Skip to content
What's on

Nucleation: past and future challenges for experiment, theory and simulation

Scientific meeting


Kavli Royal Society Centre, Chicheley Hall, Newport Pagnell, Buckinghamshire, MK16 9JJ


Theo Murphy international scientific meeting organised by Professor Angelos Michaelides, Professor Daan Frenkel ForMemRS, Professor Fiona Meldrum and Dr Gabriele Cesare Sosso.

Molecular simulations of heterogeneous ice nucleation. Credit: Gabriele Sosso and Angelos Michaelides, UCL

Nucleation is a fundamental process crucial to a surprising array of different technological and everyday phenomena, from drug design to the formation of clouds.

This meeting brought together the key players in the field, taking stock of the latest developments in experiment and simulation as well as addressing the most pressing challenges to be faced in the near future.

Speaker biographies and abstracts are available below.

Attending this event

This meeting has taken place. Recorded audio of the presentations will be added shortly.

Enquiries: contact the scientific programmes team

Schedule of talks

05 September

Session 1 09:00-12:35

Crystallisation from solution

4 talks Show detail Hide detail


Dr Gabriele Cesare Sosso, UCL, UK

09:05-09:50 The structural and energetic sources of hierarchical nucleation pathways

Professor Jim de Yoreo, Pacific Northwest National Laboratory; University of Washington, USA


Field and laboratory observations show that crystals commonly form by the attachment of particles that range from multi-ion complexes to fully formed nanoparticles. These hierarchical pathways are diverse, in contrast to those of classical models that consider only the addition of monomeric chemical species. Despite their complexity, a holistic framework for understanding particle-based pathways to crystallisation that extends classical concepts emerges when the coupled effects of complexity of free energy landscapes and the impact of dynamical factors that govern particle formation and interaction are considered. Here I describe that framework and use a series of in situ TEM and AFM studies on inorganic, organic, and macromolecular systems to illustrate the evolution in nucleation and growth processes as these complexities and dynamical factors come into play. The introduction of either size-dependent phase stability associated with the high surface-to-volume ratios of nanoparticles, or high driving force coupled with the existence of metastable polymorphs leads to true two-step pathways characterized by the initial appearance of a bulk precursor phase. The creation of micro-states, which represent local minima in free energy stabilised by configurational factors, can also lead to hierarchical pathways, but the intermediates are transient states that do not appear on a bulk phase diagram. However, small changes in molecular structure can eliminate these transient states, leading to a direct pathway of nucleation. In either of these cases, reduction in molecular mobility, either through reduced temperature or introduction of ion-binding macromolecules, can freeze non-equilibrium states into place for dynamical reasons.

Show speakers

09:50-10:35 Solute precipitation nucleation: advances in theory and simulation methods

Professor Baron Peters, University of California, Santa Barbara, USA


Nucleation is the stochastic process that creates the first stable embryo of a new phase to initiate a phase transition. The mechanism is poorly understood because rare event processes like nucleation present special challenges to both experiments and simulations. The difficulties are particularly acute for multi-component condensed phase nucleation processes. New rare events approaches that reveal how thermodynamic and dynamical factors in nucleation will be presented. Particular emphasis will be given to the factors that influence polymorph selection including disparate growth rates of competing polymorphs, and the effects of additives on the free energy landscape. Finally, we will examine assumptions about interfacial free energies as a possible origin for continuing discrepancies between experimental and theoretical nucleation rates.

Show speakers

11:05-11:50 In situ and time resolved challenges for nucleation and growth of minerals from solution

Professor Liane G. Benning, GFZ German Research Centre for Geosciences, Germany; University of Leeds, UK


The formation of mineral phases control most biogeochemical element cycles on Earth, yet surprisingly, the molecular level reactions that lead to bonds between atoms in minerals to be made or broken - essentially to nucleate, grow, transform or dissolve a mineral phase - are hard to quantify. In the last decade our ability to follow such complex and highly dynamic reactions has massively improved, and novel in situ, time resolved and high-resolution methods now allow us to follow such reactions better. However, we still lack much of the fundamental knowledge about how ions are assembled to form a first nanoparticle both from the theoretical and experimental sense and thus the validations of reaction kinetics and mechanisms that have been inferred from observations of natural processes is still missing in many important Earth systems. I will show how mineral formation and transformations in carbonates [1] and sulphate [2-3] occurs, and what knowledge extracted from in situ, time resolved and high-resolution methods can teach us about nutrient cycling, primary productivity and mineral chemistry in modern or ancient geological settings.

