Skip to main content
Dryad

Predicted room temperature electrical conductivity of molecular mixtures

Cite this dataset

Zhang, Hengrui et al. (2024). Predicted room temperature electrical conductivity of molecular mixtures [Dataset]. Dryad. https://doi.org/10.5061/dryad.pnvx0k6w1

Abstract

In the associated manuscript, we propose the MolSets machine learning model for molecular mixture properties. Using the MolSets architecture, we train a model on a dataset curated by Bradford et al. (2023) to predict the room temperature (298 K) electrical conductivity of mixtures. Here, we report the model-predicted conductivities of all equal-weight binary mixtures among 28 types of small molecules, combined with 30 types of Li+ salts (1 mol·kg-1), totaling 11,340 candidate lithium battery electrolytes. Note that the current model has a limitation of not taking salt solubility into account. This dataset is for demonstration purposes and should be used with caution.

README: Predicted room temperature electrical conductivity of molecular mixtures

https://doi.org/10.5061/dryad.pnvx0k6w1

This dataset reports the machine learning-predicted room temperature (298 K) electrical conductivity for 11,340 molecular mixtures.

Description of the data and file structure

Every mixture consists of two types of molecules and one salt. The data is organized in a spreadsheet format. The columns are organized as follows:

  • The first two columns, "Solvent 1/2" contain the SMILES strings of constituent molecules in the mixture.
  • "Weights" and "Molar Mass" columns contain the weight fractions and molar masses (g/mol) of constituent molecules, respectively.
  • The "Salt" column contains the SMILES string of the salts, whose amount (kg/mol) in the mixture is reported in the "Molality" column.
  • The last column contains the predicted logarithm conductivities, in the unit of log (S/cm).

Sharing/Access information

The data is generated by a machine learning predictive model. The model adopts the "MolSets" architecture, which is proposed in the preprint "MolSets: Molecular graph deep sets learning for mixture property modeling" (see Related Works: arXiv:2312.16473). An implementation of the model is available on GitHub (see Related Works). Data used to train the current model is acquired from the publication: Bradford et al., (2023) ACS Cent. Sci. 9(2): 206–216.

Funding

National Science Foundation, Award: 2324173, DMREF

National Science Foundation, Award: 2219489, DMR

Northwestern University, Award: Ryan Fellowship, IIN