📣 Help Shape the Future of UKRI's Gateway to Research (GtR)

We're improving UKRI's Gateway to Research and are seeking your input! If you would be interested in being interviewed about the improvements we're making and to have your say about how we can make GtR more user-friendly, impactful, and effective for the Research and Innovation community, please email gateway@ukri.org.

High Throughput Modelling of Molecular Crystals Out of Equilibrium (ht-MATTER)

Lead Research Organisation: UNIVERSITY COLLEGE LONDON
Department Name: Chemical Engineering

Abstract

Molecular crystals (MCs) are found in many everyday products, from food to pharmaceuticals, hiding in plain sight characteristics that make them ideal materials for cutting-edge technological applications. Like construction bricks, MCs' building blocks can reversibly self-assemble into a plurality of structures, thus distinguishing the chemical properties of building blocks from the physical and mechanical material properties. These characteristics open up endless possibilities in material design, with applications in pharmaceutical manufacturing, separations, catalysis, and organic electronics. Tailoring the composition of the liquid phase in which MCs assemble holds the key to designing processes able to yield materials with desired properties. However, current approaches at MCs computational design are centred on predicting the thermodynamic stability of bulk phases. This paradigm had remained essentially unchanged since its inception more than two decades ago, leaving material and process design practices to empiricism. By neglecting the role of assembly kinetics, current computational crystal structure prediction methods cannot identify attainable MC structures and the ideal conditions (i.e. solvent, composition, temperature) to obtain them.
With ht-MATTER, I will bridge this gap by developing an open, transparent and flexible molecular simulation platform that will deploy state-of-the advanced molecular simulation methods necessary to model the out-of-equilibrium processes that govern crystal precipitation from solution at the atomistic scale. ht-MATTER will catalyse a paradigm shift in computational materials design by providing the high-throughput framework necessary to identify: a) MC structures attainable at finite-temperature; b) kinetic bottlenecks associated with crystal nucleation and growth from solution; and c) their dependence on solvent choice and solute concentration.

Publications

10 25 50
 
Description During the second year of this award we have obtained substantial progress on three fronts:

1. We have demonstrated the applicability of machine learning techniques combined with statistical mechanics approaches to efficiently estimate the stability of polymorphs of molecular crystals. This result, represents a substantial step forward in increasing the efficiency and scalability of computational approaches for the design of novel crystalline materials, with applications in pharmaceutical manufacturing and functional materials design.

2. We have developed and tested an approach to obtain information on the stability of crystalline nuclei within a solution from equilibrium molecular dynamics simulations. This work has enabled us to make a substantial step towards developing polymorph ranking methods that include information on the kinetics on nucleation, rather than relying solely on thermodynamic information.

3. We have further developed our ability to compute the free energy landscape associated with physical-chemical transformations with multiple, independent simulations, improving the computational efficiency of enhanced sampling methods. Moreover we have gained a deeper understanding of the impact of machine-learned reaction coordinates on the interpretation of free energy landscapes. These are technical results that form the basis for substantially increasing our ability to obtain information on nucleation mechanisms from data.

4. We have developed a systematic assessment of what a fully digital workflow for the development of pharmaceutical crystallization processes would entail. As a part of an international team of scientists and engineers, and in collaboration with the Pharmaceutical industry, we have laid out the key steps to obtain a conceptual design of a crystallizer starting only from the molecular structure of an Active Pharmaceutical Ingredient. This exercise, has enabled to identify key strengths and weaknesses of current computational methods, highlighting priority areas that are strongly aligned with this award.
Exploitation Route This year' outcomes provide some key methodological advances for a systematic adoption of digital technologies in the development of molecular crystalline materials.
The advances on the estimate of thermodynamic stability both based on machine learning methods, and on classical thermodynamics is general and accompanied by open research code, that can be used by others in the field and that, within the context of this award, will form the basis to develop a high throughput platform for the simulation of molecular crystals.
Sectors Chemicals

Digital/Communication/Information Technologies (including Software)

Healthcare

Manufacturing

including Industrial Biotechology

Pharmaceuticals and Medical Biotechnology

URL https://pubs.acs.org/doi/full/10.1021/acs.cgd.3c01390
 
Description International Scholar Award for Doctoral Training (ISAD)
Amount £80,000 (GBP)
Organisation University College London 
Sector Academic/University
Country United Kingdom
Start 09/2024 
End 10/2027
 
