
Computational Materials and Molecular Sciences
Computational Materials and Molecular Science develops theory and methods to implement in software solutions for scientific and technological challenges. It is cross-disciplinary, multi-scale, and focused on sustainable development on modern hardware and data infrastructures.
Computational Materials and Molecular Science helps to:
- Provide theory and simulation of atomic, molecular, macromolecular and materials properties and processes
- Provide community–focused development, maintenance and training in sustainable software for molecular and materials physics and chemistry
- Provide simulation-assisted interpretation of combined or isolated large-scale and lab-scale experiments
- Develop multiscale and multi-physics simulation techniques to model chemical and physical processes in realistic conditions
- Enable reproducible workflows and open science in materials science using FAIR data principles
- Harness the potential of artificial intelligence for computational materials and molecular science
- Enable next generation software for exascale computing platforms and emerging computing paradigms
Current External Funded Projects:
EPSRC-SFI: Emergent Magnetism and Spin Interactions in Metallo-Molecular Interfaces
The interface between two materials can be used to give rise to new properties that neither component could have separately (emergence), to tune the capabilities found in of one of them (enhancement), or to share functionalities (proximity). Our range of magnetic materials is limited; only the metals iron, nickel and cobalt show spontaneous magnetic ordering at room temperature. Here, we use molecular interfaces to generate novel magnets outside the Stoner criterion, to control the spin properties of thin films and add functionalities. From a fundamental point of view, the origin of these effects is not fully explained due to the complexity of the interfaces, the materials involved and their intricate quantum-electronic properties. This project aims to fill these scientific gaps by 1) developing new theoretical framework to study magneto-molecular coupling and interfaces, 2) improving the properties of commonly used magnetic thin films via nanocarbon overlayers and, 3) creating the opportunity for switchable magnetism by turning on/off the interfacial spin ordering using electric fields.
Find out more on EP/S031081/1
Supporting research communities with large-scale DFT in the next decade and beyond
ONETEP is a world-leading UK-developed software package which uses a linear-scaling framework to enable large scale Density Functional Theory (DFT) calculations, uniquely without loss of accuracy compared to traditional methods. As with all software, ONETEP needs to be continuously evolved and updated in order to stay at the cutting edge. This is particularly challenging for a large collaborative academic project that has evolved over two decades. This will re-engineer the code in its entirety, rationalising internal structure to allow further development and enhance the interoperability of existing functionality. Modern software engineering principles will be followed throughout, in close collaboration with the computational physics and chemistry groups of STFC SCD and research software engineers in Southampton, Warwick and Imperial. At the same time developments of new functionality to enable large-scale calculations of crystalline and semicrystalline materials will satisfy a demand in this area by many researchers, such as in the CCP9 and the solid state microscopy and spectroscopy communities at STFC Facilities. Workflow tools and coupling with the ChemShell QM/MM code will be developed to allow adoption of the code by the biomolecular simulations community. The code will also be ported to emerging supercomputing architectures with GPU accelerators.
Find out more on EP/W029480/1
The UK Car-Parrinello HEC Consortium
The United Kingdom Car-Parrinello Consortium (UKCP) is a group of researchers across the UK who develop `first principles’ quantum mechanical techniques for studying atomistic systems and apply them to a very wide variety of systems. UKCP also supports experimental communities, via computational training, RSE time and computer allocations on Tier-1 and Tier-2 HPC. The close interaction between DFT theorists, software developers and users drives innovation and expands simulation capabilities, as well as magnifying the impact of the work. The research proposed does not easily fit traditional categories of “physics”, “chemistry” etc; instead, UKCP is a multidisciplinary consortium using a common theoretical foundation to advance many areas of materials-based science, with the potential for significant impact both in the short and long-term. The innovations in this proposal enable the next generation of simulations and further widen our computational horizons. UKCP will develop new algorithms, workflows & theoretical methods to increase our simulation abilities, in terms of both new functionality and dramatically improved accuracy & speed. New algorithms include embedding machine learning methods into DFT to speed up calculations, and enabling treatment of large systems (bringing together the CASTEP & ONETEP codes into a single workflow and enabling DFT codes to be embedded in multiscale, multiphysics simulations). GPU ports and improved parallelism enable UKCP software to exploit current and future HPC architectures effectively & with greater energy efficiency. New functionality includes NMR spectroscopy with spin-orbit coupling, so the full periodic table can be studied with high accuracy, and advances in excited state modelling, including temperature and environmental effects.
