A data driven machine learning approach for predicting and optimizing sulfur compound adsorption on metal organic frameworks
Abstract This study employed some machine learning (ML) techniques with Python programming to forecast the adsorption capacity of MOF adsorbents for thiophenic compounds namely benzothiophene (BT), dibenzothiophene (DBT), and 4,6-dimethyl dibenzothiophene (4,6-DMDBT). Five ML models were developed w...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-01-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-86689-2 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832585794243526656 |
|---|---|
| author | Mohsen Shayanmehr Sepehr Aarabi Ahad Ghaemi Alireza Hemmati |
| author_facet | Mohsen Shayanmehr Sepehr Aarabi Ahad Ghaemi Alireza Hemmati |
| author_sort | Mohsen Shayanmehr |
| collection | DOAJ |
| description | Abstract This study employed some machine learning (ML) techniques with Python programming to forecast the adsorption capacity of MOF adsorbents for thiophenic compounds namely benzothiophene (BT), dibenzothiophene (DBT), and 4,6-dimethyl dibenzothiophene (4,6-DMDBT). Five ML models were developed with the help of a dataset containing 676 rows to correlate the adsorbent features, adsorption conditions, and adsorbate characteristics to the MOF sample’s sulfur adsorption capability. Among the ML approaches, MLP model achieved the best performance with a low mean squared error (MSE) of 0.0032 on the test set and 0.0021 on the training set and mean relative error (MRE) of 15.26% on the test set. Also, Random Forest model yielded a higher test MSE of 0.0045 and MRE of 17.83%. Feature importance analysis was performed by utilizing MLP model and shapely additive plan (SHAP) method, and the findings revealed that “initial concentration of sulfur” (SHAP value 0.51) and “contact time” (SHAP value 0.37) were the crucial factors influenced desulfurization process efficiency. Additionally, a comparative analysis of the features utilizing the MLP network classified the factors into three primary categories: process conditions, adsorbent characteristics, and adsorbate characteristics. Consequently, the process condition was identified as the most significant group compared to others. Finally, the desulfurization process optimization indicated the maximum DBT adsorption of 161.6 mg/g for Zr-based MOF could be achieved when the features including BET, TPV, pore size, oil/adsorbent ration, and temperature were tuned around 756 m2/g, 0.955 cm3/g, 5.96 nm, 449.85 g/g, 20.1 °C, respectively. |
| format | Article |
| id | doaj-art-9ee9f2eba91b4386865b0390df52368a |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-9ee9f2eba91b4386865b0390df52368a2025-01-26T12:31:09ZengNature PortfolioScientific Reports2045-23222025-01-0115112210.1038/s41598-025-86689-2A data driven machine learning approach for predicting and optimizing sulfur compound adsorption on metal organic frameworksMohsen Shayanmehr0Sepehr Aarabi1Ahad Ghaemi2Alireza Hemmati3School of Chemical, Petroleum and Gas Engineering, Iran University of Science and TechnologySchool of Chemical, Petroleum and Gas Engineering, Iran University of Science and TechnologySchool of Chemical, Petroleum and Gas Engineering, Iran University of Science and TechnologySchool of Chemical, Petroleum and Gas Engineering, Iran University of Science and TechnologyAbstract This study employed some machine learning (ML) techniques with Python programming to forecast the adsorption capacity of MOF adsorbents for thiophenic compounds namely benzothiophene (BT), dibenzothiophene (DBT), and 4,6-dimethyl dibenzothiophene (4,6-DMDBT). Five ML models were developed with the help of a dataset containing 676 rows to correlate the adsorbent features, adsorption conditions, and adsorbate characteristics to the MOF sample’s sulfur adsorption capability. Among the ML approaches, MLP model achieved the best performance with a low mean squared error (MSE) of 0.0032 on the test set and 0.0021 on the training set and mean relative error (MRE) of 15.26% on the test set. Also, Random Forest model yielded a higher test MSE of 0.0045 and MRE of 17.83%. Feature importance analysis was performed by utilizing MLP model and shapely additive plan (SHAP) method, and the findings revealed that “initial concentration of sulfur” (SHAP value 0.51) and “contact time” (SHAP value 0.37) were the crucial factors influenced desulfurization process efficiency. Additionally, a comparative analysis of the features utilizing the MLP network classified the factors into three primary categories: process conditions, adsorbent characteristics, and adsorbate characteristics. Consequently, the process condition was identified as the most significant group compared to others. Finally, the desulfurization process optimization indicated the maximum DBT adsorption of 161.6 mg/g for Zr-based MOF could be achieved when the features including BET, TPV, pore size, oil/adsorbent ration, and temperature were tuned around 756 m2/g, 0.955 cm3/g, 5.96 nm, 449.85 g/g, 20.1 °C, respectively.https://doi.org/10.1038/s41598-025-86689-2Adsorptive desulfurizationThiophenic compounds removalMOFs adsorbentsMachine learning |
| spellingShingle | Mohsen Shayanmehr Sepehr Aarabi Ahad Ghaemi Alireza Hemmati A data driven machine learning approach for predicting and optimizing sulfur compound adsorption on metal organic frameworks Scientific Reports Adsorptive desulfurization Thiophenic compounds removal MOFs adsorbents Machine learning |
| title | A data driven machine learning approach for predicting and optimizing sulfur compound adsorption on metal organic frameworks |
| title_full | A data driven machine learning approach for predicting and optimizing sulfur compound adsorption on metal organic frameworks |
| title_fullStr | A data driven machine learning approach for predicting and optimizing sulfur compound adsorption on metal organic frameworks |
| title_full_unstemmed | A data driven machine learning approach for predicting and optimizing sulfur compound adsorption on metal organic frameworks |
| title_short | A data driven machine learning approach for predicting and optimizing sulfur compound adsorption on metal organic frameworks |
| title_sort | data driven machine learning approach for predicting and optimizing sulfur compound adsorption on metal organic frameworks |
| topic | Adsorptive desulfurization Thiophenic compounds removal MOFs adsorbents Machine learning |
| url | https://doi.org/10.1038/s41598-025-86689-2 |
| work_keys_str_mv | AT mohsenshayanmehr adatadrivenmachinelearningapproachforpredictingandoptimizingsulfurcompoundadsorptiononmetalorganicframeworks AT sepehraarabi adatadrivenmachinelearningapproachforpredictingandoptimizingsulfurcompoundadsorptiononmetalorganicframeworks AT ahadghaemi adatadrivenmachinelearningapproachforpredictingandoptimizingsulfurcompoundadsorptiononmetalorganicframeworks AT alirezahemmati adatadrivenmachinelearningapproachforpredictingandoptimizingsulfurcompoundadsorptiononmetalorganicframeworks AT mohsenshayanmehr datadrivenmachinelearningapproachforpredictingandoptimizingsulfurcompoundadsorptiononmetalorganicframeworks AT sepehraarabi datadrivenmachinelearningapproachforpredictingandoptimizingsulfurcompoundadsorptiononmetalorganicframeworks AT ahadghaemi datadrivenmachinelearningapproachforpredictingandoptimizingsulfurcompoundadsorptiononmetalorganicframeworks AT alirezahemmati datadrivenmachinelearningapproachforpredictingandoptimizingsulfurcompoundadsorptiononmetalorganicframeworks |