Hostname: page-component-7b7567c8c-fvjsw Total loading time: 0 Render date: 2026-11-16T06:21:20.495Z Has data issue: false hasContentIssue false

Surface bubble coalescence

Published online by Cambridge University Press:  25 March 2021

Daniel B. Shaw
Affiliation:
Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, NJ 08544, USA
Luc Deike*
Affiliation:
Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, NJ 08544, USAHigh Meadows Environmental Institute, Princeton University, Princeton, NJ 08544, USA
*
†Email address for correspondence: ldeike@princeton.edu

Abstract

We present an experimental study of bubble coalescence at an air–water interface and characterize the evolution of both the underwater neck and the surface bridge. We explore a wide range of Bond number, $Bo$, which compares gravity and capillary forces and is a dimensionless measure of the free surface's effect on bubble geometry. The nearly spherical $Bo\ll 1$ bubbles exhibit the same inertial–capillary growth of the classic underwater dynamics, with limited upper surface displacement. For $Bo>1$, the bubbles are non-spherical – residing predominantly above the free surface – and, while an inertial–capillary scaling for the underwater neck growth is still observed, the controlling length scale is defined by the curvature of the bubbles near their contact region. With it, an inertial–capillary scaling collapses the neck contours across all Bond numbers to a universal shape. Finally, we characterize the upper surface with a simple oscillatory model which balances capillary forces and the inertia of liquid trapped at the centre of the liquid-film surface.

Information

Type
JFM Papers
Copyright
© The Author(s), 2021. Published by Cambridge University Press
Figure 0

Figure 1.(a) Sketch of the geometry of air bubbles coalescing at a liquid–gas interface and the quantities measured in this study. Here, $r$ is the distance that the centre point of the underwater contour travels and $h$ is the vertical distance travelled upwards by the centre of the top surface. They are both measured relative to their position at the start of coalescence, $t = t_*$. In blue is a circle of radius $R_0$, which is the size of a sphere containing the same volume of air. (b,c) The shape of a single static air bubble at a water–air interface as determined by a numerical solution of the Young–Laplace equation. Bubbles with $Bo \ll 1$ are nearly spherical and almost completely surrounded by the liquid phase with a very small thin-film cap. For larger Bo, the bubble resides higher relative to the undisturbed free surface and the shape is less spherical. The solid black line is the interface between the air inside the bubble and the surrounding liquid. The solid red line is the thin-film cap between the air inside the bubble and exterior air at atmospheric conditions. The dashed line is the free surface between the exterior air and water.

Figure 1

Figure 2.(a) Two high-speed cameras record time-synced side views: one below the surface of the water and one above. (b) Example of measurements of bubble coalescence for $Bo = 1.2$. (c) The associated time evolution of the neck $r(t)$ (blue) and bridge centre $h(t)$ (red). Time is measured relative to the start of coalescence, $t_*$.

Figure 2

Figure 3.Time series of $Bo=0.13$ coalescence ($c=0.001$ mMol). Early times of coalescence are highlighted by the axisymmetric neck expansion (a–c). The free surface does not appear to alter the growth of the lower part of the neck (e–m). As time continues, the shape becomes more spherical and oscillates near its natural mode frequency (q–x).

Figure 3

Figure 4.Bubble coalescence for $Bo=2.1$ ($c=0.001$ mMol). The free surface and the neighbouring parent bubble change the initial geometry. Underwater, the parent bubbles are still roughly ellipsoidal and the neck between them expands in a fashion similar to spherical bubbles coalescing for $t/\tau _{i}<1$ (a–f). The top surface shows distinctly different behaviour. While the bottom neck has started to grow in (b), the upper surface only first shows signs of movement in (c) and the entire liquid film is moving only a short time after ($\Delta t \ll \tau _i$). The upper surface reaches its maximum (f) very quickly – well before the bottom surface (p). Note that the liquid from the meniscus between the parent bubble caps is trapped in the liquid film of the upper surface and eventually spreads out over the newly formed bubble's cap (h–r), as shown by the light-coloured region at the centre of the newly formed cap.

Figure 4

Figure 5.Underwater neck growth (a) $r(t)-r(t_*)$ as a function of $t-t_*$ for $Bo = 0.15$ to $2.38$. The solid lines are the smallest surfactant concentration, $c=0.001$ mMol SDS, and the dotted lines are $c=0.002$ mMol. The inset shows the same data rescaled by the volumetric radius, $R_0$, as is done for spherical bubble coalescence. The large spread of the data illustrates that it is not the correct normalization. (b) The in-plane curvature at the bubble's ‘corner’ (its widest part), $R_c$, is found by inspection of numerical solutions of the Young–Laplace equation for different sized bubbles, the result of which is shown by the open circles. The corner curvature increases with $Bo$ and can be described by $R_c = R_0 (1+Bo)^{-1/2}$, shown in dashed lines. The inset shows where $R_c$ (blue circle) is measured on the three-dimensional geometry of a $Bo=3$ bubble. The black region is the air–liquid interface and the red is the thin-film cap. (c) Re-scaling of the neck data with $R_c$ as the characteristic length scale and $\tau _i(R_c) = \sqrt {\rho R_c^3 / \sigma }$ as the characteristic time scale. The data collapse to a single curve, $[r-r(t_*)]/R_c = C [(t-t_*)/\tau _i(R_c)]^{1/2}$ (indicated in black), with $C=1.2$ by best fit to the data. There is no visible difference between the two concentrations of surfactant tested, showing independence of the results in this low concentration limit.

