Machine learning modeling of the CO2 solubility in ionic liquids by using σ-profile descriptors

Juho-Pekka Laakso; Ali Ebrahimpoor Gorji; Ville Alopaeus; Petri Uusi-Kyyny

doi:10.1016/j.ces.2025.121226

Machine learning modeling of the CO₂ solubility in ionic liquids by using σ-profile descriptors

Juho-Pekka Laakso^*
, Ali Ebrahimpoor Gorji
, Ville Alopaeus
, Petri Uusi-Kyyny

^*Tämän työn vastaava kirjoittaja

Tutkimustuotos: Lehtiartikkeli › Article › Scientific › vertaisarvioitu

22 Sitaatiot (Scopus)

226 Lataukset (Pure)

Abstrakti

The solubility of carbon dioxide (CO₂) in solvents is important for carbon capture and utilization technologies, with ionic liquids (ILs) being promising due to their ability to capture CO₂. Since the number of possible ILs is huge, predicting CO₂ solubility during solvent screening is essential. In this work, various machine learning (ML) models including, multiple linear regression, artificial neural network, and random forest, were developed by using 9864 data points covering 124 ILs and descriptors from the σ-profile for predicting CO₂ solubility in ILs. The random forest model produced the best performance (R² = 0.9754, MAE = 0.0257). We estimated the importance of the descriptors, highlighting that those with non-polar characteristics of the σ-profile are important. Lastly, we predicted CO₂ solubilities for 1444 unstudied ILs. The combination of ML with the σ-profile descriptors offers great generalizability for predicting CO₂ solubility in ILs. This enables IL screening for CO₂ related applications.

Alkuperäiskieli	Englanti
Artikkeli	121226
Sivumäärä	13
Julkaisu	Chemical Engineering Science
Vuosikerta	307
Varhainen verkossa julkaisun päivämäärä	24 helmik. 2025
DOI - pysyväislinkit	https://doi.org/10.1016/j.ces.2025.121226
Tila	Julkaistu - 15 maalisk. 2025
OKM-julkaisutyyppi	A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä

Rahoitus

This work was finically supported by the Academy of Finland ‘In-situ equilibrium shifting in CO2 utilization reactions by novel absorbents (CO2Shift)’ Project (351113). The authors wish to acknowledge CSC – IT Center for Science, Finland, for computational resources.

YK:n kestävän kehityksen tavoitteet

Tämä tuotos edistää seuraavia kestävän kehityksen tavoitteita:

SDG 13 – Ilmastotoimet

Pääsy asiakirjaan

10.1016/j.ces.2025.121226Lisenssi: CC BY

Machine_learning_modeling_of_the_CO2_solubilityFinal published version, 3,52 MBLisenssi: CC BY

Muut tiedostot ja linkit

Linkki julkaisuun Scopuksessa

CO2Shift, KELA: In-situ equilibrium shifting of C1 reactions by novel absorbents
Uusi-Kyyny, P. (Vastuullinen johtaja), Saad, M. (Projektin jäsen), Laakso, J.-P. (Projektin jäsen), Nguyen, H. (Projektin jäsen), Assadzadeh, B. (Projektin jäsen) & Noor, Z. (Projektin jäsen)
01/09/2022 → 31/08/2026
Projekti: RCF Academy Project

Siteeraa tätä

@article{16ea0a8adb994a0c9bd1cb34f58e739a,

title = "Machine learning modeling of the CO2 solubility in ionic liquids by using σ-profile descriptors",

abstract = "The solubility of carbon dioxide (CO2) in solvents is important for carbon capture and utilization technologies, with ionic liquids (ILs) being promising due to their ability to capture CO2. Since the number of possible ILs is huge, predicting CO2 solubility during solvent screening is essential. In this work, various machine learning (ML) models including, multiple linear regression, artificial neural network, and random forest, were developed by using 9864 data points covering 124 ILs and descriptors from the σ-profile for predicting CO2 solubility in ILs. The random forest model produced the best performance (R2 = 0.9754, MAE = 0.0257). We estimated the importance of the descriptors, highlighting that those with non-polar characteristics of the σ-profile are important. Lastly, we predicted CO2 solubilities for 1444 unstudied ILs. The combination of ML with the σ-profile descriptors offers great generalizability for predicting CO2 solubility in ILs. This enables IL screening for CO2 related applications.",

keywords = "Quantitative structure–property relationship, Machine learning, Ionic liquids, Carbon dioxide solubility prediction, σ-profile",

author = "Juho-Pekka Laakso and \{Ebrahimpoor Gorji\}, Ali and Ville Alopaeus and Petri Uusi-Kyyny",

year = "2025",

month = mar,

day = "15",

doi = "10.1016/j.ces.2025.121226",

language = "English",

volume = "307",

journal = "Chemical Engineering Science",

issn = "0009-2509",

publisher = "Elsevier",

}

TY - JOUR

T1 - Machine learning modeling of the CO2 solubility in ionic liquids by using σ-profile descriptors

AU - Laakso, Juho-Pekka

AU - Ebrahimpoor Gorji, Ali

AU - Alopaeus, Ville

AU - Uusi-Kyyny, Petri

PY - 2025/3/15

Y1 - 2025/3/15

N2 - The solubility of carbon dioxide (CO2) in solvents is important for carbon capture and utilization technologies, with ionic liquids (ILs) being promising due to their ability to capture CO2. Since the number of possible ILs is huge, predicting CO2 solubility during solvent screening is essential. In this work, various machine learning (ML) models including, multiple linear regression, artificial neural network, and random forest, were developed by using 9864 data points covering 124 ILs and descriptors from the σ-profile for predicting CO2 solubility in ILs. The random forest model produced the best performance (R2 = 0.9754, MAE = 0.0257). We estimated the importance of the descriptors, highlighting that those with non-polar characteristics of the σ-profile are important. Lastly, we predicted CO2 solubilities for 1444 unstudied ILs. The combination of ML with the σ-profile descriptors offers great generalizability for predicting CO2 solubility in ILs. This enables IL screening for CO2 related applications.

AB - The solubility of carbon dioxide (CO2) in solvents is important for carbon capture and utilization technologies, with ionic liquids (ILs) being promising due to their ability to capture CO2. Since the number of possible ILs is huge, predicting CO2 solubility during solvent screening is essential. In this work, various machine learning (ML) models including, multiple linear regression, artificial neural network, and random forest, were developed by using 9864 data points covering 124 ILs and descriptors from the σ-profile for predicting CO2 solubility in ILs. The random forest model produced the best performance (R2 = 0.9754, MAE = 0.0257). We estimated the importance of the descriptors, highlighting that those with non-polar characteristics of the σ-profile are important. Lastly, we predicted CO2 solubilities for 1444 unstudied ILs. The combination of ML with the σ-profile descriptors offers great generalizability for predicting CO2 solubility in ILs. This enables IL screening for CO2 related applications.

KW - Quantitative structure–property relationship

KW - Machine learning

KW - Ionic liquids

KW - Carbon dioxide solubility prediction

KW - σ-profile

UR - https://www.scopus.com/pages/publications/85218339374

U2 - 10.1016/j.ces.2025.121226

DO - 10.1016/j.ces.2025.121226

M3 - Article

SN - 0009-2509

VL - 307

JO - Chemical Engineering Science

JF - Chemical Engineering Science

M1 - 121226

ER -