[1] Bots et al (2012) Crystal Growth and Design doi: 10.1021/cg300676b
[2] Stawski et al (2016) title. Nature Communications. Doi: 10.1038/ncomms11177
[3] Van Driessche et al (2012) Science. doi: 10.1126/science.1215648

Show speakers

11:50-12:35 Multiscale modelling of polymer-controlled crystallisation of calcium rich minerals

Dr Davide Donadio, University of California Davis, USA


Crystallisation of calcium rich minerals, such as calcium carbonate, calcium oxalate and hydroxyapatite, affects several forms of life, from plants and animals to human beings. These minerals are fundamental constituents of shells and bones, but are also associated to pathology. Understanding and controlling crystallisation of these materials at a molecular level in biological environment would allow medical scientists to design countermeasures to prevent common diseases, such as urinary tract stones and hydroxyapatite crystal deposition.

Mineralisation in solution, assisted or inhibited by electrolytes, is a process that entails a broad range of size and time scales. To gain molecular insight by computer simulations thus requires carefully designed multiscale modelling techniques, and data analysis tools.

Here we consider the case of calcium oxalate, the main component of kidney stones, and we investigate its nucleation and growth in aqueous solution, in the presence of polypeptides, e.g. poly-glutamate, that have been shown to slow down nucleation, and to select mineral structure and morphology. With a combination of classical and ab initio molecular dynamics, and advanced sampling techniques, we unravel the interaction of poly-glutamate with ions in solution and at the surface of crystalline calcium oxalate.

Furthermore, we introduce a Hamiltonian adaptive resolution scheme (H-AdResS), which allows one to concurrently treat a liquid system at different levels of details at controlled thermodynamic conditions, thus making it possible to simulate directly the nucleation and growth of minerals in solution.

Show speakers

Session 2 13:30-18:15

Ice nucleation

5 talks Show detail Hide detail


Professor Angelos Michaelides, University College London, UK

13:30-14:15 Nucleation processes in aerosols and clouds

Professor Thomas Koop, Bielefeld University, Germany


Atmospheric clouds strongly impact the weather and climate of our planet. Their properties are dependent upon how they form from preexisting aerosol particles by way of various nucleation pathways. The basics of cloud formation are well understood, but quantitative predictions of cloud properties and precipitation remains challenging, in part because of the diversity and variable abundance of aerosol particles. Here, we will focus on cloud processes that include ice nucleation. The formation of ice crystals is a key step during the initiation of atmospheric precipitation, and ice nucleation normally is the rate-limiting mechanism for ice particle formation. Homogeneous ice nucleation is prevalent for some atmospheric conditions, but often heterogeneous ice nucleation is the dominant pathway. Such heterogeneous ice nucleation is triggered by ice-nucleating particles, which traditionally were thought to consist primarily of mineral dust particles. But more recently, a wide variety of non-crystalline ice nucleators have been identified such as amorphous solids, pollen and bacteria, surfactant monolayers, and even dissolved molecules or molecular clusters. The presentation will discuss homogeneous and heterogeneous ice nucleation processes, covering both theoretical aspects as well as experimental developments, with a focus on the fundamental physical and chemical aspects of the involved processes.

Show speakers

14:15-15:00 A state between liquid and crystal: locally crystalline but with the structure factor of a liquid

Dr Richard Sear, University of Surrey, UK


Using computer simulations of a simple model, we study a non-equilibrium state with a liquid-like S(k) but where over 95% of the molecules are in locally crystalline environments. Due to its liquid-like S(k) and slow dynamics, the state is apparently amorphous, although it contains nanocrystalline order. States such as this will have properties that are determined by local structure (possibly including electrical conductivity) that are crystal-like, despite scattering X-rays like liquids. Single crystals of two different polymorphs can form from this state. Which polymorph forms is determined long before the structure factor changes significantly. [Mithen & Sear, CGD 16, p3049 (2016)]

Show speakers

15:30-16:15 Using synthetic macromolecules to mimic (and understand) antifreeze proteins

Dr Matthew I. Gibson, University of Warwick, UK


With an ever-ageing population in the Western world, the need for regenerative medicine, especially transplantation is increasing, but the methods to cryopreserve (and hence distribute) these cells and tissue still relies on addition of high concentrations of organic solvents. We seek to create new macromolecular solutions to address this problem.