Figure 5

Figure 6.Underwater neck contour evolution for $Bo=0.25$ (a–c) and $Bo=2.0$ (d–f), showing the images (a,d) just prior to the start of coalescence, with the subsequent growth of the neck overlaid. In all other plots, the black dot at $(0,0)$ represents the origin where the neck starts at $t=t_*$. (b,e) The physical scale of the neck. (c,f) The same contours rescaled by the appropriate vertical ($\tilde {y} = R_c[(t-t_*)/\tau _{i}(R_c)]^{1/2}$) and horizontal ($\tilde {x} = R_c[(t-t_*)/\tau _{i}(R_c)]$) length scales. (g) A comparison of the rescaled data for all times at each Bond number (c,f), with the shaded area indicating the standard deviation about the mean. For all $Bo$ studied, the entire lower neck contour collapses to a single curve. Results for $Bo=0.25$ and $Bo=2.0$ are highlighted as they show collapse across roughly an order of magnitude in Bo.

Figure 6

Figure 7.Evolution of the top surface and the relevant geometric parameters. The meridian contour sequence for $Bo=0.67$ (a) and $2.0$ (b). The dark blue curve is the start of coalescence, and the light green is the first peak height of the centre (elapsed time between each line is different for (a,b)). There are many similarities in the contour evolution of each, including the trapped liquid mass and curvature reversal at the centre. (c) Evolution of the top surface centre, $h(t)$, during coalescence. The characteristic length and time scales of the behaviour both increase with Bond number. The end of the recorded signal is when travelling capillary waves start to dominate the dynamics of $h$, creating a disordered signal. As the scale of the bubble cap increases with Bond number, the larger Bond number trials exhibit more oscillations while the midpoint barely reaches its first peak before capillary waves arrive for the smallest Bond number trial ($Bo=0.38$). The dotted lines are at $c=0.002$ mMol SDS with no visible changes compared to the cases with lower surfactant.

Figure 7

Figure 8.Top surface geometry. (a) The cylindrical coordinate system of the top surface; $\phi$ is the angular coordinate measured from the vertical (the meridian contours lie on $\phi =0$), $r$ is the polar coordinate, and $z$ is the longitudinal coordinate. (b) Measured features of the top-surface geometry which are relevant to the dynamics. Here, $L$ is the horizontal spacing between the cap peaks, and $D$ is the width of the liquid trapped in the meniscus between the bubble caps. (c) Cross-section of the trapped liquid mass in the plane containing the $z$-axis and $\phi =0$. At the moment of coalescence, the bulge is estimated as a triangular toroid, which is what would be obtained by rotating the section highlighted in red $-\phi _c < \phi < +\phi$ about the $z$-axis.

Figure 8

Figure 9.Retraction of the film separating $Bo = 2.6$ parent bubbles with $c=0.002$ mMol of SDS. The bubbles were angled roughly $30^{\circ }$ away from the camera (the left bubble is further away) such that the liquid bulge at very early times of coalescence can be seen. Prior to coalescence, it is the intersection of three thin films, and after the centre film retracts the liquid is trapped. Tracking the exact shape of the trapped liquid is complex, and an initially triangular cross-section is assumed to estimate the amount of liquid in the bulge.

Figure 9

Figure 10.Measured features and the proposed time and length scales of the problem. The black filled circles correspond to experiments at $c=0.001$ mMol and the unfilled stars are $c=0.002$ mMol. (a) Measurements of the cap peak spacing, $L$, as a function of $Bo$. The grey dashed lines are the theoretical geometric limits. At $Bo \ll 1$, the caps are separated by $L=2R_0$ (the upper grey dashed line) since the geometry is nearly that of two spheres in contact. The lower grey dashed line ($L=2^{1/3}R_0$) is the $Bo \gg 1$ limit in which a Plateau border-like structure is observed with the caps intersecting close to where a single bubble cap is $120^{\circ }$ from vertical. (b) Measurements of the trapped liquid width, $D$, as a function of $Bo$ and the proposed function (dashed line): $D/R_0 = \chi (1+Bo)^{-0.5}$, with $\chi =0.75$ fit to the data. (c) Comparison of the measured time scale with that of the model, $\tau _{I,top} = 0.612\sqrt {\rho _l D^2 L / \sigma }$ with $0.612$ fit to the data. (d) Value of $\Delta H$, which is the height of the first peak, as a function of Bond number. Note that increasing surfactant over the range studied does not change the dynamics.

Figure 10

Figure 11.Collapse of $h(t)$ across Bo made dimensionless by the height ($\Delta H$) and time ($\tau _{I,top}/2$) of the first peak for each signal. As the proposed oscillator model is valid after the first peak ($(t-t_*)/\tau _{I_top}>0.5$), this part of the data is highlighted in the plot. The collapse of the data supports the proposed model which balances the change in inertia of the trapped liquid mass with the restoring force of surface tension as the liquid mass oscillates about its equilibrium point, which is determined by the balance of surface tension and gravity. The symmetry of the film reduces the model to a one-dimensional oscillator.

Shaw and Deike supplemementay movie 1

Surface bubble coalescence as viewed from above and below the free surface.

Download Shaw and Deike supplemementay movie 1(Video)
Video 44.2 MB

Shaw and Deike supplemementay movie 2

Retraction of the film separating two parent surface bubbles at the start of coalescence.

Download Shaw and Deike supplemementay movie 2(Video)
Video 5.4 MB