Find out more at EP/X035891/1 and UKCP Main/Home Page
CCP9: Computational Electronic Structure of Condensed Matter
CCP9 has a large group of researchers in electronic structure in the UK that develops, implements and applies computational methods in condensed matter. The electronic structure of condensed matter underpins a vast range of research in Materials Science, including but not limited to areas such as semiconductors, superconductors, magnetism, biological systems, surfaces and catalysis. The computational methods are very powerful in helping us to understand complex processes and develop new technologically important materials. The researchers in CCP9 develop first-principles methods to solve for the electronic structure of materials and obtain materials properties. First principles methods employ the fundamental equations of quantum mechanics as starting point and do not rely upon experimental input. Our calculations therefore predict the behaviour of materials without bias, adding insight independent from experiment that helps us to explain why materials behave as they do. The CCP9 community develops a number of major, internationally leading codes for electronic structure solution and these codes run on the whole range of computational architectures available to us today from PCs to national and international supercomputing facilities, and we support as much as possible new chip architectures such as Arm and GPU. Not only do we develop codes for these machines but also train a large number of people to understand the underlying science and use the codes through many workshops, training sessions, hands-on courses and also to present work at the CCP9 networking meetings. Throughout all of this our leading experts, both UK and internationally, engage with the community particularly our young researchers to train and enthuse. CCP9 is a strong partner with our EU colleagues in the Psi-k network reaching many thousands of electronic structure code developers, software engineers and applications scientists.
Find out more at EP/T026375/1 and https://ccp9.ac.uk/
CCP-NC: Collaborative Computational Project in NMR Crystallography
Solid-state nuclear magnetic resonance (NMR) is capable of providing extremely detailed insights into the structure and dynamics of a wide range of materials – from organic systems, such as pharmaceutical compounds and supramolecular arrays, to inorganic materials for next-generation batteries and safe storage of nuclear waste. Such information is crucial for harnessing the properties of increasingly complex new materials, and to address major challenges across the physical sciences. However, the true potential of this experimental technique is only realized through combination with advanced computational methods. In particular, first-principles electronic structure predictions of key NMR interactions, such as chemical shifts, allow experimental measurements to be directly linked to structure. In tackling challenging problems, the developing field of NMR crystallography benefits from close interaction with other experimental techniques, typically powder X-ray diffraction, and computational approaches, particularly crystal structure prediction. The Collaborative Computational Project for NMR Crystallography supports this multidisciplinary community of NMR spectroscopists, crystallographers, materials modellers and application scientists, who work both within academia and industry. We develop overarching software tools enabling a largely experimentally focused community to exploit advanced computational techniques.
Find out more at EP/T026642/1 and https://www.ccpnc.ac.uk/
CCP-QC: Collaborative Computational Project - Quantum Computing
CCP-QC is a network linking computational scientists with quantum computing scientists and engineers, to develop some of the first useful applications of quantum computers. Quantum computing is promising fundamentally faster computation as part of broader quantum technology development that includes more secure communications, and more sensitive measurements and imaging. Our conventional computers, including those in mobile phones, modern cars, and powering the internet, are based on silicon semi-conductor technology. After half a century of growth, silicon semiconductor computer chips have been at the limit of what they can do for the past decade. Faster computing requires more computers, which use more electricity and this growth is thus limited. Quantum computing uses a different logic, enabling much faster computing for some types of problems. The engineering challenges are formidable, and we are still at the stage equivalent to the first semiconductor chips in the early 1960s. Early quantum computers are already available: developing applications to suit the capabilities of this hardware is the next step, to enable us to take advantage of the opportunities they offer to speed up our computations. An important set of computational tasks in materials, chemistry, physics, biology, and engineering is developed by communities supported by collaborative computational projects (CCPs). CCP-QC will network across these CCPs and the quantum computing community, to enable the CCP communities to enhance their computations by using quantum computers. It will do this by organising joint meetings, holding training days to teach computational scientists about quantum computing, supporting small projects to develop proof-of-principle code and demonstrations on early quantum computing hardware, and providing an online information resource on early quantum computing applications.
CCP-QC will interface with the new National Quantum Computing Centre, to be launched in April 2020 and based on the STFC Harwell campus in Oxfordshire. CCP-QC will enable quantum computing hardware providers to have their hardware tested with real problems of importance to the computational science communities. The outcomes of such tests can thus influence the development of quantum computing hardware, leading to faster development of useful applications that are adapted to extract the best advantage from the early quantum hardware. The simulations carried out by the CCP communities cover a wide range of important applications, from smart materials (e.g., better solar cells and batteries) to drug design (bio-molecular simulation). CCP-QC will thus contribute to the development of faster computational methods in many important applications with wide-ranging scientific, social and economic benefits.
Find out more at EP/T026715/2 and CCP-QC
EuroScienceGateway: Leveraging the European compute infrastructures for data-intensive research guided by FAIR principles
The aim of the EuroScienceGateway project is to enable researchers, with widely ranging digital skills, to efficiently use the diverse computational infrastructures available across Europe for their research needs regarding compute and data resources, tools, and application platforms. EuroScienceGateway is working to deliver this vision through four specific project objectives, linked to measurable outcomes, that connect the work across the work packages and infrastructures:
Objective 1: Accessible e-Infrastructure resources for European scientists to enable pioneering data-driven research across scientific domains.