Title CmuMD for PLUMED2 
Description This repository hosts the implementation for PLUMED 2 of the CmuMD method. This method allows for mimicking open boundary conditions in NVT molecular simulations involving solid/fluid interfaces and enables the simulation of concentration-driven fluxes through porous membranes. The method, with an application to crystal growth, is described in the paper: [1] Molecular dynamics simulations of solutions at constant chemical potential. C Perego, M Salvalaglio, M Parrinello, J Chem Phys (14), 144113, 2015. The adaptation of CmuMD to model concentration-driven membrane fluxes is described in the paper: [2] Concentration gradient driven molecular dynamics: a new method for simulations of membrane permeation and separation A Ozcan, C Perego, M Salvalaglio, M Parrinello, O Yazaydin Chem Sci 8 (5), 3858-3865, 2017 The method is described and discussed in the review work: Non-Equilibrium Modelling of Concentration-Driven Processes with Constant Chemical Potential Molecular Dynamics Simulations T Karmakar, A Finney, M Salvalaglio, AO Yazaydin, C Perego, Accounts of Chemical Research, 2023. 
Type Of Material Computer model/algorithm 
Year Produced 2022 
Provided To Others? Yes  
Impact Reproducing the published work and enabling the adoption of methods in independently run research projects. 
URL https://github.com/mme-ucl/CmuMD
 
Title Mean Force Integration - code 
Description Repository that contains the codebase used to implement Mean Force Integration (MFI), a method for postprocessing free energy calculations that enables the calculation of free energy surface from metadynamics and their associated convergence. 
Type Of Material Computer model/algorithm 
Year Produced 2023 
Provided To Others? Yes  
Impact Enable to reproduction of results in published papers and dissemination of the methods developed in Bjola and Salvalaglio (10.26434/chemrxiv-2024-83h5q 2023). 
URL https://github.com/mme-ucl/MFI
 
Title NNucleate - Machine Learning of Nucleation Collective Variables 
Description This is the release version of NNucleate. NNucleate is a Python package developed for the training of GNN-based models approximating computationally expensive collective variables (CV). The primary intended application is enhanced sampling simulations of nucleation events. These types of simulations are typically limited by the computational cost of their CVs, since they require a computationally expensive, differentiable degree of order in the system to be calculated on the fly. 
Type Of Material Computer model/algorithm 
Year Produced 2023 
Provided To Others? Yes  
Impact Enables the reproducibility of the published results and the dissemination and adoption of the methods developed. 
URL https://github.com/mme-ucl/NNucleate
 
Title NaCl (aq) nucleation rates: computational results and experimental measurements 
Description A repository that hosts the data used to map computational predictions and experimental measurements of nucleation rates of NaCl(aq) as discussed in "Molecular simulation approaches to study crystal nucleation from solutions: Theoretical considerations and computational challenges, A. Finney, M. Salvalaglio, WIREs Computational Molecular Science, 2024." ( https://doi.org/10.1002/wcms.1697) 
Type Of Material Database/Collection of data 
Year Produced 2023 
Provided To Others? Yes  
Impact Enable a critical comparison between computational methods for the estimate of nucleation rates and their comparison with experiments. Will be periodically updated with data from new papers. 
URL https://github.com/mme-ucl/NaCl_water_Nucleation_Rates
 
Description Crystallisation Day 2024 Industry-Academia panel 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Participated in a panel discussion on Industry and academia open questions and challenges in the field of crystallization processes.
Year(s) Of Engagement Activity 2024
 
Description Presenter and Panelist at the 2025 University of Oklahoma Sustainability Forum 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact I have presented my work on the digitalisation of materials research in Pharma, in the context of the University of Oklahoma sustainability forum. I was representing academic research in a panel composed of industrialists and social scientists that discussed several aspects of sustainable practices within the Pharmaceutical industry.
Year(s) Of Engagement Activity 2025
URL https://www.ou.edu/coe/events/sustainability
 
Description Talk at 15th CFC meeting - CCDC Cambridge 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Delivered a research talk to an audience of industry professionals during the 15th CFC meeting organised by the CCDC (Cambridge).
Year(s) Of Engagement Activity 2024