Nucleation: from water vapor to droplet

Homogeneous vapor-to-liquid nucleation takes place when the water molecules in the gaseous phase undergoes condensation, resulting in the formation of nuclei of water droplets in the liquid phase.

This happens spontaneously and without interactions with other materials. Homogeneous nucleation takes place at random points that are randomly distributed in the volume.

This kind of nucleation does not occur often, because almost always there are possible points where water molecules interact with another material or surface.

The video shows how homogeneous vapor-to-liquid nucleation occurs, based on molecular dynamic simulation.

2nd meeting of the "Multiscale and Non-continuum flows" - Special Interest Group (SIG)

On September 27th, the second Special Interest Group (SIG) meeting was hosted at the University of Warwick.

This series of meetings aims at bringing together theoreticians with experimental groups who share interest in the area of multiscale and non-continuum flows.
Beside the pleasure of seeing again many members of the Micro & Nano Flows group, the meeting was a good opportunity to meet other researchers across the United Kingdom.

After a welcome and 2-minute introductions by all the participants, four main lectures were given.
Dr. Kislon Voitchovsky (Durham University) discussed the liquid behaviour at nanoscale interfaces both from the numerical and, even more interestingly, experimental standpoints.
Dr. James Sprittles (University of Warwick) showed how non-equilibrium effects of the vapor flows can (unexpectedly) play a major role on the dynamics of the collisions between microdrops and  on their impact/spreading on solid surfaces.
Dr. Sergey Karabasov and Dr. Ivan Korotkin  (Queen Mary University of London) presented an interesting hybrid model which smoothly combines molecular dynamics with the Landau-Lifshitz fluctuating hydrodynamics.

I did appreciate the stimulating and constructive environment the meeting has created, as well as the main message of encouraging researchers to reflect on the impact that their studies may have beyond the academic community and to engage with industrial partners in collaborative research.

Fast nanofluidics by travelling surface waves

During the past two decades, we see the fast development and wide applications of nanotechnologies, biological chips and lab-on-a-chip. Nanoscale transport governs the behaviour of a wide range of nanofluidic systems, but it remains less understood due to the enormous hydraulic resistance associated with the nano-confinement and the resulting minuscule flow rates in MEMS/NEMS. In addition, the huge surface-volume ratio (up to 106−109) signifcantly affects the mass and momentum transport in micro/nanoelements and makes this type of research challenging.

    Obviously, the challenge is to overcome the large surface and viscous forces that prevent the fluid from flowing at the nanoscale, wherein other driving forces can be ignored in most microfluidic systems. In microfluidic biochip engineering, a driving force for driving microscale fluid motion has been introduced by employing the surface waves. The key of this novel technology is to make a micropump that is able to position reagents on the surface of chips or in microfluidic channels without the mechanical contact. This is implemented in terms of the surface acoustic waves (SAW) that are induced using radio frequency electric signals. These waves arise through the use of piezoelectric substrate materials in the chip. It is interesting that the SAW-induced effect has similarity in live nature, for example with the skin features of fast-swimming sea animals, such as dolphins. Dolphins use travelling waves on their skin surface to damp the turbulence in the boundary layers near the skin surface.

    Our purpose is to present that the travelling surface waves propagating on the walls of nanochannels can offer a powerful method for inducing a host of extremely fast nanofluidic flow. We find that the flow rate is enlarged by increasing the amplitude of travelling surface waves and can be up to a sevenfold increase. However, the flow rate is only enhanced in the limited range of frequency of travelling surface waves such as low frequencies, and a maximum fivefold increase in flow rate is pronounced. In addition, the fluid-wall interaction (surface wettability) plays an important role in the nanoscale transport phenomena, and the flow rate is signifcantly increased under a strong fluid-wall interaction (hydrophilicity) in the presence of travelling surface waves. Moreover, the friction coefficient on the wall of nanochannels is decreased obviously due to the large slip length, and the shear viscosity of fluid on the hydrophobic surface is increased by travelling surface waves. It can be concluded that the travelling surface wave has a potential function to facilitate the fow in nanochannels with respect to the decrease in surface friction on the walls. Our results allow to defne better strategies for the fast nanofluidics by travelling surface waves. (more details please see:

Wettability of invading fluid enhances compact displacement for immiscible flow in porous media

In an enlightening article published in 2015 in Physical Review Letters, Holtzman and Segre studied the effect of the contact angle (wettability) in immiscible two-phase flow in porous media in the unstable case of a less viscous invading fluid displacing a more viscous one.  The title of the article “Wettability stabilizes fluid invasion into porous media via nonlocal, cooperative pore filling”, indicates the main conclusion of the work.

They carried out numerical experiments in a two-dimensional disordered medium constructed with an array of solid cylinders of varying diameter set within a large rectangular area.  A degree of disorder was introduced by randomly selecting the cylinder’s diameter from a uniform distribution.  The array of cylinders consisted of repeated triangular lattices of fixed spacing between vertexes (the cylinders’ centres).  The fluid dynamics modelling at the pore scale took into account three types of meniscus instabilities, namely, bursting (Haines jumps), occurring when the curvature reaches a critical value; touching, when a meniscus intersects a third solid surface, and overlapping, when adjacent menisci encounter.  The latter is essential in creating a compact, smooth, displacing front as it promotes non-local, cooperative pore filling.  They used this porous medium model to simulate the experiments of Trojer, Szulczewski, and Juanes (2015).  The radial fluid motion was produced by injecting air at a certain rate, in a small region of the domain initially saturated with a solution of water and glycerol.  The simulations stopped when air reached one of the outer boundaries.

In various applications, it is of interest to have a compact moving front in immiscible two-fluid motion as this generates the more efficient displacement mechanism.  This is desirable, for instance, in oil recovery processes.

An important contribution of the study by Holtzman and Segre is the introduction of two novel dimensionless parameters, namely, a modified capillary number, obtained by altering the classical capillary number (ratio of viscous to capillary forces) with a function of the contact angle and a dimensionless geometric ratio, and the so-called “cooperative number”, given as a function of these two parameters and the angle between two adjacent menisci. This number is a measured of the likelihood of filling the pores by menisci overlaps. In their simulations, the angle between two adjacent menisci was fixed to 120o.

By computing these two parameters for the same data used in the numerical study, Holtzman and Segre correlated their values with the salient features of the interfacial morphology from the simulations.  These predictions followed the invasion regimes observed in the experiments referred to above.  For modified capillary numbers greater than approximately 4 x 10-3, simulations showed viscous fingering, characterized by long and thin fingers, regardless of the magnitude of the cooperative number.  On the other hand, for a modified capillary number smaller than 4 x 10-3 and positive cooperative numbers, indicating invasion by a wetting fluid, simulations predicted compact displacement, characterized by short and thick “fingers”.  For the remaining sector, for which the cooperative number is negative (non-wetting invasion), results showed thinner, adjacent fingers within fractal, long interfaces, typical of the capillary fingering pattern.  Therefore, their results showed that wetting invasion (imbibition) favors the formation of a compact front for low capillary numbers.

Finally, it is noteworthy that more recently, in 2016, Zhao, MacMinn, and Juanes, conducted a series of experiments by radially displacing a very viscous silicone oil in a porous system using water while systematically varying the wettability of the medium.  They found a trend similar to the one described by Holtzman and Segre in that increasing the wettability of the invading fluid promotes cooperative pore filling and hence a more compact front displacement.  Nonetheless, Zhao et al. surprisingly found that for a strongly wetting invading fluid, this trend is dramatically reversed and the displacement becomes markedly less efficient.  They noted that this sudden change, not reported previously, is caused by the dominant presence of corner flow, for which the invading fluid moves without filling the pore bodies, over cooperative pore filling.

A primer on the "ouzo effect"

Ouzo, a grape-based liqueur popular in the Mediterranean, appears at first sight to be better suited for the evening following a day of science rather than as part of a science experiment itself.  However, one can demonstrate some fascinating fluid dynamics by merely pouring water into a glass containing some ouzo*. The initially clear-coloured liquid quickly becomes cloudy – this is known as the ‘ouzo effect’. Watch the video below:

So what’s happening here? The answer lies in the anise oil dissolved in the ouzo, which spontaneously forms droplets as soon as it makes contact with the added water. These droplets are small and numerous enough to not just refract or reflect the light, but also scatter it in complex ways, thus giving the diluted ouzo its milky white colour. The reason the droplets form is that anise oil, or its flavour-lending compound Anethole to be more precise, is only slightly soluble in water but highly soluble in ethanol. As the concentration of water in the solution increases, the solubility of anethole in the solution decreases and it eventually becomes supersaturated. If the supersaturation is significant, nucleation of oil droplets occurs. Each droplet formation event depletes some anethole from the region near it, and the droplets thus tend to form apart from each other without coalescing.

This effect is not just specific to ouzo, and wikipedia has helpfully collated a list of other anise-flavoured spirits that show the ouzo effect**. In general, the effect can be observed in any ternary mixture of hydrophobic oil (here anise oil), water-miscible solvent (here ethanol), and water. The relative concentrations of solute, solvent and water in the mixture determine if/when the ouzo effect occurs. Note that you don’t need to add water, alternatively you could just wait for the ethanol to evaporate, and watch as the ouzo effect is triggered in a sessile drop as it shrinks, as recent papers have showed.

Figure: Cover art for this paper,  with initial drop of ouzo (bottom left), subsequent drops (right and top) after sufficient ethanol has evaporated to trigger the ouzo effect. 


The ouzo effect can be used to form stable emulsions where the drop density and size can be controlled without the need for mechanical or chemical intervention. As a result, it is used in numerous applications, from the formation of beverages and perfumes to selective microextraction in forensics and biomedical science. However, despite its many uses, the fundamental mechanism of the spontaneous emulsification that underlies the ouzo effect is still not well understood. This is because the ouzo effect is an inherently multiscale phenomenon. It starts from local fluctuations of solvent concentration at the molecular level, which results in nucleation of nanodroplets in a fraction of a millisecond; these nanodroplets then grow in size (via Ostwald ripening) to form microdroplets and finally, in a few seconds, become visible at the macroscale as clouding in the ouzo glass. As folks in this research group are well aware, systems involving large time- and length-scale separation of this nature are very hard to simulate.

Hypothetically, if we can ever get to working closely with experimentalists on modelling the ouzo effect, come the end of every day, work truly would be its own reward.

* Disclaimer: DO try this at home.

**As Martin, our resident connoisseur of greek sprits, will attest, tsipouro shouldn’t be on the list as it can be made without anise oil. His home-made tsipouro did not show any ouzo effect, and I had to use a video off youtube here instead.  


The chain fountain phenomenon

A chain fountain is the name given to the counterintuitive phenomenon where a long bead chain
appears to defy gravity by first leaping out of its container before falling to the ground.

This became known as the Mould effect, after a British science presenter, Steve Mould, who made the experiment famous with a video that went viral on YouTube.
Apparently, he discovered the chain fountain phenomenon while looking for a way to explain at the molecular level the capability of viscoelastic fluid to pour itself, the so-called open siphon effect.

In this funny TED talk, Steve Mould recounts his discovery and investigation into this entertaining and counterintuitive phenomenon.

The physics behind the chain fountain is non-trivial.

At first sight one could be tempted to explain the phenomenon by observing that the falling chain has downward momentum, causing an upward momentum in beads leaving the container. This, in turn, makes them leap before gravity can slowly reverse their momentum. However, if inertia causes the flowing fountain, the chain would be stationary at the top of the curve, while this is not the case.

It has been later proposed that the fountain is not driven by inertia, momentum, or gravity but rather by an anomalous push force exerted by the container on the link of the chain about to come into motion.
The authors of the paper also posted a video on the Royal Society Youtube channel where the educational value of the demonstration and the analysis is highlighted.

Since it was first brought to widespread attention by Mould, many explanations of the chain fountain have been put forward and several papers have been published, even recently (,,
However, a complete description of the mechanisms at play appears to be still lacking.

Ordered drop arrays levitating over heated surfaces

In the famous Leidenfrost effect, liquid drops levitate over a heated solid surface without touching it. For this effect to occur, the surface should be heated well above the boiling temperature of the liquid. On the other hand, levitation over a liquid surface can occur for much lower surface temperatures. We have all seen this when drinking hot tea or coffee. There is often what looks like a misty film over the surface. It is formed when the evaporating water condenses and some of the resulting microdrops levitate over the surface. These drops are remarkably uniform in size and form arrays with a high degree of order. Leidenfrost drops do not form such ordered arrays. In a recent article in Physical Review Letters, a group of Russian physicists was able to show that ordered drop arrays can occur over solid surfaces as well, well below the Leidenfrost temperature, and give an explanation for this phenomenon. To create such arrays, they started with a layer of water on top of a copper substrate.  With a short pulse of an air jet they created a millimetre-sized dry spot on the surface. The spot stayed dry after the air flow ceased, because the surface was rough enough that the contact line was pinned. Then they heated the substrate to 85 C. As expected, they observed an array of microdrops above the liquid, but some of the hovering drops moved over the dry spot and formed an ordered array there as well, though with less order.

In order to explain the phenomenon, the authors considered the dynamics of evaporation of the drops and associated air flows. When a drop is evaporating in air, the vapour concentration is, obviously, higher near the drop. However, the total concentration of molecules (air plus vapour) is roughly the same everywhere, and therefore the air molecule concentration is lower near the drop surface and there should be diffusion of air molecules towards the surface. But since the air molecules cannot penetrate the drop, there should be a counterbalancing convective flow away from the drop surface, which is called Stefan flow, after the Austrian-Slovene scientist of the Stefan-Boltzmann law fame. The authors claim that it is this flow that repels the drops from the liquid or solid surface and from each other, making them levitate and form ordered arrays.

While the authors do some calculations to justify their explanation, there are approximations involved, so simulations avoiding these approximations may be useful. Moreover, the sizes of the drops they observe, the distances between them and the heights above the substrate are about 5-10 microns, just large enough for continuum hydrodynamics to apply. It is plausible that smaller drops may be observed (this depends, in particular, on the temperature to which the substrate is heated), in which case taking gas-kinetic effects into account may be warranted.

This work has also been covered in Physics World.

Packmol: An easy tool to make initial configuration for MD simulations

Initial configuration of an atomic system is crucial in MD simulations. Further away the initial configuration from its equilibrium position, longer will it take to reach the equilibrium state. Longer simulation means burning up more and more computational resources.

For example, If you want to simulate a drop spreading on a Platinum surface using MD and you are not interested in the transient dynamics, one obvious way is to begin with a wall and a cube of water molecules placed just above it. The attraction from the wall will bring down the cube and the “droplet” will spread.

But, if you already had a hemispherical arrangement for water molecules instead of a cube, the system would be nearer to its equilibrium state at t=0. This will only take much less time to attain equilibrium and thus will help save a lot of computational resources.

The working procedure is very simple as well. You can download and install Packmol from Suppose we want 2000 water molecules randomly arranged inside a sphere of radius 25 Angstroms, all we need is two files:

File 1:

This file contains the atomic coordinates of two Hydrogen and an Oxygen atom according to TIP4P configuration of water molecule. The first line represents the total number of atoms to follow. Second line is neglected.




O         0.0       0.0       0.0

H          0.9572 0.0       0.0

H          -0.239  0.9266 0.0


File 2: input.inp

The packmole executable reads this file as an instruction to make initial configurations. In the file, tolerance is the average distance between two molecules and filetype is the output file format. The first three arguments of inside sphere denote the coordinates of centre of the sphere and last argument is the radius. You can also arrange water molecules inside a box, cylinder etc.


tolerance 2.0

filetype xyz



            number 2000

            inside sphere 0.0 0.0 0.0 25

end structure


Now, open a terminal in the packmol folder and execute “./packmol <input.inp”.

This will create a file named, which will have 2000 water molecules arranged inside a sphere centred at (0,0,0) with radius 25 Angstroms. Next step is to write a C++/MATLAB script to read this output file and print it in a format that can be read by the MD software that you use. If you find packmol useful, don't forget to site the original work.

Starting my two years WIRL-COFUND fellowship with Institute of Advanced Study.

    This month I completed my one year at Maths@Warwick and indeed it has been a magnificent twelve months. I got to meet some wonderful people within and outside of academia. It is only fair to start this blog saying thanks to our Micro Nano Flows (MNF) for Engineering group. First and foremost I wish to express my deepest gratitude to James and Duncan for making me a part of the MNF@Warwick group and extending their support and guidance. My special thanks to Prof Emerson and Prof Reese for introducing me to this fabulous group.

    From September 4th, I shall start my WIRL-COFUND fellowship with Institute of Advanced Study at Warwick. It is a two years fellowship programme which receives funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska Curie actions COFUND scheme. I shall undertake research in energy area that is linked to one of the eleven Warwick global research priorities. The main objectives of the proposed project are develop a theoretical model for the simulation of the evaporative cooling from nano-confined liquid-gas interfaces and Implementation an open-source comprehensive simulation tool for extended fluid dynamics equations.  I shall still be working with James and Duncan  at  Mathemtics Institute, WarwickI would again like to express my appreciation to the MNF group members who helped me immensely to win this scholarship.

HDF5 and H5MD


HDF5 is a binary file format and a software library for management of large and complex data sets. The development of the library was initiated in 1987 at the National Centre for Supercomputing Applications at the University of Illinois at Urbana-Champaign. Currently, the software library is supported and developed by the not-for-profit company HDF Group.

The software library provides high-level APIs written in C, C++, Fortran 90 and Java. HDF5 includes utilities for data slicing, data compression and parallel I/O. Bindings to HDF5 are available for MathematicaMATLAB, Python and other engineering and scientific software.

HDF5 has a long history of applications in CFD and other fields of science. For example,

HDF5 is distributed under the terms of an open source license.


H5MD is a file format specification for efficient and portable storage of molecular data. The specification was developed in an attempt to simplify the exchange of data between different analysis and simulation software.

The description of the file format was published in the journal Computer Physics Communications in 2014 [1]. Currently, the specification of H5MD is maintained in the form of an open source project at a git repository. Software utilities for management of H5MD are available in the form of C and Python libraries.

Software packages for integration with H5MD are available for several molecular dynamics programs, including LAMMPS and VMD.

External Links








[1] de Buyl P, Colberg PH, Höfling F. H5MD: A structured, efficient, and portable file format for molecular data. Computer Physics Communications. 2014 Jun;185(6):1546–53.

Latest News

Recent Publications

R Pillai, JD Berry, DJE Harvie, MR Davidson (2017) Electrophoretically mediated partial coalescence of a charged microdropChemical Engineering Science, 169: 273-283. (access here)

JF Xie, BY Cao (2017) Fast nanofluidics by travelling surface wavesMicrofluidics and Nanofluidics, 21: 111 (access here)

AP Gaylard, A Kabanovs, J Jilesen, K Kirwan, DA Lockerby (2017) Simulation of rear surface contamination for a simple bluff bodyJournal of Wind Engineering and Industrial Aerodynamics, 165: 13-22. (full paper here)