Objective 2: Support the varieties of analysis types and diverse usage patterns through efficient and smart job distribution to appropriate and sustainable infrastructures.
Objective 3: The application of FAIR principles to workflows and adoption of FAIR Digital Objects to stimulate reusable and reproducible research and enable the EOSC Interoperability Framework.
Objective 4: Adoption of the EuroScienceGateway by researchers in diverse scientific disciplines.
We are mainly involved with Objective 4: working to build a community of practice for sustainable software for muon science and x-ray absorption spectroscopy. As part of this, we are adopting the Galaxy platform for managing FAIR workflows for combining different software tools for muon and x-ray science.
Find out more at:
GCRYSTAL - A GPU enabled CRYSTAL
Modern supercomputers are constantly evolving beasts in their search for increasing performance within a manageable power envelope. Recently a real boost in their performance has been achieved through the introduction of GPGPUs. But performance in hardware is just a small part of the story, without software to exploit the hardware it is, to put it bluntly, useless. To address this RSEs are porting major software packages to GPUs, a non-trivial exercise due to the markedly different architecture and a lack of fully established software tools and environments.
This project focuses on CRYSTAL, an ab initio electronic structure code used by hundreds of researchers in solid state chemistry and physics. It is funded through the eCSE program within the ARCHER2 project. The work is to produce a production quality GPU enabled version of CRYSTAL. Being such a flexible package the work can impact a wide range of areas, recent use cases including batteries, solar cells, energy storage and conversion, CO2 recycling and sequestration, homogeneous and heterogeneous catalysis, surface science, novel materials, ferro/pyroelectrics thermoelectrics, electrocalorics,nano-electronics and spintronics
Optimising Memory Use in CRYSTAL23 and The Implementation of Kubo-Greenwood Theory
This project focuses on the calculation of transport properties, such as electrical and thermal conductivity, within CRYSTAL, an ab initio electronic structure code used by hundreds of researchers in solid state chemistry and physics. Such calculations underpin research in such technologically import areas as thermoelectrics, electrocalorics and energy storage/conversion to nano-electronics, spintronics, catalysis and photo/electro-catalysis.
The work involves both the improvement of an exisiting implementation within the code, and the extension of that implementation to a higher level of theory. The exisiting implementation, based upon Boltzmann transport theory (c.f. Boltztrap2), is limited in its parallelism and very memory hungry, exacerbated by the dense k-point nets required by these calculation, so limiting the scale of system that can be addressed. The new implementation will address both these issues by distribution of the required objects across multiple processes so improving both the time and memory scalability of the code. The second objective is to implement a higher level of theory, Kubo-Greenwood (see for example), so improving the accuracy of the code.
It is funded through the eCSE program within the ARCHER2 project.
Many-body theory of antimatter interactions with molecules
Positrons are unique probes of matter and have important use in fundamental physics, materials science, medical imaging and astrophysics. The proper interpretation and development of the (difficult and costly) positron-based experiments/technologies requires accurate theoretical support and predictive capability. However, low-energy positron-matter interactions are characterised by strong correlations, which make their theoretical description a challenging many-body problem. We have recently developed a many-body approach to positron interactions (binding/scattering/annihilation) in molecules that accounts for the important correlations, implemented in our currently unrivalled code “EXCITON+” [Nature 606, 688 (2022)]. The code, adapted from EXCITON (authored by Co-I Patterson), is in its infancy: focus has been on developing scientific capabilities; it is not optimised for large-scale HPC architectures.
This project has two strands:
1) a major software engineering objective to improve the usability, robustness and maintainability of EXCITON/EXCITON+ to enhance public release (including introducing unit tests/regression testing, development of documentation, and improving build and deployment);
2) profiling and optimisation of the codes (including re-engineering of the parallelisation structure), to optimise efficiency, scaling and load balancing, reducing the code’s carbon footprint, and making calculations on larger molecules, e.g., polymers, feasible.
The project is currently in progress and a project report will be submitted in due course. See ARCHER2 eCSE and eCSE Panels for details of the ARCHER2-eCSE programme.
