enhancement of the NCL that is consistent with its known enhancement of the bulk conduction.

Models for Ion Transport in Amorphous Materials: Recent Advances

By the development and investigation of coarse-grained models for ion motion in disordered systems, important progress has been achieved in the past to explain a large variety of ion transport properties in amorphous materials. In this work we discuss recent theoretical advances, which allow one to better understand the nearly constant dielectric loss in disordered solid ionic conductors, the mixed glass former effect in glassy electrolytes, and certain effects of ion motion in polymer electrolytes.

Nearly constant loss effects in borate glasses

Different nearly constant loss phenomena are investigated in borate glasses with compositions xNa(2)O.(1-x)B(2)O(3), for 0 < or =x< or = 0.3. The ionic conductivities caused by these effects are studied in wide ranges of temperature and frequency, spanning 4.3 K to 573 K and 100 mHz to 1 MHz, respectively. In a first step, we show how to identify the nearly constant loss (NCL) in 0.3Na(2)O.0.7B(2)O(3) glass. In the procedure, the scaling property of the conductivity caused by ordinary hopping is used to remove this component from the total conductivity as measured as a function of temperature at fixed frequency. The resulting NCL component is seen to be proportional to frequency and to display no temperature dependence. In a second step, a broad-band relaxation process is shown to exist in amorphous boron oxide and in sodium borate glasses with x< or = 0.1. It is most probably due to the presence of traces of water, with hydrogen ions behaving as reorienting and interacting local dipoles. In a third step, we propose a simple formal treatment of the NCL phenomenon, tracing it back to a large number of interacting ions, each of them moving locally. The key aspect is a "see-saw-type" time dependence of their individual single-particle double-well potentials, which is due to their Coulomb interactions. The individual ion does, therefore, not require thermal activation and is thus kept in motion even at cryogenic temperatures.

Conductivity dispersion in supercooled calcium potassium nitrate: caged ionic motion viewed as part of standard behaviour

Conductivity spectra of ionic materials with disordered structures are usually thought to consist of several parts, i.e., the DC conductivity, a power-law component, a nearly-constant-loss feature (if identified) and the (far-)infrared conductivity caused by vibrational motion. Such a decomposition may, however, easily lead to a misinterpretation of the underlying dynamics. Here, we discuss broad-band conductivity data of the supercooled glass-forming melt calcium potassium nitrate, of composition 0.4 Ca(NO(3))(2).0.6 KNO(3), often abbreviated as CKN. Data have been taken at frequencies up to the far infrared. We show that the frequency-dependent conductivity is very well reproduced by a superposition of only two components. One of them is due to vibrations, the other is caused by displacements of the mobile ions. The latter component, which does not follow a power law, is described in terms of a physical model called the MIGRATION concept. This model treatment has been found to apply in many solid electrolytes as well and is, therefore, considered to provide a "standard" formulation of the ion dynamics. The gradual transition from a correlated forward-backward ("caged") ionic motion to a stepwise translational motion may be regarded as the main feature of the MIGRATION concept.

Conductivity Spectra of Polyphosphazene-Based Polyelectrolyte Multilayers

Polyelectrolyte multilayers are built up from ionically modified polyphosphazenes by layer-by-layer assembly of a cationic (poly[bis(3-amino-N,N,N-trimethyl-1-propanaminium iodide)phosphazene] (PAZ+) and an anionic poly[bis(lithium carboxylatophenoxy)phosphazene] (PAZ-). In comparison, multilayers of poly(sodium 4-styrenesulfonate) (PSS) and poly(allylamine hydrochloride) (PAH) are investigated. Frequency-dependent conductivity spectra are taken in sandwich geometry at controlled relative humidity. Conductivity spectra of ion-conducting materials generally display a dc plateau at low frequencies and a dispersive regime at higher frequencies. In the present case, the dispersive regime shows a frequency dependence, which is deviating from the typical behavior found in most ion-conducting materials. Dc conductivity values, which can be attributed to long-range ionic transport, are on the order of sigmadc = 10-10-10-7 S.cm-1 and strongly depend on relative humidity. For PAZ+/PAZ- multilayers sigmadc is consistently larger by one decade as compared to PSS/PAH layers, while the humidity dependence is similar, pointing at general mechanisms. A general law of a linear dependence of log(sigmadc) on relative humidity is found over a wide range of humidity and holds for both multilayer systems. This very strong dependence was attributed to variations of the ion mobility with water content, since the water content itself is not drastically dependent on humidity.

