welcome !


I am a Senior Postdoctoral Research Associate at the University of Bristol (UK), working on deeply supercooled liquids, vitrification, gelation and crystallisation. I am also interested in cross-disciplinary problems of data reduction and visualization.

The blog below collects (hopefully) plain-English explanations of my research, useful snippets and ideas for computational physics and soft matter.

Find more about my cv or my research interests.

Dynamical solid–liquid transition through oscillatory shear

During the summer of a few years ago, Eric Brillaux, a student of the Ecole Normale Superieure de Lyon, visited Bristol for a summer project. Thinking of something moderately ambitious that could (in theory) be achieved in a few months, we started to explore how simple crystals respond to oscillatory shear.

The original motivation of the work was rather speculative: I was wondering if it could be possible to transform mechanically an ordered system (a crystal) into an amorphous system (a glass) whilst preserving some degree of local order. To this purpose, the ideal model to consider was an atomistic binary mixture whose supercooled liquid state presents local structural motifs that are identical to the repeated units of its crystalline state (as we have previously shown, see here) .

In particular, we considered an oscillatory shear protocol. This is interesting not only because it represent more closely actual experimental methods, but also because it allows the system to behave either reversibly or irreversibly: if the oscillations are small, the crystal survives, if the deformations are large, amorphisation takes place.

In our article just out on Soft Matter, we discuss how this dynamical transition between the reversible and irreversible regime takes place in actual three dimensional crystals and how it depends on the crystal composition. For example, single-component crystals can transform structurally, from face-centred cubic structures to more body-centred cubic structures before becoming disordered. Instead, crystals of two components either transform reversibly or become amorphous at a critical oscillation amplitude.

The dynamics is also rather interesting: for instance, the growth of amorphous regions follows a coarsening pattern that is reminiscent of spinodal decomposition in non-driven systems.

In the end, the amorphous states that we obtain are not as rich in local structural motifs as I hoped, but the dynamical transition in itself has appeared to be very intriguing!

More details in the original article

Eric Brillaux, Francesco Turci, Soft Matter, 2019, doi 10.1039/C8SM01950A

Progress of the amorphisation of a binary crystal upon several oscillations of period τ0.

Morphometric approach to many-body correlations in hard spheres

Supercooled liquids become more and more viscous as their temperature is reduced. The increased viscosity corresponds to an enormous increase in the characteristic time for the relaxation of density fluctuations. What is often puzzling is that, differently from many other physical phenomena, this dramatic change in the correlations in time appears to be weakly reflected in conventional measures of spatial correlations. These are typically so-called pair or two-body correlations, measuring how likely it is to find randomly chosen pairs of particles at particular characteristic distances.

The lack of strong correlations between two-body spatial correlation and the emergent, enormous relaxation times of supercooled liquids suggests that more complex, eventually multi-body correlations may be at play.

Thanks to the work of a very gifted PhD student in Paddy Royall‘s group, Joshua F. Robinson, we have obtained a first theoretical insight on the origin of such emergent correlations in a reference model for supercooled liquids, i.e. hard spheres, which are often employed to understand the behaviour of colloidal particles and as a basis to develop approximate theories of liquids.

We rooted our work in a geometric approach to treat the free energy of thermal hard spheres developed by Roland Roth (a co-author of our work) termed morphometric theory and this has allowed us to study the free energy of a certain number of thermal structural motifs of hard spheres immersed in an effective medium and predict with a high degree of precision their respective populations. Furthermore, the approach that we have used has revealed that it is possible to follow local deformations of the motifs and compute the free energy barriers between them.

The work appeared as an Editor’s Suggestion in Physical Review Letters:

Joshua F. Robinson, Francesco Turci, Roland Roth, C. Patrick Royall, Phys. Rev. Lett. 122, 068004 (2019) doi: 10.1103/PhysRevLett.122.068004

and it has also been paired by a Viewpoint in Physics by Thomas Speck.

Coupling of sedimentation and liquid structure: Influence on hard sphere nucleation

Colloidal hard spheres at high volume fractions (beyond ~0.49) can crystallise: up to to 0.54, they coexist with a fluid phase, and at even higher densities they completely crystallise. The way the hard sphere fluid transforms into the solid is called crystal nucleation: due to thermal fluctuations, every now and then locally denser and more ordered regions appear and disappear; occasionally, these are large enough to further grow irreversibly, and form a crystalline region.

Nucleation is a generic process: in hard spheres, it should present its simplest traits, as it can be driven only by entropic forces. Yet, nucleation rates in colloidal hard spheres are rather different from what can be predicted from theory and numerical simulations. In particular, the discrepancy between the two increases of many orders of magnitude with decreasing the volume fraction from, for example, 0.55 to 0.53. Simulations normally consider idealised hard spheres in an ideal solvent. What could possibly go wrong?

