Machine Learning Predictions of Simulated Self-Diffusion Coefficients for Bulk and Confined Pure Liquids.

Diffusion properties of bulk fluids have been predicted using empirical expressions and machine learning (ML) models, suggesting that predictions of diffusion also should be possible for fluids in confined environments. The ability to quickly and accurately predict diffusion in porous materials would enable new discoveries and spur development in relevant technologies such as separations, catalysis, batteries, and subsurface applications. In this work, we apply artificial neural network (ANN) models to predict the simulated self-diffusion coefficients of real liquids in both bulk and pore environments. The training data sets were generated from molecular dynamics (MD) simulations of Lennard-Jones particles representing a diverse set of 14 molecules ranging from ammonia to dodecane over a range of liquid pressures and temperatures. Planar, cylindrical, and hexagonal pore models consisted of walls composed of carbon atoms. Our simple model for these liquids was primarily used to generate ANN training data, but the simulated self-diffusion coefficients of bulk liquids show excellent agreement with experimental diffusion coefficients. ANN models based on simple descriptors accurately reproduced the MD diffusion data for both bulk and confined liquids, including the trend of increased mobility in large pores relative to the corresponding bulk liquid.

Phospholipid headgroups govern area per lipid and emergent elastic properties of bilayers.

Phospholipid bilayers are liquid-crystalline materials whose intermolecular interactions at mesoscopic length scales have key roles in the emergence of membrane physical properties. Here we investigated the combined effects of phospholipid polar headgroups and acyl chains on biophysical functions of membranes with solid-state 2H NMR spectroscopy. We compared the structural and dynamic properties of phosphatidylethanolamine and phosphatidylcholine with perdeuterated acyl chains in the solid-ordered (so) and liquid-disordered (ld) phases. Our analysis of spectral lineshapes of 1,2-diperdeuteriopalmitoyl-sn-glycero-3-phosphoethanolamine (DPPE-d62) and 1,2-diperdeuteriopalmitoyl-sn-glycero-3-phosphocholine (DPPC-d62) in the so (gel) phase indicated an all-trans rotating chain structure for both lipids. Greater segmental order parameters (SCD) were observed in the ld (liquid-crystalline) phase for DPPE-d62 than for DPPC-d62 membranes, while their mixtures had intermediate values irrespective of the deuterated lipid type. Our results suggest the SCD profiles of the acyl chains are governed by methylation of the headgroups and are averaged over the entire system. Variations in the acyl chain molecular dynamics were further investigated by spin-lattice (R1Z) and quadrupolar-order relaxation (R1Q) measurements. The two acyl-perdeuterated lipids showed distinct differences in relaxation behavior as a function of the order parameter. The R1Z rates had a square-law dependence on SCD, implying collective mesoscopic dynamics, with a higher bending rigidity for DPPE-d62 than for DPPC-d62 lipids. Remodeling of lipid average and dynamic properties by methylation of the headgroups thus provides a mechanism to control the actions of peptides and proteins in biomembranes.

Symbolic regression development of empirical equations for diffusion in Lennard-Jones fluids.

Symbolic regression (SR) with a multi-gene genetic program has been used to elucidate new empirical equations describing diffusion in Lennard-Jones (LJ) fluids. Examples include equations to predict self-diffusion in pure LJ fluids and equations describing the finite-size correction for self-diffusion in binary LJ fluids. The performance of the SR-obtained equations was compared to that of both the existing empirical equations in the literature and to the results from artificial neural net (ANN) models recently reported. It is found that the SR equations have improved predictive performance in comparison to the existing empirical equations, even though employing a smaller number of adjustable parameters, but show an overall reduced performance in comparison to more extensive ANNs.

Prediction of Self-Diffusion in Binary Fluid Mixtures Using Artificial Neural Networks.

Artificial neural networks (ANNs) were developed to accurately predict the self-diffusion constants for individual components in binary fluid mixtures. The ANNs were tested on an experimental database of 4328 self-diffusion constants from 131 mixtures containing 75 unique compounds. The presence of strong hydrogen bonding molecules may lead to clustering or dimerization resulting in non-linear diffusive behavior. To address this, self- and binary association energies were calculated for each molecule and mixture to provide information on intermolecular interaction strength and were used as input features to the ANN. An accurate, generalized ANN model was developed with an overall average absolute deviation of 4.1%. Forward input feature selection reveals the importance of critical properties and self-association energies along with other fluid properties. Additional ANNs were developed with subsets of the full input feature set to further investigate the impact of various properties on model performance. The results from two specific mixtures are discussed in additional detail: one providing an example of strong hydrogen bonding and the other an example of extreme pressure changes, with the ANN models predicting self-diffusion well in both cases.

