Geometry and Topology of Two-Dimensional Dry Foams: Computer

Mar 27, 2017 - The simulation data is shifted horizontally by +0.1 sides per bubble. .... the Rio Tinto Centre for Advanced Mineral Recovery, is grate...
0 downloads 0 Views 5MB Size
Subscriber access provided by The University of Melbourne Libraries

Article

Geometry and topology of 2D dry foams - computer simulation and experimental characterisation Mingming Tong, Katie Cole, Pablo R. Brito-Parada, Stephen Johan Neethling, and Jan J. Cilliers Langmuir, Just Accepted Manuscript • DOI: 10.1021/acs.langmuir.6b03663 • Publication Date (Web): 27 Mar 2017 Downloaded from http://pubs.acs.org on April 4, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Langmuir is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

Geometry and topology of 2D dry foams - computer simulation and experimental characterisation Mingming Tong*§1,2, Katie Cole§3,4, Pablo R. Brito-Parada 3, Stephen Neethling3, Jan J. Cilliers3

1. School of Mechanical and Materials Engineering, University College Dublin, Ireland.

2. Mechanical Engineering, College of Engineering and Informatics, NUI Galway, Ireland 3. Froth and Foam Research Group, Department of Earth Science and Engineering,

Imperial College London, UK.

4. Department of Physics, University of Cape Town, South Africa.

Abstract

Pseudo-2D foams are commonly used in foam studies as it is experimentally easier to measure the bubble size distribution and other geometric and topological properties of these foams than it is for a 3D foam. Despite the widespread use of 2D foams in both simulation and experimental studies, many important geometric and topological relationships are still not well understood. Film size, for example, is a key parameter in the stability of bubbles and the overall structure of foams. The relationship between the size distribution of the films in a foam and that of the bubbles themselves is thus a key relationship in the modelling and simulation of unstable foams. This work uses structural simulation from Surface Evolver to statistically analyse this relationship and to ultimately formulate a relationship for the film size in 2D foams that is shown to be valid across a wide range of different bubble poly-dispersities.

ACS Paragon Plus Environment

1

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 26

These results and other topological features are then validated using digital image analysis of experimental pseudo-2D foams produced in a vertical Hele-Shaw cell, which contains a monolayer of bubbles between two plates. From both the experimental and computational results it is shown that there is a distribution of sizes that a film can adopt and that this distribution is very strongly dependent on the sizes of the two bubbles to which the film is attached, especially the smaller one, but that it is virtually independent of the underlying poly-dispersity of the foam.

Keywords: foam, pseudo-2D, bubble, topology, film size distribution 1. INTRODUCTION Structure and stability influence the behavior of foams in many applications and products. For example, in froth flotation, these properties play important roles in the separation of valuable mineral species, while in many solid foam products, such as metal and polyurethane foams, the evolution of the bubble size distribution in the liquid precursor influences the ultimate physical properties of the material. Subprocesses such as bubble coalescence are, however, not yet completely understood. Previously Tong et al. (2011)1 developed a numerical model for the drainage and stability of dry 2D foams with the aim of predicting changes in the bubble size distribution due to coalescence in froth flotation. The numerical model was based on the two-way coupling of a population balance model for the bubble size with a model of liquid drainage. One aspect of the model requiring improvement was the bubble topology and geometry. In this previous work all bubbles were assumed to be hexagonal prisms when calculating the size of the films between the bubbles, the length of the Plateau borders and the number of neighboring bubbles; all of which are important parameters when modelling liquid drainage and coalescence in foams. The shape assumption is relaxed in this work to take into account the actual topological and geometric properties of bubbles within pseudo-2D foams. In particular the distribution of the number of films surrounding the bubbles and the size distribution of these films is investigated.

