We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Estimation of the Surface Free Energy Components for Solid Surfaces: A Machine Learning Approach.
- Authors
Aydin, Ebubekir Siddik; Korkut, Ibrahim; Ozbay, Salih
- Abstract
Components of surface free energy (SFE) are critical to reveal the interactions between liquids and solids, and many interface phenomena can be interpreted by using these components. However, calculation of SFE components precisely with semiempirical approximations requires special expertise in surface science field. In this study, various machine learning algorithms (SVM, GPR) were utilized to construct models for the prediction of SFE from published data sets including contact angle values of water, formamide and diiodomethane. While creating the model data set, care was taken to choose different solid surface types. To objectively assess the validity of the constructed models, the tenfold cross-validation method was used for Case 1 and Case 2, respectively. For performance evaluations, R2, adjusted R2, RMSE and N-RMSE values of each model were calculated. The SVM and GPR models performed extremely well in the calculation of SFE components based on acid–base approach with regression coefficient values of about 0.99. In addition, estimation of the SFE components from experimental data that were not used in model training was made with the created MATLAB GUI interface. The γ S LW and γ S Tot parameters were predicted more accurately than the γ S + , γ S - and γ S AB components. The Case 1 model was more successful, especially in surface energy calculations of polymer-based surfaces. Finally, the proposed approach was evaluated by considering its advantages and disadvantages.
- Subjects
SURFACE energy; FREE surfaces; CONTACT angle; SOLIDS; KRIGING; MACHINE learning
- Publication
Arabian Journal for Science & Engineering (Springer Science & Business Media B.V. ), 2024, Vol 49, Issue 6, p7863
- ISSN
2193-567X
- Publication type
Article
- DOI
10.1007/s13369-023-08502-4