Enhanced parallelisation for R-matrix with time-dependence, double ionisation
R-matrix with time-dependence (RMT) is the most sophisticated code in the world for the treatment of laser-driven multielectron dynamics. The code can describe electron dynamics in atomic and molecular systems, driven by arbitrarily polarised laser pulses, and include relativistic effects in atomic systems. Recently, we have extended the code to account for double-ionisation phenomena, wherein two electrons may be ejected from a target atom simultaneously. This massively increases the complexity and computational workload of the calculations. The description of the second electron at large distances from the parent ion requires a two-dimensional grid, one dimension for each emitted electron. To handle this complexity, the RMT code divides physical space into three distinct regions: an inner region close to the nucleus, a single-ionization region wherein one electron is far from the nucleus, and a double-ionization region where two electrons are far from the nucleus. Each region adopts a numerical and parallel scheme appropriate to the physics it describes. This set-up requires sophisticated parallelisation and communication to manage the flow of information between regions. In this project, we will address the parallelisation in the double-ionization region, and its associated communication, to enable larger calculations harnessing a greater number of processing elements more efficiently, and profile those calculations to identify optimal strategies for load-balancing.
The project is currently in progress and a project report will be submitted in due course. See ARCHER2 eCSE and eCSE Panels for details of the ARCHER2-eCSE programme.
Battery Cell Assembly Twin (BatCAT)
BatCAT, is led by the Norwegian University of Life Sciences and STFC has a major role in it. STFC aim to gain experience and competitive advantage in using modern technologies in a novel way and help battery stakeholders improve their design processes and understanding of the materials composing the batteries and their full manufacture and operational lifecycles.
The project, which is funded by Horizon Europe, will leverage modern data technologies to support decision making and product/process optimisation in battery manufacturing. It will allow all the involved actors (as scientists, engineers, technicians) to integrate their data, knowledge and models in a smart way, enabling faster and trustworthy innovation routes.
Battery cells and storage are big business, but they are also expensive and require huge amounts of research and development time and investment in order to maximise their capacity and safety over their lifecycles while minimising health and safety risks with forward thinking of reusability and recycling. This is where a digital twin comes in.
A digital twin brings together a tangible physical entity with an identical virtual landscape; created by scientific computing experts using data from experiments and simulations, the Internet of Things and models. It aims to be a digital replica of the product and the process it is representing, and as such it is crucial to build in as much detail and data as possible to ensure accuracy.
By creating virtualised versions of products or situations first, developers, researchers, technologists, business decision makers etc can see the entire process, foresee and ward off issues before they ever exist in the real world.
By building a digital twin for battery production much of that development could happen in the digital world. It would allow developers to identify and avoid issues even before a prototype is produced. These could include:
- the optimum designs for a battery cell or storage device
- the most prospective manufacturing processes and environments
- how the batteries would behave in ‘real life’ rather than in standardised lab tested use. This would cover the effects of important aspects such as operational and environmental temperatures, usage and storage on the life span and degradation of the battery components
- how safe the digitally designed batteries will be
- environmental disposal and recycling costs
By mid 2027, the BatCAT consortium aims to have created a proof-of-concept digital twin for lithium-ion batteries (LIBs) and redox flow batteries (RFBs). In the longer term the consortium aims to use the learnings to create a live digital twin of cell manufacturing for LIBs and RFBs while also developing methodologies that will be adapted for the manufacture of new battery technologies.
The consortium is made up of 18 organisations from across Europe, including research bodies, universities and small businesses. By working together, the small businesses will be able to access far more data and potentially reap bigger outcomes than trying to undertake a development of this scale alone. A substantial fraction of non-commercial data coming out of the project will be open to all to benefit society as a whole.
A team of six staff from STFC’s Scientific Computing department will be taking the lead on data and aspects of materials modelling for the digital twin. They will also ensure that real-time data from battery production and testing facilities are properly integrated within the digital twin workflows that combine these with data from trusted sources, experiments and computational modelling.
DOME 4.0
DOME 4.0 aims to offer an intelligent semantic industrial data ecosystem for knowledge creation across the entire materials to manufacturing value chains. The ecosystem intends to demonstrate a sustainable solution to the information silos problem related to the past efforts and puts forward a formal, ontology-based documentation for open and confidential data spaces applicable to future and current projects thereby delivering added value. The flexibility of the semantic architecture of DOME 4.0 is adapted to the emerging Industry Commons developments and can facilitate the scale-up of the ecosystem to large amounts of data, tools and services applicable to wider sectors of the European economy.
The unique offerings of DOME 4.0 are twofold. Primarily, to instigate wider market impact, stakeholder adoption and engagement, while aggregating a critical mass community in the DOME 4.0 ecosystem than the individual-encompassed showcases would ever achieve alone. In parallel, the DOME 4.0 exemplified, novel business models cross-cutting the individual marketplaces, complemented with transparent and fair compensation schemes will augment the operations and effectiveness by adding value to the individual marketplaces, data repositories and platforms.
