ConfSolv: Prediction of Solute Conformer-Free Energies across a Range of Solvents.

Predicting Gibbs free energy of solution is key to understanding the solvent effects on thermodynamics and reaction rates for kinetic modeling. Accurately computing solution free energies requires the enumeration and evaluation of relevant solute conformers in solution. However, even after generation of relevant conformers, determining their free energy of solution requires an expensive workflow consisting of several ab initio computational chemistry calculations. To help address this challenge, we generate a large data set of solution free energies for nearly 44,000 solutes with almost 9 million conformers calculated in 41 different solvents using density functional theory and COSMO-RS and quantify the impact of solute conformers on the solution free energy. We then train a message passing neural network to predict the relative solution free energies of a set of solute conformers, enabling the identification of a small subset of thermodynamically relevant conformers. The model offers substantial computational time savings with predictions usually substantially within 1 kcal/mol of the free energy of the solution calculated by using computational chemical methods.

Analyzing Learned Molecular Representations for Property Prediction.

Advancements in neural machinery have led to a wide range of algorithmic solutions for molecular property prediction. Two classes of models in particular have yielded promising results: neural networks applied to computed molecular fingerprints or expert-crafted descriptors and graph convolutional neural networks that construct a learned molecular representation by operating on the graph structure of the molecule. However, recent literature has yet to clearly determine which of these two methods is superior when generalizing to new chemical space. Furthermore, prior research has rarely examined these new models in industry research settings in comparison to existing employed models. In this paper, we benchmark models extensively on 19 public and 16 proprietary industrial data sets spanning a wide variety of chemical end points. In addition, we introduce a graph convolutional model that consistently matches or outperforms models using fixed molecular descriptors as well as previous graph neural architectures on both public and proprietary data sets. Our empirical findings indicate that while approaches based on these representations have yet to reach the level of experimental reproducibility, our proposed model nevertheless offers significant improvements over models currently used in industrial workflows.

Homoleptic borates and aluminates containing the difluorophosphato ligand - [M(O2PF2)x](y-) - synthesis and characterization.

Weakly coordinating anions (WCAs) with the difluorophosphato ligand (O2PF2) were the target of this study. Initial experiments were conducted towards the preparation of homoleptic aluminates of the well-studied [Al(OR)4](-)-type. The preparation of the initial target structure Li[Al(O2PF2)4] failed due to the remaining Lewis acidic character of the central aluminum atom. Instead, the formation of Li3[Al(O2PF2)6] and Al(O2PF2)3 was observed with hexacoordinate aluminum atoms and verified by NMR, IR and X-ray crystallography. A possible mechanism towards these compounds was postulated in the solvent induced dismutation of the tetracoordinate Li[Al(O2PF2)4]. A singly charged WCA was realized by the exchange of the central aluminum atom for boron. The [B(O2PF2)4](-) anion was prepared starting from BH3·S(CH3)2 and boron tribromide leading to the protic room temperature Ionic Liquid (IL) [H(S(CH3)2)][B(O2PF2)4] and the neat liquid Brønsted acid H[B(O2PF2)4], respectively, representing a significantly improved synthesis with regard to the first experiments of Dove et al. The basicity of the [B(O2PF2)4](-) anion and its WCA quality were investigated on the basis of the IR-spectroscopic NH-scale and the salt [H(N(Oct)3)][B(O2PF2)4] that places it better than all oxyanions and close to the carboranate based WCAs. A pathway to the solvent free pure Li[B(O2PF2)4] salt was established on a multi-gram scale with excellent purities enabling electrochemical applications (verified by NMR, IR, X-ray crystallography and cyclovoltammetry).

Li[B(OCH2CF3)4]: synthesis, characterization and electrochemical application as a conducting salt for LiSB batteries.