Nick Wood, a talented PhD student here in Bristol, has taken care of some potential origins of the discrepancy analysing the effect of sedimentation on local order. Together with John Russo and Paddy Royall, we have considered in our recent paper how the flow induced by sedimentation may, via hydrodynamics interactions, transform the structure of the liquid, compared to the case in absence of sedimentation, and how such changes would impact on the nucleation barriers. The result is that some structural signatures clearly vary as a function of the density mismatch between colloids and solvent and that this leads to an estimated correction of the rates in the right direction, but by an amount that is not sufficent to address the entire discrepancy.

The complete article can be found here:
Nicholas Wood, John Russo, Francesco Turci and C. Patrick Royall J. Chem. Phys. 149, 204506 (2018); https://doi.org/10.1063/1.5050397

Structure of the colloidal hard-sphere fluid with (left) and without (right) sedimentation. In green are highlighted motifs that hinder the crystal formation and anti-correlate with it, as demonstrated here.

Local structure in deeply supercooled liquids exhibits growing lengthscales and dynamical correlations

Computer simulations are very powerful: in the case of molecular dynamics, we can model the positions and velocities of atoms or molecules and observe the emergence of pattern and structures in situ, following each individual atom in its trajectory.

However, when we study supercooled liquids or glasses, it is hard to probe in computer simulations very low temperatures or very tightly packed systems, unless we opt for indirect and clever routes to glean some information on the low temperature behaviour. It would be great if one could directly take a very cold (or, similarly, very dense) liquid at equilibrium and see how the constituent particles are arranged.

This is precisely what James Hallett has managed to do during his stay in Bristol using super-resolution microscopy: this method can access the coordinates of equilibrium themal packings so dense that a direct simulation would never do. Thanks to James’s clever imaging, we have then carefully analysed the individual coordinates and trajectories of dense repulsive colloids and managed to clearly show how the local environment of these densely packed equilibrium systems changes as the density is increased.

We have found some notable features: as we take denser samples, the liquid becomes gradually richer in regions where particles are arranged into five-fold symmetric structures; those regions display reduced mobility compared to other regions of the sample; randomly selected domains of the system become more and more “similar to each other” as the density is increased, accompanied by the decrease in the number of distinct configurations the liquid can take, a quantity related to its so-called configurational entropy.

This work has just appeared in Nature Communications. Full text here:

James E. Hallett, Francesco Turci, and C. Patrick Royall. Nature Communications 9(1), 3272 (2018).

Measuring the similarity between randomly sampled regions: (a) picking at random a blue and yellow group of particles from the (beautiful) experimental coordinates we can use methods based on singular value decomposition (SVD) to find the best-fitting rotation and translation. The overlap Q between blue and yellow particles is then calculated. In (b) (c) and (d) we see how the probability distribution of the overalap changes as (b) we change the volume fraction φ; (c) when we look at 5-fold symmetric regions (LFS-rich) at low φ and high φ. Only at high φ LFS-rich regions are significantly more similar.

The race to the bottom: approaching the ideal glass?

Together with CP Royall, S Tatsumi, J Russo and PhD student J Robinson we have just published on the Journal of Physics: Condensed Matter a handy review on recent approaches to the exploration of very stable glasses in experiments and simulations.

We cover a variety of topics, including vapor deposited glasses in experiments, importance sampling in trajectory space, random pinning, representing distinct attempts to address the following question: is glassiness linked to some kind of novel thermodynamic transition in very low temperature liquids?

Find out more here:

CP Royall, F Turci, S Tatsumi, J Russo, J Robinson, Journal of Physics: Condensed Matter, Volume 30, Number 36 (2018)

Structural-dynamical transition in the Wahnström mixture

Supercooled liquids present dynamical heterogeneities at low temperatures: on a certain length and timescale, some areas are very mobile (active) while others are much more solid-like (inactive). This feature is often interpreted as the signature of the fact that the liquid, when supercooled, starts exploring different metastable regions of the free energy landscape.

A possible route to illustrate this effect is through large deviations of structural-dynamical obserables, as we first did in the case of a canonical atomistic model for glassformers, the Kob-Andersen mixture. A main observation of that work was that dynamical heterogeneities correspond to a first order phase transition in a (reweighted) space of possible steady states between high energy trajectories that are rich in structure and low energy trajectories that are poor in structure. Moreover, such a transition has a strong temperature dependence, so that the structure-rich trajectories become more and more likely to be observed as the temperature decreases.

Now, we have published a follow-up work on the European Physical Journal E, where we show that the same mechanism is at play in another model glass-former (the Wahnström mixture), showing that while the overall qualitative picture may be general, the details depend on the nature of the interactions between the constituents. Moreover, we also show that configurations extracted from the structure-rich trajectories have much larger yield stresses than the normal supercooled liquid: the emerging rigidity of glasses appears to be strongly related to the structural-dynamical transition that we have highlighted.

