11:00 AM - 11:15 AM
[SCG49-02] First-principles molecular dynamics study of the CO2-water interface using unsupervised machine learning
Keywords:first principles molecular dynamics, machine learning, carbon neutral, liquid liquid interface
Carbon dioxide (CO2) sequestration has been widely recognized as one of key approaches to net-zero carbon emissions. An important factor in CO2 sequestration is proper control of the CO2 injection into the pore space of the reservoir rocks that is saturated by saline fluids (e.g., brine). Understanding the behavior of CO2 in contact with the reservoir fluids such as brine is, therefore, vital for successful CO2 sequestration.
Molecular dynamics (MD) simulations are one of suitable approaches to gaining insights into the CO2-water interface at a microscopic level. However, no attempt has been made to characterize the CO2-water interface within the accuracy of ab initio calculations. In this study, we performed first-principles MD simulations based on density functional theory (DFT) to investigate the structural and dynamical properties of the CO2-water interface, which were compared to those obtained in classical force field (FF) MD simulations.
The density profile of the CO2-water interface shows that CO2 molecules prefer being close to the interface at a low pressure in the DFT model, while this tendency is less prominent in the FF model. This difference is considered to come from an insufficient description of the attractive interaction between CO2 and water molecules in the FF models employed in this study, indicating the importance of accurate description of the CO2-water molecular interaction under low density conditions.
The distribution of molecular orientation shows that both the DFT and FF models exhibit essentially the same profile of the orientation of water and CO2 at the CO2-water interface. We found that on either side of the CO2 and water phase in the interface region, the water dipole tends to point to the water phase with the molecular plane parallel to the interface.
Analyses using multidimensional scaling (MDS) and time dependent PCA (TDPCA) [1] for the water reorientation revealed that the DFT model exhibits substantial structural fluctuations of water molecules at the interface, while the FF model does not. TDPCA further clarified that collective modes of water reorientation greatly change in the structural fluctuations.
The present findings would provide a profound understanding of the behavior of the CO2-water system under conditions typical of subsurface storage.
[1] T. Morishita, J. Chem. Phys. 155, 134114 (2021).
Molecular dynamics (MD) simulations are one of suitable approaches to gaining insights into the CO2-water interface at a microscopic level. However, no attempt has been made to characterize the CO2-water interface within the accuracy of ab initio calculations. In this study, we performed first-principles MD simulations based on density functional theory (DFT) to investigate the structural and dynamical properties of the CO2-water interface, which were compared to those obtained in classical force field (FF) MD simulations.
The density profile of the CO2-water interface shows that CO2 molecules prefer being close to the interface at a low pressure in the DFT model, while this tendency is less prominent in the FF model. This difference is considered to come from an insufficient description of the attractive interaction between CO2 and water molecules in the FF models employed in this study, indicating the importance of accurate description of the CO2-water molecular interaction under low density conditions.
The distribution of molecular orientation shows that both the DFT and FF models exhibit essentially the same profile of the orientation of water and CO2 at the CO2-water interface. We found that on either side of the CO2 and water phase in the interface region, the water dipole tends to point to the water phase with the molecular plane parallel to the interface.
Analyses using multidimensional scaling (MDS) and time dependent PCA (TDPCA) [1] for the water reorientation revealed that the DFT model exhibits substantial structural fluctuations of water molecules at the interface, while the FF model does not. TDPCA further clarified that collective modes of water reorientation greatly change in the structural fluctuations.
The present findings would provide a profound understanding of the behavior of the CO2-water system under conditions typical of subsurface storage.
[1] T. Morishita, J. Chem. Phys. 155, 134114 (2021).