A new Li salt with views to success in electrolytes is synthesized in excellent yields from lithium borohydride with excess 2,2,2-trifluorethanol (HOTfe) in toluene and at least two equivalents of 1,2-dimethoxyethane (DME). The salt Li[B(OTfe)4 ] is obtained in multigram scale without impurities, as long as DME is present during the reaction. It is characterized by heteronuclear magnetic resonance and vibrational spectroscopy (IR and Raman), has high thermal stability (Tdecomposition >271 °C, DSC) and shows long-term stability in water. The concentration-dependent electrical conductivity of Li[B(OTfe)4 ] is measured in water, acetone, EC/DMC, EC/DMC/DME, ethyl acetate and THF at RT In DME (0.8 mol L(-1) ) it is 3.9 mS cm(-1) , which is satisfactory for the use in lithium-sulfur batteries (LiSB). Cyclic voltammetry confirms the electrochemical stability of Li[B(OTfe)4 ] in a potential range of 0 to 4.8 V vs. Li/Li(+) . The performance of Li[B(OTfe)4 ] as conducting salt in a 0.2 mol L(-1) solution in 1:1 wt % DME/DOL is investigated in LiSB test cells. After the 40th cycle, 86 % of the capacity remains, with a coulombic efficiency of around 97 % for each cycle. This indicates a considerable performance improvement for LiSB, if compared to the standard Li[NTf2 ]/DOL/DME electrolyte system.

Modeling the influence of salts on the critical micelle concentration of ionic surfactants.

We show for the first time that a phenomenological, augmented volume-based thermodynamics (aVBT) model is capable to predict the critical micelle concentrations of ionic surfactants, including ionic liquids, with added salts. The model also adjusts for the type of salt added by including its molecular volume, which might form a connection to the Hofmeister effect. The other physico-chemically relevant quantities included in the model include surface area and solvation enthalpies.

Establishing consistent van der Waals volumes of polyatomic ions from crystal structures.

Based on temperature (T) dependent crystal structure data of seven organic salts, a radii-based scheme for the calculation of the van der Waals volume (V(vdw)) is analyzed. The obtained volumes (V(vdw,r), r=radius-based) are nearly T independent. An ion volume partitioning scheme is proposed by fixing the anion volumes of [Cl](-), [Br](-), [I](-), [BF(4)](-), [PF(6)](-), [OTf](-) and [NTf(2)](-). The van der Waals volumes (V(vdw,r) (+/-)) of 48 ions are established, with low standard deviations (0.2-3.6 Å(3), 0.1-4.5 % of V(vdw,r) (+/-)). The ion volumes are independent of the counterion and one crystal structure already suffices for their derivation. Correlations of the viscosity with V(vdw,r) via a Litovitz ansatz and our recently derived Arrhenius-type approach prove that these volumes are suitable for the volume-based description and prediction of IL properties. The corresponding correlation coefficient for the latter is R(2)=0.86 for 40 ILs (354 data points) in the T range of 253-373 K.

Temperature dependence of the viscosity and conductivity of mildly functionalized and non-functionalized [Tf2N](-) ionic liquids.

A series of bis(trifluoromethylsulfonyl)imide ionic liquids (ILs) with classical as well as mildly functionalized cations was prepared and their viscosities and conductivities were determined as a function of the temperature. Both were analyzed with respect to Arrhenius, Litovitz and Vogel-Fulcher-Tammann (VFT) behaviors, as well as in the context of their molecular volume (V(m)). Their viscosity and conductivity are highly correlated with V(m)/T or related expressions (R(2) ≥0.94). With the knowledge of V(m) of new cations, these correlations allow the temperature-dependent prediction of the viscosity and conductivity of hitherto unknown, non- or mildly functionalized ILs with low error bars (0.05 and 0.04 log units, respectively). The influence of the cation structure and mild functionalization on the physical properties was studied with systematically altered cations, in which V(m) remained similar. The T(o) parameter obtained from the VFT fits was compared to the experimental glass temperature (T(g)) and the T(g)/T(o) ratio for each IL was calculated using both experimental values and Angell's relationship. With Walden plots we investigated the IL ionicity and interpreted it in relation to the cation effects on the physical IL properties. We checked the validity of these V(m)/T relations by also including the recently published variable temperature viscosity and conductivity data of the [Al(OR(F))(4)](-) ILs with R(F) =C(H)(CF(3))(2) (error bars for the prediction: 0.09 and 0.10 log units, respectively).