Using Computationally-Determined Properties for Machine Learning Prediction of Self-Diffusion Coefficients in Pure Liquids.

The ability to predict transport properties of liquids quickly and accurately will greatly improve our understanding of fluid properties both in bulk and complex mixtures, as well as in confined environments. Such information could then be used in the design of materials and processes for applications ranging from energy production and storage to manufacturing processes. As a first step, we consider the use of machine learning (ML) methods to predict the diffusion properties of pure liquids. Recent results have shown that Artificial Neural Networks (ANNs) can effectively predict the diffusion of pure compounds based on the use of experimental properties as the model inputs. In the current study, a similar ANN approach is applied to modeling diffusion of pure liquids using fluid properties obtained exclusively from molecular simulations. A diverse set of 102 pure liquids is considered, ranging from small polar molecules (e.g., water) to large nonpolar molecules (e.g., octane). Self-diffusion coefficients were obtained from classical molecular dynamics (MD) simulations. Since nearly all the molecules are organic compounds, a general set of force field parameters for organic molecules was used. The MD methods are validated by comparing physical and thermodynamic properties with experiment. Computational input features for the ANN include physical properties obtained from the MD simulations as well as molecular properties from quantum calculations of individual molecules. Fluid properties describing the local liquid structure were obtained from center of mass radial distribution functions (COM-RDFs). Feature sensitivity analysis revealed that isothermal compressibility, heat of vaporization, and the thermal expansion coefficient were the most impactful properties used as input for the ANN model to predict the MD simulated self-diffusion coefficients. The MD-based ANN successfully predicts the MD self-diffusion coefficients with only a subset (2 to 3) of the available computationally determined input features required. A separate ANN model was developed using literature experimental self-diffusion coefficients as model targets. Although this second ML model was not as successful due to a limited number of data points, a good correlation is still observed between experimental and ML predicted self-diffusion coefficients.

NMR spectroscopy of coin cell batteries with metal casings.

Battery cells with metal casings are commonly considered incompatible with nuclear magnetic resonance (NMR) spectroscopy because the oscillating radio-frequency magnetic fields ("rf fields") responsible for excitation and detection of NMR active nuclei do not penetrate metals. Here, we show that rf fields can still efficiently penetrate nonmetallic layers of coin cells with metal casings provided "B1 damming" configurations are avoided. With this understanding, we demonstrate noninvasive high-field in situ 7Li and 19F NMR of coin cells with metal casings using a traditional external NMR coil. This includes the first NMR measurements of an unmodified commercial off-the-shelf rechargeable battery in operando, from which we detect, resolve, and separate 7Li NMR signals from elemental Li, anodic ß-LiAl, and cathodic LixMnO2 compounds. Real-time changes of ß-LiAl lithium diffusion rates and variable ß-LiAl 7Li NMR Knight shifts are observed and tied to electrochemically driven changes of the ß-LiAl defect structure.

Modes of Disorder in Poly(carbon monofluoride).

Poly(carbon monofluoride), or (CF)n, is a layered fluorinated graphite material consisting of nanosized platelets. Here, we present experimental multidimensional solid-state NMR spectra of (CF)n, supported by density functional theory (DFT) calculations of NMR parameters, which overhauls our understanding of structure and bonding in the material by elucidating many ways in which disorder manifests. We observe strong 19F NMR signals conventionally assigned to elongated or "semi-ionic" C-F bonds and find that these signals are in fact due to domains where the framework locally adopts boat-like cyclohexane conformations. We calculate that C-F bonds are weakened but are not elongated by this conformational disorder. Exchange NMR suggests that conformational disorder avoids platelet edges. We also use a new J-resolved NMR method for disordered solids, which provides molecular-level resolution of highly fluorinated edge states. The strings of consecutive difluoromethylene groups at edges are relatively mobile. Topologically distinct edge features, including zigzag edges, crenellated edges, and coves, are resolved in our samples by solid-state NMR. Disorder should be controllable in a manner dependent on synthesis, affording new opportunities for tuning the properties of graphite fluorides.

