Ab initio geochemistry of hydrous phases

Hydrous phases are among the most important Earth components. They are important for a broad suite of Earth processes, including the origin of life. From both thermodynamic and structural perspectives, however, they represent some of the most complex naturally occurring materials: their bonding often includes a combination of strong covalent, weak ionic, van der Waals, and hydrogen bonding, all within large unit cells. Most are solid solutions, and many are prone to variations in layer packing. Thermodynamic modeling of these materials is fundamental for understanding present and past natural processes, including those that shaped—and continue to shape—the structure and evolution of our planet. Yet, thermodynamic properties of these materials at appropriate conditions are difficult to measure. To make significant progress and attain a deep understanding of these materials requires an atomistic theoretical approach.

This proposal focuses on understanding, from an atomistic perspective, the way water combines with minerals and oxides at crustal and upper mantle conditions. First, we want to shed light on fundamental geochemical problems related to the circulation of water in the planet's interior. This task starts with ab initio computations of heats of formation, free-energies, and thermodynamic properties at high pressures and temperatures (P,T). Second, we want to facilitate the seismic detection of the presence of water in hydrous phases in the initial stages of water ingestion at subduction zones. This research involves calculations of thermoelastic properties of these phases. Third, we want to advance the interpretation of laboratory experiments based on vibrational spectroscopy investigating the incorporation and release of water from nominally hydrous minerals at high P,T. This work will shed light on reactions leading to the retention and volcanic release of water from Earth's interior.

These proposed studies are computationally intensive. They are not only high performance (HPC), but also involve high throughput calculations (HTC). In this regard, we will need to maintain and enhance a computational infrastructure for our calculations. Given the effort involved and expertise required to develop this framework, we hope to make it a community infrastructure. Previously, we have developed the VLab cyberinfrastructure (VLab-CI) to facilitate such calculations. The VLab-CI is a software ecosystem that abstracts the manual work involved in these HPC/HTC calculations. It consists of a Web portal from where calculations are submitted and monitored, workflows automating thermodynamic and thermoelastic calculations, plus several databases that store all the information (data, metadata, post-processed data, etc.) produced. We will revamp and deploy the VLab–CI in the Oak Ridge leadership computing facility (ORNL-LCF), Titan/Summit. This work will require further developing the latest GPU version of the Quantum ESPRESSO software to execute optimally on Summit. The VLab deployment at ORNL represents a notable enhancement in computing power and stability and will serve as a long-term robust HPC platform for us and the broader community. The outstanding HPC resources of ORNL-LCF will be vital for the studies proposed here and for a future community infrastructure. The VLab-CI will run simulations and host databases containing all the detailed results produced by the users, past, present, and future.