In silico predictions of the temperature-dependent viscosities and electrical conductivities of functionalized and nonfunctionalized ionic liquids.

The viscosity (η) and electrical conductivity (κ) of ionic liquids are, next to the melting point, the two key properties of general interest. The knowledge of temperature-dependent η and κ data before their first synthesis would permit a much more target-oriented development of ionic liquids. We present in this work a novel approach to predict the viscosity and electrical conductivity of an ionic liquid without further input of experimental data. For the viscosity, only some basic physical observables like the Gibbs solvation energy (ΔG(solv)(*,∞)), which was calculated at the affordable DFT-level (RI-)BP86/TZVP/COSMO, the molecular radius, calculated from the molecular volume V(m) of the ion volumes, and the symmetry number (σ), according to group theory, are necessary as input. The temperature dependency (253-373 K) of the viscosity (4-19000 mPa s) was modeled by an Arrhenius approach. An alternative way, which avoids the deficits of the Arrhenius relation by a series expansion in the exponential term, is also presented. On the basis of their close connection, the same set of parameters is suitable to describe the electrical conductivity as well (238-468 K, 0.003-193 mS/cm). Nevertheless, more elegant alternatives like the usage of the Stokes-Einstein/Nernst-Einstein relation or the Walden rule are highlighted in this work. During this investigation, we additionally found an approach to predict the dielectric constant ε* of an ionic liquid at 298 K by using V(m) and ΔG(solv)(*,∞) between ε* = 9 and 43.

An experimental and theoretical study of the aluminium species present in mixtures of AlCl3 with the ionic liquids [BMP]Tf2N and [EMIm]Tf2N.

It is known that nano- or microcrystalline aluminium may be electrodeposited from mixtures of AlCl(3) and the ionic liquids 1-butyl-1-methylpyrrolidinium bis(trifluoromethylsulfonyl)amide ([BMP]Tf(2)N) and 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)amide ([EMIm]Tf(2)N), and that two phases form with higher formal concentrations of AlCl(3) (at 1.6 mol L(-1) (x(Al)=0.33) and 2.5 mol L(-1) (x(Al)=0.39), respectively). This account analyzes the hitherto unknown molecular nature of these mixtures by a detailed experimental (multinuclear NMR and Raman spectroscopies) and theoretical study (BP86/TZVP DFT calculations, including COSMO solvation energies). The addition of AlCl(3) to the two liquids first leads to complexation with [Tf(2)N](-) and then disproportionation of the initial [AlCl(x)(Tf(2)N)(y)](-) complexes give Al(Tf(2)N)(3) and [AlCl(4)](-). At high concentrations of AlCl(3), the lower phase consists almost completely of Al(Tf(2)N)(3), whereas in the upper phase [AlCl(4)](-) is the dominant species. Electrodeposition of aluminium in the upper phase occurs from mixed AlCl(x)(Tf(2)N)(y) species, most likely from [AlCl(2)(Tf(2)N)(2)](-) formed in small concentrations at the phase boundary between the [AlCl(4)](-) and the Al(Tf(2)N)(3) layers. All the findings are supported by DFT calculations as well as an X-ray crystal structure determination of Al(Tf(2)N)(3). The latter was separated from the mixture by sublimation on a preparative scale. It was independently prepared from AlEt(3) and HNTf(2) and fully characterized. Moreover, the ionic liquids [BMP]AlCl(4) (m.p. 74 degrees C) and [EMIm]AlCl(4) (m.p. -7 degrees C), which mainly form the upper layer in the biphasic regime, were independently prepared and also fully characterized.