Artificial neural network prediction of self-diffusion in pure compounds over multiple phase regimes.

Artificial neural networks (ANNs) were developed to accurately predict the self-diffusion constants for pure components in liquid, gas and super critical phases. The ANNs were tested on an experimental database of 6625 self-diffusion constants for 118 different chemical compounds. The presence of multiple phases results in a heavy skew in the distribution of diffusion constants and multiple approaches were used to address this challenge. First, an ANN was developed with the raw diffusion values to assess what the main drawbacks of this direct method were. The first approach for improving the predictions involved taking the log 10 of diffusion to provide a more uniform distribution and reduce the range of target output values used to develop the ANN. The second approach involved developing individual ANNs for each phase using the raw diffusion values. Results show that the log transformation leads to a model with the best self-diffusion constant predictions and an overall average absolute deviation (AAD) of 6.56%. The resultant ANN is a generalized model that can be used to predict diffusion across all three phases and over a diverse group of compounds. The importance of each input feature was ranked using a feature addition method revealing that the density of the compound has the largest impact on the ANN prediction of self-diffusion constants in pure compounds.

Machine Learning-Based Upscaling of Finite-Size Molecular Dynamics Diffusion Simulations for Binary Fluids.

Molecular diffusion coefficients calculated using molecular dynamics (MD) simulations suffer from finite-size (i.e., finite box size and finite particle number) effects. Results from finite-sized MD simulations can be upscaled to infinite simulation size by applying a correction factor. For self-diffusion of single-component fluids, this correction has been well-studied by many researchers including Yeh and Hummer (YH); for binary fluid mixtures, a modified YH correction was recently proposed for correcting MD-predicted Maxwell-Stephan (MS) diffusion rates. Here we use both empirical and machine learning methods to identify improvements to the finite-size correction factors for both self-diffusion and MS diffusion of binary Lennard-Jones (LJ) fluid mixtures. Using artificial neural networks (ANNs), the error in the corrected LJ fluid diffusion is reduced by an order of magnitude versus existing YH corrections, and the ANN models perform well for mixtures with large dissimilarities in size and interaction energies where the YH correction proves insufficient.

Formation of monomeric Sn(ii) and Sn(iv) perfluoropinacolate complexes and their characterization by 119Sn Mössbauer and 119Sn NMR spectroscopies.

The synthesis and characterization of a series of Sn(ii) and Sn(iv) complexes supported by the highly electron-withdrawing dianionic perfluoropinacolate (pinF) ligand are reported herein. Three analogs of [SnIV(pinF)3]2- with NEt3H+ (1), K+ (2), and {K(18C6)}+ (3) counter cations and two analogs of [SnII(pinF)2]2- with K+ (4) and {K(15C5)2}+ (5) counter cations were prepared and characterized by standard analytical methods, single-crystal X-ray diffraction, and 119Sn Mössbauer and NMR spectroscopies. The six-coordinate SnIV(pinF) complexes display 119Sn NMR resonances and 119Sn Mössbauer spectra similar to SnO2 (cassiterite). In contrast, the four-coordinate SnII(pinF) complexes, featuring a stereochemically-active lone pair, possess low 119Sn NMR chemical shifts and relatively high quadrupolar splitting. Furthermore, the Sn(ii) complexes are unreactive towards both Lewis bases (pyridine, NEt3) and acids (BX3, Et3NH+). Calculations confirm that the Sn(ii) lone pair is localized within the 5s orbital and reveal that the Sn 5px LUMO is energetically inaccessible, which effectively abates reactivity.

Arene Substitution Design for Controlled Conformational Changes of Dibenzocycloocta-1,5-dienes.