ACS Paragon Plus Environment

2

Page 3 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

There has been extensive research carried out with respect to the topological and geometrical structure of 3D foams. Using computational methods, a variety of geometrical and topological features of 3D foams were formulated by Kraynik’s pioneering work 2-5 , including the number of sides per face, number of faces per bubble, bubble surface area and the length of the Plateau borders. The work by Tong and Neethling (2010) 6 mathematically formulated the dependence of film size on the size of the bubble to which the film is attached for 3D bubbles. It was recently7 found that biological tissues, foams, compressed emulsions and granular matter all show topological characteristics that can be categorized by the dominant source of disorder through a mean-field approach, and correspondingly the number of sides per bubble can be mathematically formulated as a function of bubble size. Optical tomography with a numerical reconstruction procedure8 has been used to characterize the topology of real foam, including the average number of faces per bubble and edges per face, which confirmed the original experimental work of Matzke (1946)9. While digital image analysis is commonly used as the mode of measurement of foam topology, it has limited application in 3D foams due to the inherent ability of films to scatter light. Digital imaging can capture bubbles either at the free surface of the foam or along the transparent walls of a container, but it is difficult to measure bubbles directly inside the foam. These “wall” bubbles are not a direct representation of the internal structure of the foam due to the strong influence of wall effects on the bubble topology10. On the other hand, the structure of pseudo-2D foams can be measured directly. For the example of a Hele-Shaw cell, the bubbles moving along the wall are constrained in shape with respect to an axis normal to the wall, and hence they can be assumed to be of 2D geometry. This is a reasonable simplification, particularly in the dry foam regime (liquid fraction less than 5%) in which the Plateau borders account for most of the volume of liquid in the foam and the vertices and films account for relatively little of the liquid volume. The structure of a pseudo-2D foam can be readily imaged because it is composed of a monolayer of bubbles and the influence of the “wall” effects on the bubble topology is

ACS Paragon Plus Environment

3

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 26

relatively small compared with the case of a 3D foam. Pseudo-2D foams thus offers opportunities for foam research, particularly for the purpose of validating computational models. They have been used extensively to study the geometry and topology of foams computationally11,12 and theoretically13,14. While related experimental research has been conducted in order to study bubble coarsening 15,16,17 , the permeability of films 18, the flow and stability of bubbles

19

as well as the liquid content and electrical

conductivity 20-22, the combination 1,23 of experiment and computational simulation has turned out to be a powerful method for investigating the key features of pseudo-2D foams. However, none of the previous research in 2D has taken into account the size distribution of the films between the bubbles of poly-dispersed foams, which is a crucial structural parameter in the computational modelling of bubble coalescence. In this paper, a combined computational and experimental approach is presented in which a relationship between the bubble size and film size distribution for dry 2D foams is developed and validated.

2. SIMULATION AND EXPERIMENTAL METHODOLOGY The overall strategy of this work is to use the simulation of the structure of 2D foams to predict key topological and structural parameters of interest and to use corresponding experimental measurements for validation.

2.1 Computer simulation method

By assuming that the 2D foam is in the dry limit (liquid fraction less than 5%), each thin liquid film degenerates to a 1D edge constrained by Plateau’s laws. The bubbles can then be represented as trivalent polygons with edges that form segments of circular arcs. The initial structure of the foam is produced by applying a Voronoi tessellation onto a 2D square domain, which is periodic in both

ACS Paragon Plus Environment

4

Page 5 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

directions. Using a periodic domain eliminates edge effects that would otherwise be present in the simulations. In order to produce a foam structure with a given level of poly-dispersity, the size of every bubble in the foam (area in 2D) is gradually adjusted according to a skewed normal distribution. The obtained foam, with an assigned target poly-dispersity, is then relaxed and annealed using Surface Evolver2-6. One thousand 2D bubbles were used in each simulation. The poly-dispersity of the bubbles is represented by the normalized standard deviation (NSD) of the equivalent circular radius of the bubbles. Eight different levels of poly-dispersity were studied in this work, ranging from mono-dispersed (i.e. NSD of 0) to a NSD of 0.7. In the simulations with the highest poly-dispersity, the largest bubbles were more than 100 times the area of the smallest ones. As the seed points for the Voronoi tessellation are random, at each level of poly-dispersity, 10 different simulations were carried out in order to increase the amount of data available for analysis. Figures 1(a–c) show the structure of three 2D foams with different levels of poly-dispersity (NSD = 0.1, 0.4 and 0.7).

