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.
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.
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.
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.
A series of networks is introduced with systematically varied network heterogeneity and high overall values of average glass transition temperature (Tg), based on polymerization of rigid acrylate and aromatic thiol monomers. The curing behavior, chain dynamics, and microstructure of these networks were investigated through a combination of dynamic mechanical analysis and infrared spectroscopy, nuclear magnetic resonance (NMR) spectroscopy, and x-ray scattering, respectively. The maximum Tg achieved during cure can be related to the breadth of the mechanical loss tangent, as others have previously suggested, as well as the temperature dependence of the chain dynamics in the network as monitored by 1H NMR. In addition, the microstructures of the networks are characterized by periodic, fractal microgels with characteristic length scales of ca. 20–40 nm. Intriguingly, this structural motif persists in the more homogeneous networks exhibiting comparatively narrow glass transitions and chain dynamics, indicating that dynamically homogeneous networks can still exhibit significant compositional heterogeneity at the mesoscale.
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.
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.