Antifreeze (glyco) proteins are found in many species adapted to live in the polar regions or at high altitudes, in sub-zero temperatures. These act to slow ice growth and also suppress the freezing point. Ice nucleating proteins are also found in many species, including Psuedomonus syringae, which promote frost formation on plants as a feeding mechanism. The exact mechanism (or mechanisms) of action of these proteins is still not clear. The relationships between the different macroscopic properties, such as ice shaping and ice growth inhibition are also not clear. To probe these questions, and to enable translation into applications, we have developed a range of synthetic polymer materials which can reproduce the properties of both antifreeze proteins and ice nucleating proteins, but by using simplified structures, and scalable synthetic methods. We have also developed ‘small molecule’ ice growth inhibitors which also raise question about the exact mechanisms of action.

Guided by the above, we have applied these polymers to enhance cellular cryopreservation and enable solvent-freeze cryopreservation in some cases.

Show speakers

16:15-17:00 Role of stacking disorder in ice nucleation

Professor Valeria Molinero, The University of Utah, USA


Freezing of water is central to the processes that determine Earth's climate. Accurate predictions of changes in weather and climate therefore hinge on good predictions of the rate of ice nucleation. Such rate estimates are based on extrapolations using classical nucleation theory (CNT), which assumes that the structure of nanometer-sized ice nuclei corresponds to that of bulk hexagonal ice, the thermodynamically stable form of ice. Recent simulations with various water models show that ice nucleated and grown under atmospheric temperatures is stacking disordered at all sizes; i.e., it consists of random sequences of cubic and hexagonal ice layers. This implies that either stacking disordered nuclei are more stable than hexagonal ice nuclei, or that they form because of non-equilibrium dynamical effects. Both scenarios challenge central tenets of Classical Nucleation Theory (CNT) and its validity for analysing and predicting ice nucleation rates. Here we use rare events sampling and free energy calculations with the mW water model to show that stacking disordered ice is the stable phase for nuclei with less than at least ~50,000 molecules. Stacking disordered critical nuclei at 230 K are 14.3±0.5 kJmol-1 more stable than hexagonal nuclei, favored by the entropy of mixing of cubic and hexagonal layers. This results in over three orders of magnitude higher nucleation rates with respect to CNT predictions. We find that the correction to CNT nucleation rates is significant over the whole range of temperatures relevant to homogeneous nucleation, and the most pronounced at the warmest conditions. This should have a strong impact on climate models, which are very sensitive to the parameterisation of ice nucleation rates. We conclude that CNT must be corrected for the dependence of the crystallisation driving force on nucleus size when interpreting and extrapolating ice nucleation rates from experimental laboratory conditions to temperatures important to clouds.

Show speakers

17:00-18:15 Poster session

06 September

Session 3 09:00-12:30

Simulation methods and experimental techniques

4 talks Show detail Hide detail


Professor Fiona Meldrum, University of Leeds, UK

09:00-09:45 Clusters, complexes and primary particles in the early stages of bioinspired mineral growth

Professor Nico Sommerdijk, Technische Universiteit Eindhoven, The Netherlands


Biominerals possess shapes, structures and properties not found in synthetic minerals. The dream of exploiting the biological principles of controlled mineral formation in materials chemistry inspires a large community to investigate the underlying mechanisms through biomimetic mineralisation experiments.[1] The defining characteristics of biominerals arise from the interplay of the mineral with a macromolecular matrix, which directs crystal nucleation and growth. Within this three dimensional biomolecular assembly, the developing mineral interacts with acidic macromolecules, either dissolved in the crystallisation medium or associated with an insoluble templating structure.[2]

CryoTEM has proven to be a powerful tool to investigate – with great detail – the nucleation and growth of different mineral systems, including calcium carbonate[3], calcium phosphate,[4] iron oxide[5] and silica.[6] It also allows us to study how organic-inorganic interactions at interfaces affect the crystallisation from solution.[7-9] Interestingly we find that all these pathways involve nanometer sized building blocks that have been termed prenucleation clusters,[7-9] prenucleation complexes[4] and primary particles.[5-6] Using time resolved cryoTEM we now also demonstrate how multistep nucleation pathways are altered through the influence of polypeptide based additives.[10] Controlling the pathways of nucleation and growth may help us to ultimately to control the size, shape and orientation of the crystals and optimise them for specific technological applications.[11]

[1] Nudelman and Sommerdijk, Angewandte Chemie-International Edition 2012, 51, 6582.
[2] Chem. Rev. 2008, 108, 4329.
[3] Pouget, et al., J. Am. Chem. Soc. 2010, 132, 11560.
[4] Habraken, et al., Nat. Commun. 2013, 4.
[5] Baumgartner, et al., Nat. Mater. 2013, 12, 310.
[6] Carcouet, et al., Nano Lett. 2014, 14, 1433.
[7] Pouget, et al., Science 2009, 323, 1455.
[8] Dey, et al., Nat. Mater. 2010, 9, 1010.
[9] Nudelman, et al., Nat. Mater. 2010, 9, 1004.
[10] Dey, et al., Faraday Discuss. 2015, 179, 215.
[11] De Yoreo, et al., Science 2015, 349, 498.

