2 Overview

The paper of Marston et al. (Reference Marston, Truscott, Speirs, Mansoor and Thoroddsen2016) studies the markedly thin splash that is created by a sphere impacting onto a deep liquid pool (figure 2). While penetrating into the liquid, the sphere creates a void in its wake which is open to the ambient air. As this void deepens and expands, air needs to flow into the cavity – indicated by the red arrows in figure 2(b) – which creates a lower pressure at the inside of the splash, causing it to close in upon itself, the so-called surface seal. In order to systematically study this phenomenon the authors controlled the ambient pressure during the impact experiment. More specifically, they built a 6 m tall vacuum chamber, allowing for ambient pressures ranging from just above the vapour pressure ( $1/16$  bar) to atmospheric values (1 bar).

Figure 2. Surface seal and buckling instability of a 10 mm diameter sphere impacting on water. The images are taken at $t=2.7$ , $3.5$ , and 4.4 ms from impact and the red arrows in (b) indicate the airflow into the void (from Marston et al. Reference Marston, Truscott, Speirs, Mansoor and Thoroddsen2016).

As expected Marston et al. (Reference Marston, Truscott, Speirs, Mansoor and Thoroddsen2016) found a significant dependence of the surface seal phenomenon on the ambient pressure. However, the delay of the crown splash seal at lower pressures was not as large as one would expect if only the pressure difference were responsible for the closure. Together with the observation that the closure speed is largely independent of ambient pressure, this strongly points to the importance of yet another force for the surface seal, namely that due to surface tension. They argue however that both forces are significant, and even if surface tension in the end may drive the collapse of the splash, the surface seal phenomenon should be viewed as the confluence of pressure and surface tension effects.

The paper is not only ingenious in its conclusions, but also in the smaller details. Such as the velocity of the sheet of which the splash is built, measured by tracking particles that happen to be present in the crown. Or the thickness of the sheet, estimated from the opening speed of holes that spontaneously form inside the splash.

The second part of the article is about the buckling of the splash, which occurs approximately when the splash, during closure, progresses through a cylindrical shape, as is clearly visible as the wavy pattern along the circumference of the splash in figure 2(b) and the vertical ribbon-like structures of figure 2(c). The onset of the buckling is an example of a hydrodynamic instability. This is a phenomenon in which the liquid spontaneously changes its shape and/or appearance. The waves that emerge on an otherwise quiet lake subject to the action of the wind result from such an instability, just like the break up of a thin water stream from a faucet into droplets. The size of the ultimately observed structures, such as the length of the wave and the diameter of the droplets, are already determined by this initial instability. More precisely, by that unstable wavelength that has the largest growth rate.

But Marston et al. (Reference Marston, Truscott, Speirs, Mansoor and Thoroddsen2016) went beyond the mere characterisation of the buckling instability and connected the wavy pattern that is visible in the crown in figure 2(b) all the way down the cavity to another instability, namely one that occurs on the surface of the sphere. The waves created by this contact line instability could be traced over the entire cavity wall and connected to the wavelength of the buckling instability arising in the splash. Thus it appeared that the buckling instability inherited its wavelength from the contact line instability on the sphere.

This, in itself, is already a remarkable conclusion, but the authors did not stop there. They asked themselves what wavelength they would observe in the splash in the absence of the contact line instability. This would have been a rather artificial question if the authors had not come up with a trick to actually realise such a situation in experiment. They exploited the Leidenfrost effect, the observation that a droplet can be made to hover for minutes on its own vapour, when deposited on a plate which is heated to a temperature above the Leidenfrost point. When a metal sphere heated to well above the Leidenfrost point impacts on a liquid, a vapour layer forms around the entire sphere, which creates a void and a splash without ever coming into physical contact with the liquid. This almost magical trick completely avoids the contact line instability. The conclusions that the authors could draw from this experiment was that, in absence of the contact line instability, the buckling still occurred with a wavelength slightly different from that observed in the original experiment, but that its growth was significantly postponed. This stands to reason, since now the buckling instability needed to occur on its own, without being assisted by the wave pattern imposed by the contact line instability.

3 Future

Many strong conclusions were drawn in Marston et al. (Reference Marston, Truscott, Speirs, Mansoor and Thoroddsen2016) but arguably the most significant one is that of the link between the contact line and buckling instabilities, which appears to be connected to the fact that the most unstable wavelengths in both cases happen to be similar. An intriguing point is what would happen if this were not the case. The authors tried to control the wavelength of the contact line instability by etching patterns and creating alternating hydrophobic–hydrophilic stripes on the surface of the sphere. A great idea which unfortunately failed, yet strengthens the diameter scaling for the buckling that was proposed in the article. However, since the buckling and the contact line instability appear to be governed by different physics, would it be possible to separate the most unstable wavelengths by going to a different scale? Or would it be conceivable to induce a different type of instability on the cavity wall by adapting the shape of the object?

One of the remaining puzzles in the field is determining what controls the number of droplets that is produced from the crown splash. The work of Marston et al. (Reference Marston, Truscott, Speirs, Mansoor and Thoroddsen2016) suggests that the number of droplets in some cases may not be controlled by an intrinsic instability of the crown’s rim, something that has been assumed in most of the works that have dealt with this problem. Instead it may very well be connected to another instability, such as the contact line or buckling instability. In this light, it would be challenging to see if the number of droplets can actually be controlled by imposing another instability onto the splash. It is even conceivable that a link between the two instabilities as observed by the authors may actually be more common, and something that the fluid dynamics community should be aware of.

A final word of praise must go to the superb, almost artistic, quality of the images that are presented in Marston et al. (Reference Marston, Truscott, Speirs, Mansoor and Thoroddsen2016). They bring experimental research based on high-speed imaging to a very high standard, which is commendable for all areas of science in which imaging techniques play an important role.