We report that an agile eight-membered cycloalkane can be stabilized by fusing a benzene ring on each side, substituted with proper functional groups. The conformational change of dibenzocycloocta-1,5-diene (DBCOD), a rigid-flexible-rigid organic moiety, from its Boat to Chair conformation requires an activation energy of 42 kJ/mol, which is substantially lower than those of existing submolecular shape-changing units. Experimental data corroborated by theoretical calculations demonstrate that intramolecular hydrogen bonding can stabilize Boat, whereas electron repulsive interaction from opposing ester substituents favors Chair. Intramolecular hydrogen bonding formed by 1,10-diamide substitution stabilizes Boat, spiking the temperature at which Boat and Chair can readily interchange from -60 to 60 °C. Concomitantly this intramolecular attraction raises the energy barrier from 42 kJ/mol for unsubstituted DBCOD to 68 kJ/mol for diamide-substituted DBCOD. Remarkably, this value falls within the range of the activation energy of highly efficient enzyme-catalyzed biological reactions. With shape changes once considered only possible with high energy, our work reveals a potential pathway exemplified by a specific submolecular structure to achieve low-energy-driven shape changes for the first time. The intrinsic cycle stability and high-energy output systems that would incur damage under high-energy stimuli could particularly benefit from this new kind of low-energy-driven shape-changing mechanism. This work has laid the basis to construct systems for low-energy-driven stimuli-responsive applications, hitherto a challenge to overcome.

Heterogeneous Polymer Dynamics Explored Using Static 1H NMR Spectra.

NMR spectroscopy continues to provide important molecular level details of dynamics in different polymer materials, ranging from rubbers to highly crosslinked composites. It has been argued that thermoset polymers containing dynamic and chemical heterogeneities can be fully cured at temperatures well below the final glass transition temperature (Tg). In this paper, we described the use of static solid-state 1H NMR spectroscopy to measure the activation of different chain dynamics as a function of temperature. Near Tg, increasing polymer segmental chain fluctuations lead to dynamic averaging of the local homonuclear proton-proton (1H-1H) dipolar couplings, as reflected in the reduction of the NMR line shape second moment (M2) when motions are faster than the magnitude of the dipolar coupling. In general, for polymer systems, distributions in the dynamic correlation times are commonly expected. To help identify the limitations and pitfalls of M2 analyses, the impact of activation energy or, equivalently, correlation time distributions, on the analysis of 1H NMR M2 temperature variations is explored. It is shown by using normalized reference curves that the distributions in dynamic activation energies can be measured from the M2 temperature behavior. An example of the M2 analysis for a series of thermosetting polymers with systematically varied dynamic heterogeneity is presented and discussed.

Machine learning prediction of self-diffusion in Lennard-Jones fluids.

Different machine learning (ML) methods were explored for the prediction of self-diffusion in Lennard-Jones (LJ) fluids. Using a database of diffusion constants obtained from the molecular dynamics simulation literature, multiple Random Forest (RF) and Artificial Neural Net (ANN) regression models were developed and characterized. The role and improved performance of feature engineering coupled to the RF model development was also addressed. The performance of these different ML models was evaluated by comparing the prediction error to an existing empirical relationship used to describe LJ fluid diffusion. It was found that the ANN regression models provided superior prediction of diffusion in comparison to the existing empirical relationships.

Quantification of Uncoupled Spin Domains in Spin-Abundant Disordered Solids.

Materials often contain minor heterogeneous phases that are difficult to characterize yet nonetheless significantly influence important properties. Here we describe a solid-state NMR strategy for quantifying minor heterogenous sample regions containing dilute, essentially uncoupled nuclei in materials where the remaining nuclei experience heteronuclear dipolar couplings. NMR signals from the coupled nuclei are dephased while NMR signals from the uncoupled nuclei can be amplified by one or two orders of magnitude using Carr-Meiboom-Purcell-Gill (CPMG) acquisition. The signal amplification by CPMG can be estimated allowing the concentration of the uncoupled spin regions to be determined even when direct observation of the uncoupled spin NMR signal in a single pulse experiment would require an impractically long duration of signal averaging. We use this method to quantify residual graphitic carbon using 13C CPMG NMR in poly(carbon monofluoride) samples synthesized by direct fluorination of carbon from various sources. Our detection limit for graphitic carbon in these materials is better than 0.05 mol%. The accuracy of the method is discussed and comparisons to other methods are drawn.

Computational and Experimental 1H-NMR Study of Hydrated Mg-Based Minerals.

Magnesium oxide (MgO) can convert to different magnesium-containing compounds depending on exposure and environmental conditions. Many MgO-based phases contain hydrated species allowing 1H-nuclear magnetic resonance (NMR) spectroscopy to be used in the characterization and quantification of proton-containing phases; however, surprisingly limited examples have been reported. Here, 1H-magic angle spinning (MAS) NMR spectra of select Mg-based minerals are presented and assigned. These experimental results are combined with computational NMR density functional theory (DFT) periodic calculations to calibrate the predicted chemical shielding results. This correlation is then used to predict the NMR shielding for a series of different MgO hydroxide, magnesium chloride hydrate, magnesium perchlorate, and magnesium cement compounds to aid in the future assignment of 1H-NMR spectra for complex Mg phases.

Trapped Intermediate of a Meerwein-Pondorf-Verley Reduction of Hydroxy Benzaldehyde to a Dialkoxide by Titanium Alkoxides.

A series of titanium alkoxides ([Ti(OR)4] (OR = OCH(CH3)2 (OPri), OC(CH3)3 (OBut), and OCH2C(CH3)3 (ONep)) were modified with a set of substituted hydroxyl-benzaldehydes [HO-BzA-Lx: x = 1, 2-hydroxybenzaldehyde (L = H), 2-hydroxy-3-methoxybenzaldehyde (OMe-3), 5-bromo-2-hydroxybenzaldehyde (Br-5), 2-hydroxy-5-nitrobenzaldehyde (NO2-5); x = 2, 3,5-di-tert-butyl-2-hydroxybenzaldehyde (But-3,5), 2-hydroxy-3,5-diiodobenzaldehyde (I-3,5)] in pyridine (py). Instead of the expected simple substitution, each of the HO-BzA-Lx modifiers were reduced to their respective diol [(py)(OR)2Ti(κ2(O,µ-O')(OC6H4-x(CH2O)-2)(L)x] (OR = OPri, x = 1, L = H (1a), OMe-3 (2a), Br-5 (3a·py), NO2-5 (4a·4py); x = 2, But-3,5 (5a), I-3,5 (6a), ONep; x = 1, L = H (1b), OMe-3 (2b), Br-5 (3b·py), NO2-5 (4b); x = 2, But-3,5 (5b), I-3,5 (6b·py)), as identified by single crystal X-ray studies. The 1H NMR spectral data were complex at room temperature but simplified at high temperatures (70 °C). Diffusion ordered spectroscopy (DOSY) NMR experiments indicated that 2a maintained the dinuclear structure in a solution independent of the temperature, whereas 2b appears to be monomeric over the same temperature range. On the basis of additional NMR studies, the mechanism of the reduction of the HO-BzA-Lx to the dioxide ligand was thought to occur by a Meerwein-Pondorf-Verley (MPV) mechanism. The structures of 1a-6b appear to be the intermediate dioxide products of the MPV reduction, which became "trapped" by the Lewis basic solvate.

Hydration and Hydroxylation of MgO in Solution: NMR Identification of Proton-Containing Intermediate Phases.

Magnesium oxide (MgO)-engineered barriers used in subsurface applications will be exposed to high concentration brine environments and may form stable intermediate phases that can alter the effectiveness of the barrier. To explore the formation of these secondary intermediate phases, MgO was aged in water and three different brine solutions and characterized with X-ray diffraction (XRD) and 1H magic angle spinning (MAS) nuclear magnetic resonance (NMR) spectroscopy. After aging, there is ∼4% molar equivalent of a hydrogen-containing species formed. The 1H MAS NMR spectra resolved multiple minor phases not visible in XRD, indicating that diverse disordered proton-containing environments are present in addition to crystalline Mg(OH)2 brucite. Density functional theory (DFT) simulations for the proposed Mg-O-H-, Mg-Cl-O-H-, and Na-O-H-containing phases were performed to index resonances observed in the experimental 1H MAS NMR spectra. Although the intermediate crystal structures exhibited overlapping 1H NMR resonances in the spectra, Mg-O-H intermediates were attributed to the growth of resonances in the δ +1.0 to 0.0 ppm region, and Mg-Cl-O-H structures produced the increasing contributions of the δ = +2.5 to 5.0 ppm resonances in the chloride-containing brines. Overall, 1H NMR analysis of aged MgO indicates the formation of a wide range of possible intermediate structures that cannot be observed or resolved in the XRD analysis.

Resolving Confined 7Li Dynamics of Uranyl Peroxide Capsule U24.

