Periodic Reporting for period 4 - CC4SOL (Towards chemical accuracy in computational materials science)
Période du rapport: 2022-01-01 au 2023-06-30
Computational materials science makes it possible to predict properties of real materials by solving the underlying many-particle Schrödinger equation in an approximate but computationally efficient manner. These predictions allow to understand and even design novel materials with the desired properties for their usage in, e.g. energy conversion, transport or storage applications. Therefore, computational materials science can have a significant impact on the environment and our society. However, the approximations involved often limit its predictive power, making it necessary to improve upon the accuracy further. This can be achieved by employing more accurate theories that exhibit in general a less beneficial trade-off between accuracy and computational cost.
The present project explored ideas and developed new methods that reduce the computational cost of highly accurate coupled cluster theories employed in computational materials science simulations of periodic crystals and surfaces. The demand and prospects for these methods are excellent given that they can predict atomization- and reaction energies in a wide range of solids and molecules with chemical accuracy (≈43 meV). This project successfully reduced their computational cost by developing more efficient representations and convergence acceleration techniques for the calculation of the underlying many-electron wavefunctions and expectation values. In addition to the methodological developments, the study of challenging solid-state physics and chemistry problems formed an important part of this project. We employed the newly developed methods directly to investigate molecular adsorption and reactions on surface and pressure-driven solid-solid phase transitions.
The present project explored ideas and developed new methods that reduce the computational cost of highly accurate coupled cluster theories employed in computational materials science simulations of periodic crystals and surfaces. The demand and prospects for these methods are excellent given that they can predict atomization- and reaction energies in a wide range of solids and molecules with chemical accuracy (≈43 meV). This project successfully reduced their computational cost by developing more efficient representations and convergence acceleration techniques for the calculation of the underlying many-electron wavefunctions and expectation values. In addition to the methodological developments, the study of challenging solid-state physics and chemistry problems formed an important part of this project. We employed the newly developed methods directly to investigate molecular adsorption and reactions on surface and pressure-driven solid-solid phase transitions.
A multitude of developments that reduce the computational cost of coupled cluster theory calculations for real materials have been explored in this project. The most notable developments include novel corrections to computed electronic correlation energies that accelerate their convergence to the thermodynamic limit and complete basis set limit, reducing the computational cost significantly. The research on finite basis set corrections was carried out by Andreas Irmler et. al. and has been published in PRL 123, 156401 (2019), JCP 151, 104107 (2019) and JCP 154, 234103 (2021). Furthermore, the work on finite size corrections for coupled cluster theory calculations of periodic systems has been published in PRX 8, 021043 (2018).
In addition to the methodological developments, several more applied research projects have been carried out and resulted in published research articles. These applied projects focus on solid-solid phase transitions, molecular adsorption on surfaces and defects. Two articles about solid-solid phase transitions have been published in PRB 98, 134108 (2018) and npj Computational Materials 5, 1-6 (2019). In these studies, we have computed enthalpy differences of different carbon and boron nitride allotropes as well as high pressure phases of hydrogen that can be used to predict temperature- pressure phase diagrams. In addition, we have performed highly accurate ab initio investigations on the interaction between single water molecules and sheets of h-BN and graphene. The obtained findings have been published in PRX 8, 021043 (2018) and JPCL 3, 358 (2019). Furthermore, an investigation of dissociative hydrogen adsorption on Silicon has been published in JCP 149, 244105 (2018). A recent work focused on the study of defects in solids and is published in PRB 108, 115125 (2023). The reported applications have clearly demonstrated the potential and high accuracy of the investigated coupled cluster methods for computational materials science. We hope that these findings will serve as useful reference values in future studies and help to improve the accuracy and efficiency of widely used methods used in computational materials science in general.
In the final phase of this project the developed computer codes used to perform massive parallel coupled cluster theory calculations have been published and made freely available at https://github.com/cc4s. Alongside the computer code, we have published a documentation describing the code and how it can be used.
In addition to the methodological developments, several more applied research projects have been carried out and resulted in published research articles. These applied projects focus on solid-solid phase transitions, molecular adsorption on surfaces and defects. Two articles about solid-solid phase transitions have been published in PRB 98, 134108 (2018) and npj Computational Materials 5, 1-6 (2019). In these studies, we have computed enthalpy differences of different carbon and boron nitride allotropes as well as high pressure phases of hydrogen that can be used to predict temperature- pressure phase diagrams. In addition, we have performed highly accurate ab initio investigations on the interaction between single water molecules and sheets of h-BN and graphene. The obtained findings have been published in PRX 8, 021043 (2018) and JPCL 3, 358 (2019). Furthermore, an investigation of dissociative hydrogen adsorption on Silicon has been published in JCP 149, 244105 (2018). A recent work focused on the study of defects in solids and is published in PRB 108, 115125 (2023). The reported applications have clearly demonstrated the potential and high accuracy of the investigated coupled cluster methods for computational materials science. We hope that these findings will serve as useful reference values in future studies and help to improve the accuracy and efficiency of widely used methods used in computational materials science in general.
In the final phase of this project the developed computer codes used to perform massive parallel coupled cluster theory calculations have been published and made freely available at https://github.com/cc4s. Alongside the computer code, we have published a documentation describing the code and how it can be used.
Most methodological developments carried out in this research project go beyond the state of the art and are strongly interdisciplinary, situated at the cross-roads of theoretical chemistry and condensed matter physics. It would not have been possible to obtain the results described above without the newly developed techniques that reduce the computational cost of the employed methods substantially.