An artifical neural network (ANN) is trained on the Binary Polymer Solution Cloud Points Dataset.1,2 All model details, data preprocessing, and training protocols can be found in Ref. 1 and 2. The ANN model outputs estimated cloud point temperatures as a function of volume fraction for select polymer-solvent systems.
The model is trained on 21 linear polymer chemistries and 97 unique polymer-solvent systems. Therefore, extrapolation to new polymers will likely be poor. However, the model will be updated as additional data is processed, reducing error for new linear polymers.1,2
The code generates a binodal curve for a binary mixture of a linear polymer in solvent. Required inputs are: polymer repeat unit SMILES, solvent SMILES, polymer Mw, volume fraction range, polydispersity, and pressure. Number of volume fractions is not required (default is 100) and determines the number of predicted values between φmin and φmax (larger numbers result in smoother curves). Predictions will be made using the artifical neural network (ANN) model.
To generate a SMILES notation for the repeat unit and solvent, several tools are available such as the BigSMILES converter on the CRIPT website, or search for a specific polymer-solvent system in the Binary Solution Cloud Points Dataset.
UCS/LCS - predict the upper or lower critical solubility cloud points (miscible region found with increasing or decreasing temperature, respectively). Default is UCS only.
Find Nearest Experimental Data (by Mw) - search Binary Solution Cloud Points Dataset for experimental data closest to input Mw value.