STFC leads on a chemical information showcase that enables the integration of data sources in a dynamic table. The dynamic table enables comparison of values for the same variable coming from different sources, with a clear tracking of data provenance (the source is explicitly tracked, the eventual upstream ones are available in the raw data, depending on the provider). The approach is generalisable and extensible by design with results shown in a coherent view, and a specialised “Further information” box to elucidate the major concepts used. This is not generally possible otherwise, for example, using a generic Internet search engine, and must be done manually or by bespoke codes. This capability improves the findability of reliable information, its integration (for the cases where multiple sources are complementary), comparison (for the cases where multiple sources overlap) and extendibility (where information from multiple sources does not overlap but is complementary).
Developing next-generation DL_POLY for the benefit of the modelling community
Molecular dynamics (MD) simulations have become an important modelling tool to understand, predict and optimise the properties of condensed matter phases. There is a growing demand in simulating very large system sizes where important effects operate, including in the areas of energy, environment and advanced materials. Whereas high-performance computing facilities can currently handle very large systems, these simulations produce the amounts of data which can not be stored or analysed. This severely limits the ability of users to run and analyse MD simulations of very large systems. Here, we propose to solve problem by changing the paradigm of how MD simulations are run. Instead of the traditional way of writing the trajectory file and analysing it after the simulation has ended, we will calculate some important physical properties of the system on fly during an MD simulation. We will implement the calculation of these properties in the UK flagship MD code, DL_POLY. This will be accompanied by enhancing and optimising the software engineering, porting and supplying various tools for efficient running of DL_POLY. We will distribute the newly developed code to 5,100 registered users including in academia and industry, develop documentation and conduct training events. As a result of this development, our newly developed MD code will gain a competitive edge over its competitors and will enable its users to access new larger length and energy scales.
Find more at EP/W029006/1 and DLPOLY gitlab repo
Goldilocks convergence tools and best practices for numerical approximations in Density Functional Theory calculations
Within the field of materials and molecular science, modelling and simulation based on Density Functional Theory (DFT) is key in the R&D of functional materials for environmental sustainability, such as green computing, environment remediation, and energy production, conversion and storage. DFT-based research currently consumes a considerable amount of resources on supercomputers globally. In the UK, DFT calculations use over 45% of ARCHER2, the Tier1 UK National Supercomputing service. DFT also features heavily in the usage of Tier2 systems and lower-tier institutional computers. As ever more powerful computers become available, the environmental impact of DFT-based research is increasing rapidly. It is paramount to improve the efficiency of this research and develop means of assuring that energy-intensive compute resources are distributed and used responsibly. The proposed work will provide practical tools and evidence-based best practices towards these aims for researchers and the compute-resources distribution chain.
DFT calculations contain numerical approximations that need to be converged according to the accuracy required for each study. Without more support for inexperienced users, the risk of is of over-convergence, leading to unnecessarily more costly calculations, or under-convergence, leading to entirely useless calculations, which are a waste of compute resource and electricity. A conservative estimate of the proportion of under- or over-converged DFT calculations is in the 10% range. Given the large proportion of compute resource invested in this research, even a relatively small increase in efficiency will result in a large reduction of wasted compute resource, and significant improvements in the environmental sustainability of research infrastructure.
This project will result in a tool and evidence-based best practices to provide automatic, expert guiding in the ‘Goldilocks’ choice of these convergence parameters. This will be achieved by training machine learning (ML) models to predict the convergence parameters for DFT numerical approximations for the required accuracy in common types of scientific investigations. Given that numerical approximations requiring convergence are present in all codes, this tool will be applicable across all DFT codes in common use in the UK. The primary contribution of this project will be to increase considerably the efficiency and assurance levels of responsible use of UKRI and EPSRC hardware and software infrastructure, now and in the future.
Comparison of the compute resources usage for typical jobs run before and after the adoption of this tool, will enable baseline quantification and extrapolation of the efficiency gained. Outcomes of this analysis will be disseminated globally, leading to best practices across international compute Facilities, so as to extend world-wide the gains in environmental sustainability of compute infrastructure.
This project will be a significant step towards ML-based automatic generation of inputs for DFT calculations, as well as an automatic a priori calculator of compute resources and carbon footprint. This automation will contribute to democratisation in the use of this research method in parts of the world where digital research infrastructure may be more accessible than experimental facilities.
Find more at at EP/Z530657/1
CCP5/CCP5++: Integrating Computer Simulation of Condensed Phases with experiments and data science
Molecular modelling has established itself as a powerful predictive tool for a large range of materials and phenomena whose intrinsic multiscale nature requires modelling tools able to capture their chemical, morphological and structural complexity. In the UK, the molecular modelling community, supported by the software, training and networking activities coordinated by the CCP5, has become, over the past 40 years, international-leading in this field. Building upon these successes, the new CCP5++ network will revolutionise the field of materials molecular modelling creating a new integrated community of modellers, experimentalists and data scientists that together will identify the new frontiers of the field and will transform the way these disciplines work together.
