Scalable Quantum Chemistry with Flexible Embedding Stage 2

In stage two of the "Scalable Quantum Chemistry with Flexible Embedding" project we propose to refine the software we have developed in stage one for molecular modelling of reactivity in complex systems, in particular surface chemistry and heterogeneous catalysis in the presence of a solvent such as water. This software combines quantum chemistry (also known as first principles) techniques, which can treat the chemically reacting centres, embedded in an empirical (molecular mechanics) model for the rest of the system, i.e. the slab of material which models the surface and the solvent layers on top of it.

In stage one we are extending our embedding model to include new treatments for spin-polarised systems such as magnetically ordered materials, an implementation of a low-cost quantum method to describe anisotropic and spin-polarised environments and periodic boundary conditions in order to perform calculations on solvated surfaces. Examples of the science this software targets includes metal oxide catalysis that work in the presence or water; the binding of pollutant species (heavy metals or toxic organics) to natural minerals in the environment, or specially designed inorganic materials that could potentially remove them from the environment; catalysis by more complex minerals, not well treated by existing treatments but of great industrial importance; and the design of sensors and photoelectric materials by organic layers on inorganic surfaces.

The aim of stage two is to make it is as straightforward as possible for new users to set up and run flexible embedding calculations based on the methods developed in stage one, and for advanced users to adapt the code to their own needs. We will achieve this by replacing the user interface the stage one code, which is able to achieve the scientific goals of stage one but is many-layered and relies on a legacy software package, with a new direct user interface written in the popular scripting language Python. The new interface will radically simplify use and development of the code without compromising on the primary goal of stage one, namely achieving high performance for the new embedding models on large-scale parallel machines.

We will incorporate all the key functionality from our legacy codebase into the new version of the code, including utilities to set up embedded cluster calculations, an advanced geometry optimisation library, and the ability to interface to multiple quantum mechanical and molecular mechanical software packages, in particular the FHI-AIMS package which will be coupled to the code in collaboration with the FHI-AIMS developers. We also aim to make the new code as user-friendly as possible through the development of comprehensive documentation, tutorials and case studies from the demonstration applications program began in stage one.

By the end of stage two we will have a new version of the code that is ready for release to the community, with all the key functionality of the legacy version implemented, and we will be in a position to transition users to the new code and use it as our platform for all future QM/MM and multiscale developments.

Materials research is an underpinning technology for a wide range of UK industries. In the chemicals industry economic and environmental sustainability of processes depends on innovation in the area of catalysis. The emerging area of supported nanoparticulate catalysts depends on the construction of specific metal/support combinations with the particle size and interface with the oxide playing an important role in catalyst performance. Advances in this field will require fundamental understanding of the connection between catalyst structure and activity for which modelling at the quantum level is an indispensible tool. The new software package developed as a result of stage 2 will help improve the range of problems that can be addressed by both academic and industrial researchers in this area, without needing the intimate knowledge of the methods and codes required to operate the transitional workflow in stage 1.

For systems of interest, new fundamental insights will be obtained. Materials are widely used in industrial context, including electronics, catalysis, energy materials, environmentally friendly materials and materials used as filters and sorbents tackling chemical and or radioactive waste. New classes of minerals will be addressed along with their interfaces with water thus impacting on our knowledge in earth sciences (calcite and water cycles) and biominerals (apatite in tooth enamel and bones etc.)

The partners are all active participants in the Collaborative Computational Projects, in particular CCP5 (SCP is chair and CRAC is a past chair). The developments described here are believed to be highly relevant to the use of condensed phase simulation techniques for systems of technological and commercial interest, and it is anticipated that many within the CCP5 network will be able to exploit the software within their own commercial interactions.
STFC and UCL have worked together on an earlier embedding scheme for solids and implemented it in ChemShell. Commercial interest in the model was the main motivation for Accelrys to licence ChemShell as the basis of their QMERA product, which they are now selling commercially. Once the ability of the stage 2 package to interface to multiple QM packages has been demonstrated, it should be straightforward to introduce the required additional functionality into the Accelrys QM code, DMol3, so that the new methods can be exploited by users of Materials Studio, which is widely used in industry.