We obtained a kerosene-soluble form of the lithium salt [UO2(O2)(OH)2]24 phase (Li-U24), by adding cetyltrimethylammonium bromide surfactant to aqueous Li-U24. Interestingly, its variable-temperature solution 7Li NMR spectroscopy resolves two narrowly spaced resonances down to -10 °C, which shift upfield with increasing temperature, and finally coalesce at temperatures > 85 °C. Comparison with solid-state NMR demonstrates that the Li dynamics in the Li-U24-CTA phase involves only exchange between different local encapsulated environments. This behavior is distinct from the rapid Li exchange dynamics observed between encapsulated and external Li environments for Li-U24 in both the aqueous and the solid-state phases. Density functional theory calculations suggest that the two experimental 7Li NMR chemical shifts are due to Li cations coordinated within the square and hexagonal faces of the U24 cage, and they can undergo exchange within the confined environment, as the solution is heated. Very different than U24 in aqueous media, there is no evidence that the Li cations exit the cage, and therefore, this represents a truly confined space.

Computational Study of Microhydration in Sulfonated Diels-Alder Poly(phenylene) Polymers.

The nature of microhydration in sulfonated Diels-Alder poly(phenylene) (SDAPP) polymer membranes is explored using ab initio and density functional theory (DFT) electronic structure calculations. The impact of the aromatic poly(phenylene) structure, including cooperative effects between multiple spatially adjacent sulfonic groups, on the hydration environment is addressed using a series of DFT B3LYP/6-311**-optimized structures for different SDAPP· nH2O clusters. In addition, larger SDAPP polymer fragments, along with selected hydrophilic domain structures extracted from molecular dynamic (MD) simulations, are also evaluated using ONIOM HF/PM6 semiempirical calculations. The SDAPP clusters reveal that spontaneous proton dissociation occurs at low levels of hydration to form sulfonic-acid-associated H3O+ contact ion pairs (CIPs), which then evolve into solvated CIPs at higher hydration levels. For multiple sulfonic acid groups located on the poly(phenylene) side chains, the hydration energies are a function of the relative acid location and backbone configuration. Variations in the phenylene backbone torsional angles allow remote sulfonic acids to adopt an optimal separation to produce an extended hydrogen bonded network of waters between the SDAPP acids groups. These calculations provide a baseline to help describe the proton transport and hydration behavior of SDAPP membranes.

Synthesis and characterization of thallium-salen derivatives for use as underground fluid flow tracers.

A pair of thallium salen derivatives was synthesized and characterized for potential use as monitors (or taggants) or as models for Group 13 complexes for subterranean fluid flows. These precursors were isolated from the reaction of thallium ethoxide with N,N'-bis(3,5-di-tert-butylsalicylidene)-ethylenediamine (H2-salo-But), or N,N'-bis(3,5-di-tert-butylsalicylidene)-1,2-phenylenediamine (H2-saloPh-But). The products were identified by single crystal X-ray diffraction as: [((µ-O)2,κ1-(N)(N')salo-But)Tl2] (1) and {[((µ-O)2saloPh-But)Tl2][((µ-O)2,κ1-(N)(N')saloPh-But)Tl2]} (2). Both structures are similar, wherein each O atom of the salo moiety bridges the two Tl atoms, leading to a TlTl interaction, which is further stabilized by an intramolecular π-bond with neighboring phenyl rings. For 1, an additional TlN interaction was solved for each metal center; whereas, for 2, one of the two molecules in the matrix has a weak TlN interaction but no bonding noted in the other molecule. Both Density Functional Theory (DFT) calculations and variable temperature solution 205Tl NMR studies of 1 and 2 further confirmed the TlTl interaction. The UV-vis absorbance spectra of these compounds had an absorbance peak at 392 nm for 1 and a broad absorbance peak centered at 469 nm for 2, which were found to be in good agreement with the DFT calculated spectra that were dominated by the singlet state. Fluorescence emission and excitation studies reveal absorptions at 360 and 380 nm for 1 and 2, respectively, which are attributed to the TlTl metal centers. To demonstrate practicality, fluorescence spectra of 1 and 2 were obtained using a handheld 405 nm cw laser pointer and portable spectrometer where compound 1 was found to glow 15 times brighter than compound 2. Only compound 1 was found to survive the simulated deep-well conditions explored, which was attributed to the TlN interaction noted for 1 but not for 2.