Oxygen-deficient perovskites: linking structure, energetics and ion transport

The present review focuses on links between structure, energetics and ion transport in oxygen-deficient perovskite oxides, ABO(3-delta). The perfect long-range order, convenient for interpretations of the structure and properties of ordered materials, is evidently not present in disordered materials and highly defective perovskite oxides are spatially inhomogeneous on an intermediate length scale. Although this makes a fundamental description of these and other disordered materials very difficult, it is becoming increasingly clear that this complexity is often essential for the functional properties. In the present review we advocate a potential energy barrier description of the disordered state in which the possible local (or inherent) structures are seen to correspond to separate local minima on the potential energy surface. We interpret the average structure observed experimentally at any temperature as a time and spatial average of the different local structures which are energetically accessible. The local structure is largely affected by preferences for certain polyhedron coordinations and the oxidation state stability of the transition metals, and the strong long-range electrostatic interactions present in non-stoichiometric oxides imply that only a small fraction of the local energy minima on the potential energy surface are accessible at most temperatures. We will show that models neglecting the spatial inhomogeneity and thus the local structure serve as useful empirical tools for particular purposes, e.g. for understanding the main features of the complex redox properties that are so crucial for many applications of these oxides. The short-range order is on the other hand central for understanding ionic transport. Oxide ion transport involves the transformation of one energetically accessible local structure into another. Thus, strongly correlated transport mechanisms are expected; in addition to the movement of the oxygen ions giving rise to the transport, other ions are involved and even the A and B atoms move appreciably in a cooperative fashion along the transition path. Such strongly correlated or collective ionic migration mechanisms should be considered for fast oxide ion conductors in general and in particular for systems forming superstructures at low temperatures. Structural criteria for fast ion conduction are discussed. A high density of low-lying local energy minima is certainly a prerequisite and for perovskite-related A(2)B(2)O(5) oxides, those containing B atoms that have energetic preference for tetrahedral coordination geometry are especially promising.

Low-Temperature Phases of Rubidium Silver Iodide: Crystal Structures and Dynamics of the Mobile Silver Ions

Recently, broad-band conductivity spectra have been taken in the low-temperature gamma-phase of the archetypal fast ion conductor RbAg4I5. Attempts to reproduce the experimental data in a simple model calculation have led to the conclusion that strictly localized displacive movements of interacting ionic charge carriers should play an important role in the low-temperature phase. However, with no detailed structural study of gamma-RbAg4I5 available, the relevant processes could not be identified within the crystal structure. This state of affairs has triggered the present investigation of the structures of all three phases of rubidium silver iodide. Powder diffraction data of RbAg4I5 have been collected at the high-resolution powder diffractometer at ID31 at the European Synchrotron Radiation Facility (ESRF). The structure of the gamma-phase has been solved by successive Rietveld refinements in combination with difference Fourier analyses. The same structural principle is found to prevail in all three phases, interconnected distorted RbI6 octahedra forming a three-dimensional framework, which undergoes only displacive structural changes during the alpha-beta and beta-gamma phase transitions. With decreasing temperature, the disorder in the silver sublattice is found to decrease, and a clustering of the disordered silver ions is found to develop. In the gamma-phase, "pockets" containing partially occupied silver sites have been identified, and it is suggested that the localized displacive motion detected by conductivity spectroscopy is performed by the silver ions located within these pockets.