Show speakers

09:45-10:30 Challenges for the simulation of nucleation pathways for biominerals

Professor Julian Gale, Curtin University, Australia


The nucleation of biominerals, such as calcium carbonate and calcium phosphate, has attracted significant attention in the last decade following the proposal of so-called 'non-classical' mechanisms. For the case of calcium carbonate it has been proposed that the pathway to formation under homogeneous conditions can include stable pre-nucleation clusters (PNCs) [1], liquid-liquid phase separation [2] and the formation of amorphous calcium carbonate as a thermodynamically stable phase for nanoparticles in a given size range [3]. Given the difficulties of probing such species in situ with experimental techniques, atomistic simulation is a valuable complementary approach. However, simulation faces a number of challenges of it’s own due to the limitations of molecular dynamics with the currently available computing power. For example, the concentrations of ions at saturation for most biominerals are below 1 mM, meaning that unbiased sampling of ion association in aqueous solution at experimental conditions is unfeasible. In this presentation several challenges for simulation of ion association from ion pairing, through pre-nucleation species, to determining critical nucleus size will be examined. This includes issues spanning the underlying accuracy of the free energy landscape through to how to map cluster stability in terms of a manageable set of collective variables that capture the association/dissociation pathways and connect them to identifiable thermodynamic states. Here examples will be drawn from aqueous calcium systems with carbonate, phosphate and oxalate.

[1] D. Gebauer et al, Science, 322, 1819 (2008)
[2] A.F. Wallace et al, Science, 341, 885 (2013)
[3] P. Raiteri & J.D. Gale, J. Am. Chem. Soc., 132, 17623 (2010)

Show speakers

11:00-11:45 Nucleation and growth processes of gas clathrate hydrates

Professor Carolyn Ann Koh, Colorado School of Mines, USA


Gas clathrate hydrates are crystalline inclusion compounds comprised of a three-dimensional network of hydrogen-bonded water molecules that can trap small gas molecules in the water cavities. The ability to control clathrate hydrate nucleation and growth processes is important in several energy applications, including during the production and transportation of oil/gas in subsea flowlines where gas hydrates can form blockages in the flowline, as well as energy storage of fuels in gas hydrate crystals. The nucleation and growth processes and inter-particle interactions of gas hydrate crystals on gas bubbles and water droplets in water and oil continuous systems are examined at high pressure and low temperature conditions. Addition of polymers and surface-active molecules can be used to modify these processes, e.g. delaying the nucleation and growth processes, or reducing the inter-particle interactions. The clathrate hydrate formation synthesis pathways can be a key strategy to designing higher storage capacity materials, and/or stabilising new stable and metastable crystal structures. Structure metastability has been observed through spectroscopic and computational studies. Examples of the use of different promoter guest molecules, synthesis methods, and pressure conditions are presented for the production of stable and metastable clathrate hydrate phases. These studies can help further our knowledge for developing clathrate materials for storage and other technologies.

Show speakers

11:45-12:30 Free energy landscape and molecular pathways of ice and gas hydrate nucleation

Professor Tianshu Li, George Washington University, USA


Gas hydrate and ice are similar in many aspects: both appear 'ice-like', and both form in aqueous environments as a result of ordering of water molecules. However the nucleation processes of two solids could proceed with very different pathways. Employing advanced molecular simulation methods, this talk explores the free energy landscape and molecular pathways of both ice and hydrate nucleation. This talk shows that ice nucleation, both homogeneous and heterogeneous, appears to follow a pathway described very well by classical nucleation theory (CNT). The nucleation of hydrate, on the other hand, has been often found to involve multiple steps, thus appearing non-classical. Indeed, structural analysis show that on average, hydrate formation is facilitated by a 'two-step' like mechanism involving a gradual transition from amorphous to crystalline structure. However analysis also shows the existence of direct nucleation pathways where hydrate crystallises without going through the amorphous stage. Interestingly, the calculated free energy profile was also found to fit reasonably well against CNT. The structural diversity and the CNT-like free energy profile imply that hydrate nucleation could be an entropically driven, kinetic process that proceeds via multiple pathways that have similar free energy profiles.