More information in

F Turci, T Speck and C P Royall, Structural-dynamical transition in the Wahnström mixture Eur. Phys. J. E , 41: 54 (2018) https://doi.org/10.1140/epje/i2018-11662-3

A structure-rich configuration of the Wahnström binary mixture under shear: red areas are icosahedral motifs while flashing yellow particles are those that manifest the largest non-affine square displacement

Structural Covariance in the Hard Sphere Fluid

Hard spheres are one of the simplest model of fluid. When the density is increased, the motion is slowed down and non-trivial local arrangements of particles characteristic of the liquid emerge. We call these local structural motifs.

In hard spheres, special role is played by structures with pentagonal rings, such as icosahedra or pentagonal pyramids.

In collaboration with Pierre Ronceray (Princeton), to appear on the Journal of Chemical Physics, we have shown that if we study the cross-correlations between the variation of populations of different motifs we can understand more of the phase behaviour of this model liquid: for example, we can predict its propensity to form more motifs of a given type if an external field is applied; or, we can understand and classify different motifs that coexist in the liquid phase.

More details can be found in the JCP Editor’s Pick:

B. Carter, F. Turci, P. Ronceray, C.P. Royall, Structural Covariance in the Hard Sphere Fluid, The Journal of Chemical Physics 148, 204511 (2018); https://doi.org/10.1063/1.5024462

Filtering long sentences with regular expressions

It happens often to me that in order to make an article clearer or more incisive, I may need to identify very long sentences in a draft and break them down to smaller, simpler units.

In \LaTeX  editors, there may not be an option by default to identify long sentences. However, using regular expressions, it is possible to circumvent this issue.

In editors that allow to search for regular expressions (such as Sublime Text or Texpad or others) the following snippet would allow us to search for sentences with more that 20 words:


It is not too hard to break this regular expression down to its elementary constituents. Let us just recall a few ideas concerning regular expressions

  • remember that () enclose groups
  • | indicates the OR operation
  • \w+ corresponds to a series of one or more occurrences of an alphanumeric character (a word)
  • \s+ corresponds to a series of one or more occurrences of spaces
  • \. is the dot character
  • {number_1, number_2} looks for at least number_1 repetitions of the previous element (with at most number_2 repetitions)

Therefore, in plain language, the above regular expression is

(a word followed by a comma followed by some space) OR (a word followed by some space) REPEATED AT LEAST 20 TIMES (a word followed by a full stop OR a question mark OR an exclamation mark)

This clearly allows us to detect sentences that may be long, very long, very very long, at least as long as this very sentence!

Screen Shot 2018-03-31 at 11.43.49
An example of a match in Sublime Text. Notice that the regular expression button .* on the bottom left corner of the search field is pressed.

Experimental determination of configurational entropy in a two-dimensional liquid under random pinning

A glass is (broadly speaking) mostly composed of particles that very slowly move due to surrounding cages formed by the disordered structure of the neighbouring particles. A way to study the glass transition is to actually freeze-in a subset of the particles and observe how this induces changes to the slow relaxation and how this relates to the emergence of local order.

The freezing-in procedure (also called pinning) has a secondary important effect: the pinned liquid has a “simplified” configurational space, as many configurations become forbidden. The number of available configurations (and hence the configurational entropy of the liquid) is therefore reduced by the simple pinning procedure and if a thermodynamic origin of dynamical arrest is to be surmised, such a reduction would be a necessity.

Some years ago, Ian Williams designed a novel, clever way to pin large two-dimensional colloidal supercooled liquids. In an article recently accepted in the Journal of Physics: Condensed Matter we have shown that the technique allows us to observe the crossover from a free-flowing liquid to a pinned glass and that this is accompanied by very limited structural changes. However, if we map the configurational entropy of the experiments with the entropy measured from model numerical simulations, we do observe that accounting for the fraction of pinned particles leads to a reduction in the estimated configurational entropy.

Full reference:

I. Williams, F. Turci et al. (2018) Experimental determination of configurational entropy in a two-dimensional liquid under random pinning, J. Phys.: Condens. Matter in press https://doi.org/10.1088/1361-648X/aaa869

Segmenting 3d biological data

I have recently been given the opportunity to study the segmentation of 3D data.  The group of Dr. C. Hammond of the School of Physiology, Pharmacology and Neuroscience in Bristol studies malformation in tissues of  Zebrafish  a model organism which can be genetically manipulated relatively easily .

A major task is to identify bone malformation or osteoarthritis. Hammond’s group manages to image hundreds of Zebrafish in three dimensions so that bone structures can be visualised. Identifying bone deformations in the spine, for example, is key to associate them to specific genetic marker. To do so, a quantitative analysis of the structure of the individual vertebrae is necessary.

It turned out that it is possible to do this via image analysis techniques that are publicly available in Python: the key libraries that I employed are scipy. ndimage and scikit-image.  Identifying the vertebrae in 3d means to perform  a segmentation of volumes and surfaces in 3d images.

An example of the vertebrae, individually resolved, can be visualised in 3d here below: