N.B. Sorry, but the workshop has reached the capacity of the room at CECAM (50) and we are currently unable to register further participants. We apologise for this.

Crystallisation is a fundamental and complex problem, and it is of importance in fields from materials and pharmaceuticals to climate science and ice-cream production. Crystallisation starts with nucleation, and there are two possible mechanisms for nucleation. The first is where the crystal nucleus forms in the bulk of the undercooled liquid or supersaturated solution, and the second is where the nucleus forms at a interface. Nucleation at an interface is called heterogeneous nucleation, and the free energy cost of forming the nucleus is typically dramatically lower there than in the bulk. Given the exponential dependence of the nucleation rate upon the free energy of forming a critical nucleus, this lowering means that in all but a few idealised systems, heterogeneous nucleation dominates. However, it is the idealised (no interface) case of homogeneous nucleation that has received most theoretical attention, not nucleation at interfaces, even though it is nucleation at interfaces that is the more practically important problem. In addition, not only is heterogeneous nucleation typically dominant in the formation of bulk crystals, but it is also, of course, crucial in the ever-growing field of crystallisation in confinement.

Here we seek to bring together communities of scientists studying nucleation in different systems and via computer simulation and experiment, as well as those concerned with practical challenges in the many systems where crystallisation is influenced by an interface. We identify three key areas where the role of interfaces is crucial: (i) The freezing of water, particularly in aerosol droplets. (ii) Crystallisation in confined molecular systems, such as nanoporous materials and microemulsions. Of course, in these nanoscale systems the ratio of interface to bulk is extremely high. (iii) Heterogeneous nucleation at surfaces in colloidal systems and the effect of self-generated grain boundaries.

There are important open questions about how interfaces can control the nucleation and growth of bulk crystals, and there is an ever-growing interest in interface-dominated, nanoscale confined systems. Despite this there has never been a workshop focusing only on the role of interfaces in crystallization. Our proposed meeting will bring together computer simulators and experimentalists from a wide range of fields that typically do not talk to each other, so that they can learn from one another. By doing this, we believe that our workshop will achieve a step change in research in crystallisation and heterogeneous nucleation.

Over the last last few years, the state of the art in all three areas of interest outlined in the motivation has rapidly advanced. These areas are: both experimental and simulation studies of heterogeneous nucleation; the simulation and experimental study of water freezing; and crystallisation in confinement. We will briefly summarise the state of the art of each in turn.

So we start with heterogeneous nucleation. Until recently almost all computer simulation studies of nucleation considered homogeneous nucleation, where there is no interface. Only very recently has attention begun to focus on crystallisation at interfaces (Page and Sear, 2009, Sear, 2012, Dorosza and Schilling, 2012), and in pores (van Meel et al., 2010). In both cases it is clear that interfaces dramatically affect nucleation. The nucleation barrier at interfaces is much lower than in the bulk, and so nucleation at the interfaces should dominate nucleation in the bulk.

Also, experiments on colloidal systems are now starting to look at nucleation at interfaces in a systematic and quantitative way (Ivlev et al., 2012, Franke et al., 2011). For example, the Egelhaaf and Lowen groups have demonstrated quantitative agreement between simulation and experiment, for hard sphere crystals nucleated on a surface (Sandomirski et al., 2011). This agreement is particularly important: hard spheres are the only system for which quantitative data exist for both experiment and simulation, and these heterogeneous results contrast with the notorious discrepancy in homogenous nucleation rates, which was discussed in a 2010 CECAM workshop we organized. To our knowledge, the work of Sandormirski et al. is the only particle-resolved study of heterogeneous nucleation of a crystal. But there are other examples of recent work in colloidal systems, on the role of interfaces. For example, that on self-generated interfaces in binary colloidal systems formed from particles expelled from nuclei (Yoshizawa et al., 2011).

Now we turn to the state of the art in the study of the nucleation of ice. There have been many attempts to simulate the crystallisation of water, but until recently these have been handicapped by limitations of both the model used (Vega and Abascal, 2011) and of computer power. But now models are better (Vega and Abascal, 2011, O'Neill et al. 2011) and computing power is improving so simulations are now beginning to provide exciting and reliable knowledge of crystallisation (Moore and Molinero, 2011). However, although we are now able to undertake simulation studies of homogeneous nucleation of ice with unprecedented accuracy, the heterogeneous nucleation of ice is still essentially unstudied. This is in spite of the fact that in experiment ice presumably nucleates at an interface and apparently often nucleates preferentially at the three-phase line where a solid/water interface hits an air/water interface (Shaw et al., 2005).

Recent experimental results of Shaw and co-workers (Gurganus et al., 2011) have reopened this question of the underlying mechanism for this observation that ice grows from the line where two interfaces meet. As nucleation at the interface can often determine the temperature at which atmospheric water droplets freeze, a better understanding here would feed into, for example, improved models for our climate and hence for the extent and impact of global warming. Thus it is important to understand this phenomenon better. The state-of-the-art in experiment is limited here by the inability to probe the microscopic dynamics of nucleation. These dynamics can now be studied by computer simulations of the newly improved models.

The final area we consider here is that of crystallisation in confinement. Here interfaces can not only influence the nucleation barrier but also confine the crystal. It is particularly timely to discuss crystallisation in confinement. Recent work on crystallisation confined in nanoscale pores (Hamilton et al., 2012) has shown that confinement can change the polymorph that forms, an intriguing observation given that polymorph control is very important in fields such as the pharmaceutical industry. For example, the formation of a previously unsuspected polymorph of the anti-HIV drug Ritonavir disrupted supply and was extremely costly. The work of Ward and coworkers (Hamilton et al., 2012) in particular has emphasised the thermodynamic role of interfaces in regular pore systems where polymorph selection can be controlled on a larger scale.

Cooper and coworkers have recently studied crystallisation in a different system but one where there is also nanoscale confinement. Their system consists of nanoscale volumes of a crystallising solution inside microemulsion droplets (Nicholson et al., 2011). Cooper and coworkers emphasise the role of kinetics. Also, in their systems although the crystallising system is initially confined, the droplets can coalesce allowing the crystals to grow, and ultimately resulting in bulk crystals. Like Ward and coworkers, they also observe different polymorphs in their system to those obtained in bulk crystallisation. Thus both nanoscale pores and microemulsions potentially offer control not only nucleation but over which polymorph is formed.

Indeed, work on biomineralisation has shown that living organisms manipulate crystallization, apparently by using the surfaces of proteins to control nucleation (in a polymorph specific way), and can do so in volume confined by lipid bilayers (Nudelman and Sommerdijk, 2012). There are many examples in living organisms, such as sea urchin spines, mollusc shells, etc, where the control over crystallisation is extremely impressive, and where we know crystallisation is occurring at interfaces. However, our ability to copy this control is limited by our very poor understanding of the microscopic dynamics that underly this control.

We expect in the near future, both simulation and experiment will be used in combination to advance the the state-of-the-art described above. Interfaces are ubiquitous and very often dominate nucleation. For example, the recent developments in our understanding of the role of microstructural details in colloidal systems is ready to be applied in simulations of water, since there now exist more accurate water models. Furthermore, these same concepts of local structure and dynamics and their role in the formation of nuclei can now be used to deepen our currently limited understanding of crystallization in confined molecular systems. These are just two examples of where the state-of-the-art is ready to be advanced by collaboration between scientists working on different systems where crystallisation is crucial.