Show speakers

Session 4 13:30-17:00

Colloids and biomolecules

4 talks Show detail Hide detail


Professor Daan Frenkel ForMemRS, University of Cambridge, UK

13:30-14:15 Physical determinants of amyloid nucleation

Dr Anđela Šarić, University of Cambridge; UCL, UK


The assembly of normally soluble proteins into large fibrils, known as amyloid aggregation, is associated with a range of pathologies, including Alzheimer's and Parkinson's diseases. Computer simulations, in combination with quantitative experiments, can provide valuable insights into the mechanisms of amyloid formation, helping to bridge experimental scales with microscopic mechanisms.

Substantial evidence shows that disordered pre-amyloid oligomers − not the fully grown amyloid fibrils − are cytotoxic and involved in pathological processes. Yet, their relationship to fibrils is not well understood. Using computer simulations, we showed that at physiological conditions, disordered oligomers serve as nucleation centres for fibrils, and are crucial on-pathway species to amyloid formation, governing its kinetics [1].

Moreover, recent experiments have revealed that amyloid fibrils are able to catalyse formation of their copies from soluble peptides. By combining simulations with biosensing and kinetic measurements of the aggregation of Alzheimer’s Aβ peptide, we proposed a mechanistic explanation for the self-replication of protein fibrils. We find that the process is dominated by a single physical determinant − the adsorption of monomeric proteins onto the surface of fibrils [2]. Such mechanistic understanding not only has implications for future efforts to control pathological protein aggregation, but is also of interest for the rational assembly of nanomaterials, where achieving self-replication is one of the unfulfilled goals.

[1] A. Šarić, Y. C. Chebaro, T. P. J. Knowles and D. Frenkel, PNAS 111, 17869 (2014)
[2] A. Šarić, et al., Nat. Phys., in press (2016).

Show speakers

14:15-15:00 Kinetics of protein aggregation

Professor Tuomas Knowles, University of Cambridge, UK


Filamentous protein aggregation underlies a number of functional and pathological processes in nature. This talk focuses on the formation of amyloid fibrils, a class of beta-sheet rich protein filament. Such structures were initially discovered in the context of disease states where their uncontrolled formation impedes normal cellular function, but are now known to also possess numerous beneficial roles in organisms ranging from bacteria to humans. The formation of these structures commonly occurs through supra-molecular polymerisation following an initial primary nucleation step. In recent years it has become apparent that in addition to primary nucleation, in secondary nucleation events which are catalysed the presence of existing aggregates can play a significant role in the dynamics of such systems. This talk describes our efforts to understand the nature of the nucleation processes in protein aggregation as well as the dynamics of such systems and how these features connect to the biological roles that these structures can have in both health and disease.

Show speakers

15:30-16:15 Numerical simulation studies of ice nucleation at normal and extreme conditions

Dr Chantal Valeriani, Universidad Complutense de Madrid, Spain


Combining simulations of spherical crystal seeds embedded in the metastable fluid with classical nucleation theory (seeding technique), we are able 1) to successfully describe the nucleation rate in a wide range of metastability;  2) to estimate the crystal-fluid interfacial free energy, in good agreement with explicit direct calculations. 

Among all freezing transitions, that of water into ice is probably the most relevant to biology, physics, geology, or atmospheric science. 

Making use of the seeding technique, we evaluate the homogeneous ice nucleation rate for several water models (TIP4P/2005,TIP4P/ice, TIP4P and mW) for temperatures between 15 and 35 K below melting. The values agree (within statistical error) with experimental measurements and confirm that water freezing above 20 K below melting has necessarily to happen heterogeneously.

Estimating the ice-liquid interfacial free-energy, we conclude that 1) it depends on the chosen water model and is a key factor in the nucleation rate; 2) for all water models it decreases as the temperature decreases; 3) extrapolating our results of the interfacial free-energy to the melting temperature, we obtain a value between 25 and 32 mN/m, in reasonable agreement with experiments. 

We then establish the ice growth rate for TIP4P/ICE and mW water, and use the Avrami’s expression to estimate the crystallisation time. We find a crossover between a nucleation-controlled and a growth-controlled crystallisation regime, and argue that this could explain the apparent discrepancy observed among experimental values of the nucleation rate for temperatures below 230K.  

The avoidance of water freezing is the holy grail in the cryopreservation of biological samples, food and organs. Performing computer experiments to investigate ice nucleation at high pressures, we find a slowing down of the nucleation rate mainly due to the increase of the ice I-water interfacial free energy with pressure. 

Show speakers

16:15-17:00 Panel discussion and closing remarks

Nucleation: past and future challenges for experiment, theory and simulation Kavli Royal Society Centre, Chicheley Hall Newport Pagnell Buckinghamshire MK16 9JJ