To achieve its mission, the CCP5++ will coordinate and support an ambitious plan of meetings, sandpits, coding workshops, secondments and visitor schemes to cater for the large community of modellers, experimentalists and data scientists working on advanced materials. This support has proved to be vital to enable the UK condensed matter community to attain and maintain an international position at the forefront of such an intensely competitive field and will enable the UK researchers to identify and tackle major world challenges in-silico materials discovery.
From the start the network memberships include key representatives of the experimental and data science communities, international software and modelling institutes, industrial collaborators and national HPC consortia and CCPs, that working together will shape the future of materials molecular modelling.
Find out more at EP/V028537/1 and ccp5.ac.uk
Particles At eXascale on High Performance Computers (PAX-HPC)
Many recent breakthroughs would not have been possible without access to the most advanced supercomputers. For example, for the Chemistry Nobel Prize winners in 2013, supercomputers were used to develop powerful computing programs and software, to understand and predict complex chemical processes or for the Physics Nobel Prize in 2017 supercomputers helped to make complex calculations to detect hitherto theoretical gravitational waves.
The advent of exascale systems is the next dramatic step in this evolution. Exascale supercomputing will enable new scientific endeavour in wide areas of UK science, including advanced materials modelling, engineering and astrophysics. For instance, solving atomic and electronic structures with increasing realism to solve major societal challenges – quantum mechanically detailed simulation and steering design of batteries, electrolytic cells, solar cells, computers, lighting, and healthcare solutions, as well as enabling end-to-end simulation of transients (such as bird strike) in a jet engine, to simulation of tsunami waves over-running a series of defensive walls, or understanding the universe at a cosmological scale. Providing a level of detail to describe accurately these challenging problems can be achieved using particle-based models that interact in complicated dance that can be visualised or analysed to see how our model of nature would react in various situations. To model problems as complex as outlined the ways the particles interact must be flexible and tailored to the problem and vast quantities of particles are needed (and or complicated interactions). This proposal takes on the challenge of efficiently calculating the interacting particles on vast numbers of computer cores. The density of particles can be massively different at different locations, and it is imperative to find a way for the compute engines to have similar amounts of work – novel algorithms to distribute the work over different types of compute engines will be developed and used to develop and run frontier simulations of real-world challenges.
There is a high cost of both purchasing and running an exascale system, so it is imperative that appropriate software is developed before users gain access to exascale facilities. By definition, exascale supercomputers will be three orders of magnitude more powerful than current UK facilities, which will be achieved by a larger number of cores and the use of accelerators (based on gaming graphic cards, for example). This transition in computer power represents both an anticipated increase in hardware complexity and heterogeneity, and an increase in the volume of communication between cores that will hamper algorithms used on UK’s current supercomputers. Many, if not all, of our software packages will require major changes before the hardware architectures can be fully exploited. The investigators of this project are internationally leading experts in developing (enabling new science) and optimising (making simulations more efficient) state-of-the-art particle-based software for running simulations on supercomputers, based here and abroad. Software that we have developed is used both in academia and in industry. In our project we will develop solutions and implement these in our software and, importantly, train Research Software Engineers to become internationally leading in the art of exploiting exascale supercomputers for scientific research.
Find out more at EP/W026775/1 or PAX HPC
Materials Chemistry HEC Consortium (MCC)
The Materials Chemistry Consortium is the UK’s High End Computing (HEC) consortium for Materials Chemistry. We exploit high performance computing in a broad programme of work modelling and predicting the structures, properties and reactivities of materials. The project comprises application-driven and cross-cutting themes focused on fundamental challenges in contemporary materials chemistry and physics and advanced methodology. It brings together the UK’s materials academic community, currently representing 38 universities.
Tuning properties of materials forms the backbone of research in Energy Conversion, Storage and Transport, a key application theme for the UK’s economy and net-zero targets. We will aim to improve the performance of materials used in both batteries and fuel cells, as well as novel types of solar cells. In Reactivity and Catalysis, we will develop realistic models of several key catalytic systems. Targets relate strongly to the circular economy and include CO2 activation and utilisation, green ammonia production, biomass conversion and enhancement of efficiency in industrial processes and more effective reduction in air pollution. We will develop environment protecting materials to contain toxic and/or radioactive waste, capture greenhouse gases for long-term storage, remove toxins and pollutants from the biosphere to improve wildlife and human health, and control transmission of solar energy through windows. Work on Biomaterials will reveal the fundamental processes of biomineralisation, which drives bone repair and bone grafting, with a focus on synthetic bone replacement materials. Materials Discovery will support screening materials using global-optimisation-based approaches to develop tailored chemical dopants, improving the desired property of a device, and searching the phase diagram for solid phases of a pharmaceutical drug molecule.
Crosscutting themes will focus on basic issues in the physics and chemistry of matter that underlie the application themes. They will address: challenges in predicting the morphology, atomic structure and stability of different phases; defects and their role in material growth, corrosion and dissolution in Bulk, Surfaces and Interfaces, and at Nano- and meso-scales. Our simulations will investigate materials far from equilibrium, as well as quantum and nano-materials with links to topological spintronics. Software developments will include utilising machine learnt potentials, significantly increasing the time- and length-scales of simulations (compared to electronic structure-based calculations) without compromising their accuracy and predictive power. We will continue to develop new functionalities and optimise performance of internationally leading materials software and link to research exploiting quantum computers.
Find out more at EP/X035859/1 or the MCC website.
Predictive multiscale free energy simulations of hybrid transition metal catalysts
Catalysis is a key area of fundamental science which underpins a high proportion of manufacturing industry. Developments in catalytic science and technology will also be essential in achieving energy and environmental sustainability. Progress in catalytic science requires a detailed understanding of processes at the molecular level, in which computation now plays a vital role. When used in conjunction with experiment, computational modelling is able to characterise structures, properties and processes including active site structures, reaction mechanisms and increasingly reaction rates and product distributions. However, despite the power of computational catalysis, currently available methods have limitations in both accuracy and their ability to model the reaction environment. Also, it is practically difficult to model hybrid catalysts, which combine elements of different types of catalyst (e.g. unnatural metal centres incorporated in natural enzymes). Advances in technique are essential if the goal of catalysis by design is to be achieved.
A powerful, practical approach to modelling catalytic processes is provided by Quantum Mechanical/Molecular Mechanical (QM/MM) methods, in which the reaction and surroundings are described using an accurate quantum mechanical approach, with the surrounding environment modelled by more approximate classical forcefields. QM/MM has been widely and successfully employed in modelling enzymatic reactions (recognised in the 2013 Nobel prize for Chemistry) but has an equally important role in other areas of catalytic science.
The flagship ChemShell code, developed by the STFC team in collaboration with UCL, Bristol and other groups around the world, is a highly flexible and adaptable open source QM/MM software package which allows a range of codes and techniques to be used in the QM and MM regions (www.chemshell.org). The software has been widely and successfully used in modelling enzymatic reactions and catalytic processes in zeolites and on oxide surfaces. It will provide the ideal platform for the developments in this project which will take computational catalysis to the next level. These will include the use of high level QM techniques to achieve chemical accuracy, accurate modelling of solvent effects, calculation of spectroscopic signatures allowing direct interaction with experiment, and dynamical approaches for free energy simulations. Crucially, we are bringing together methods from different spheres of computational catalysis to enable modelling of hybrid catalytic systems. We are developing flexible and rigorous methods that meet the twin challenges of high-level QM treatment for accuracy with the ability to sample dynamics of the reacting system. Together these methods will allow accurate and predictive modelling of catalytic reactions under realistic conditions. The project will also anticipate the software developments needed to exploit the next generation of exascale high performance computing.
We are applying these new techniques to model the catalytic behaviour of a range of engineered heterogeneous, homogeneous and biomolecular catalysts, currently under study in the UK Catalysis Hub. The Hub supports experimental and computational applications across the whole UK catalysis community. This project will provide method development and software engineering that is not covered by the Hub, and thus will complement EPSRC investment in the Hub. Specific systems include methanol synthesis using homogeneous ruthenium complexes, Cu-based artificial enzymes for enantioselective Friedel-Crafts reactions, fluorophosphite-modified rhodium systems for hydroformylation catalysis of alkenes, and non-canonical substitutions in non-heme iron enzymes for C-H functionalisations. These highly topical and potentially industrially relevant systems will allow us both to test and exploit the new software, which promises a step change in our ability to model catalytic systems and reactions.
Find out more at EP/W014378/1 and chemshell.org.
BEORHN: Bacterial Enzymatic Oxidation of Reactive Hydroxylamine in Nitrification via Combined Structural Biology and Molecular Simulation
The nitrogen cycle is critical to the environment and global health. The majority of nitrogen used in modern agriculture comes from artificial fertiliser comprised primarily of ammonia or ammonium compounds. This is converted into nitrogen-containing chemicals that are useful to plants (e.g. nitrate) by the action of nitrifying bacteria in soils and water and is then returned to nitrogen gas in the atmosphere through further bacterial action. Losses or imbalances in these processes lead to the release of the pollutant and greenhouse gas nitrous oxide (N2O), the pollutant nitric oxide (NO), the toxic intermediate hydroxylamine (NH2OH), or nitrites/nitrates into freshwater, resulting in algal blooms. Understanding the nitrification process is therefore critically important for agriculture, food security, the environment and human health.
In the nitrification process, the second step involves the oxidation of the reactive compound hydroxylamine, catalysed by metal-containing proteins which contain a highly unusual iron-heme structure where the heme contains an additional bond or ‘cross-link’ to the protein. Two families of structurally very different proteins, hydroxylamine oxidoreductase (HAO) and cytochrome P460 (CytP460), carry out this chemical reaction to yield different reaction products (NO for HAO and N2O for CytP460). Each functional unit of HAO contains seven iron-heme units that function to transfer or ‘shuttle’ electrons and one P460 heme unit where the heme is further modified via cross-linking to a tyrosine amino acid residue and where the oxidation of hydroxylamine occurs. In CytP460s each functional unit contains one catalytic P460 unit but, in this case, cross linked to a different kind of amino acid (lysine). Furthermore, to add to the complexity, within the CytP460 family, the two proteins so far identified in different families of bacteria (N. europaea and M. capsulatus), have different heme environments despite carrying out exactly the same chemical reaction.
Our project addresses this poorly understood second step in the nitrification process, namely the catalytic oxidation of hydroxylamine by HAO and CytP460. We will target these protein systems by combining integrated spectroscopic and structural biology approaches and computational chemistry using high performance computing. We will use X-ray crystallography with near-simultaneous measurement of spectroscopic data of the same crystal to assign correct electronic states to the enzyme’s active site. We will use thousands of very small (micro)crystals to obtain structures of enzymes at room temperature and to produce structural movies of the enzymes in action (more traditional techniques produce an average structure more similar to a single movie frame). These spectroscopic and structural data will be combined with state-of-the-art computational methods (molecular dynamics and recently developed quantum mechanics/molecular mechanics approaches) to better understand at the atomic level how these enzymes work. Linking experiments and simulations in this way, we will obtain a fundamental understanding of the function of these enzymes, and why the reactions they catalyse result in different products. Our ultimate goal is to design new, mutated enzymes, using our knowledge of how their structure affects the reactions they catalyse, to change their products from NO to N2O and vice versa, so demonstrating the potential for control of catalysis in future biotechnological applications.
Find out more at BB/V016660/1.
Quantum Enhanced Computing Platform for Pharmaceutical R&D – QuPharma
The covid 19 pandemic underlined the importance of quick and efficient development of drugs and vaccines, that are safe to deploy for wider use. Despite impressive developments during the pandemic, drug discovery remains a very long and expensive with very low probability of success. Identifying useful substances with suitable properties for specific diseases is very difficult task, even for the most powerful supercomputer. More than half of the few drugs that enter the phase of human trials, do not get approval for commercial use, with all the effort related to that going to waste.
Quantum computing is a new type of supercomputer that works differently than current computers. Based on the exotic properties of quantum mechanics, they will be able to solve very complicated problems, that are currently unsolvable in a short amount of time; modelling the properties of drugs is one of these problems. Quantum computers can help scientists select and study more and better substances in order to deliver faster more efficient drugs for the benefit of all.
In this project, we will develop a quantum computer and use it alongside a classical supercomputer to solve problems that are of real value to the pharmaceutical companies. STFC is providing the expertise to integrate quantum and classical computing, and use it for realistic chemical simulations using embedding methods we are developing in the ChemShell multiscale modelling package. Working with our colleagues at the Hartree Centre and NQCC and our partners SeeQC, Riverlane, the University of Oxford, the Medicines Discovery Catapult and Merck, we will identify some of the pain points of the drug discovery process where quantum computers can help. The UK is a world leader in the pharmaceutical sector and a pioneer in developing the quantum technology industry. This project is of real national value as it will boost the development of quantum computers, while showing how useful they can be in solving major proble
Find out more: GtR.

The Theme encompasses the scientific, software and data expertise underpinning the Theoretical and Computational Physics and Chemistry elements of the CoSeC, ALC and PSDI Programmes. In addition, we also lead or co-lead scientific and software research activities across the wider area of Computational Materials and Molecular Science with UK and overseas partners in academia and industry. Our research activities span a very wide remit ranging from Atomic, Molecular and Optical Physics to Materials Science, Materials Informatics, Solid State Physics, Magnetism and Light-Matter interactions; from Chemistry, Electrochemistry, Photochemistry and Biochemistry to Chemical Engineering and Processing. We are a team of 40+ computational and software scientists with a unique set of complementary skills ranging from classical and multiscale modelling to density functional theory (DFT), time-dependent DFT, and beyond-DFT Green’s function and wavefunction methods; from empirical, perturbative or real-time approaches to the latest artificial intelligence techniques. We develop and apply our solutions using various hardware from traditional high performance computing to cutting edge heterogeneous and quantum architectures. As fundamental research becomes increasingly cross-disciplinary and scientific computing more heterogenous, we are in a remarkably privileged position to continue pushing the forefront of Computational Materials and Molecular Science to the benefit of STFC and our partners.
Dr Gilberto Teobaldi, Theme Lead, Computational Materials and Molecular Sciences