Figure 1. 2D foam structures of different poly-dispersities (a) NSD = 0.1; (b) NSD = 0.4; (c) NSD = 0.7, as generated by the structural computation.

2.2 Experimental method

ACS Paragon Plus Environment

5

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 26

A Hele-Shaw style foam column1,24 was employed to create a vertical pseudo-2D foam trapped between a pair of parallel Perspex plates with a separation of 5 mm. Conventionally horizontal HeleShaw foam columns are used when trying to produce pseudo-2D foams (e.g. Osei-Bonsu et al.

25

).

However, vertical Hele-Shaw cells have been used to produce pseudo-2D foams26 when the drainage of liquid together with its coupling to coalescence was under investigation. A schematic of the design of the foam column is shown in Fig.2. Narrow capillary tubes of diameter 1.5 mm were arranged in parallel to provide an even distribution of air between two internal weirs, and the air was supplied from an air compressor at a rate of 5.5 litres per minute. The column contained two internal weirs so that the foam overflowed continuously in two places. After overflowing, the foam collapsed and the liquid was collected at the lower corners of the column, where it was recycled for foam production by a small gap at the base of the weirs. The aqueous foaming solution was made from 300 ml deionised water with 1 g/L methyl isobutyl carbinol and 0.1 g/L xanthan gum. The solution was chosen to produce a foam that exhibited a reasonable amount of coalescence and bursting at the chosen air rate without being too unstable to overflow. At the base of the column, bubbles rose through a high liquid content boundary layer to form a foam where the cross-section of the bubbles was roughly circular with an average area of approximately 15 mm2. Above the high liquid content boundary layer, the bubbles became increasingly polygonal in shape due to liquid drainage causing the shrinking of the Plateau borders. At the same time, the bubbles experienced coalescence, with bubble areas increasing to about 250 mm2 at the top of the column. With these bubble sizes and the given separation between the plates, the bubbles formed geometric prisms with close to constant polygonal cross-sections

27

.

Examples of bubbles from the foam are shown in Fig. 3, where the numbers of sides found ranged from 3 to 18. The column was positioned 200 mm in front of a light source which was composed of four bright white fluorescent strip lights and a diffuser screen. The column was operated for 25 minutes in order to achieve steady state overflowing height and velocity14 before images were captured for analysis.

ACS Paragon Plus Environment

6

Page 7 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

Figure 2. A schematic of the foam column with measurements in millimetres. The grid shows where the images of the bubbles were taken.

ACS Paragon Plus Environment

7

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 26

Fig. 3. 2D foam images showing bubbles with increasing number of sides. The star shapes indicate the bubble with the indicated number of sides, n.

The area of the foam column was split into a grid of 3 (horizontal) and 6 (vertical) focus elements (Fig. 2). High resolution photographs (4272 × 2848 pixels) of each element were captured from a distance of approximately 500 mm with a Canon EOS 450D D-SLR camera mounted on a tripod (Fig. 4). Each image had the focus element centered in the field of view to minimise parallax effects that would increase the apparent width of Plateau borders near the edges of each photograph. Three sets of photographs were taken for each focus element to enable repeat measurements.

ACS Paragon Plus Environment

8

Page 9 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

Figure 4. High resolution images of the focus elements of foam arranged according to the grid structure of the column.

ACS Paragon Plus Environment

9

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 26

The images of the foam taken in front of the light source showed sharp differences in intensity between the gas in the bubble and the liquid of the Plateau borders and films. A range of pre-processing steps were performed in Image J28 to enhance these differences and to create the image “mask”. A watershed algorithm was then applied to the mask to identify each unique bubble using the Matlab Image Processing toolbox. The number of neighbors per bubbles was found by passing a (6 × 6) kernel over the watershed image and counting bubbles occurring adjacently. Bubbles were ignored in an edge zone around all four sides of each image, the width of which was larger than the equivalent circular radius of the largest measured bubble. This avoided the measurement of bubbles that were not wholly within the field of view or which were exhibiting significant parallax effects. For those image edges that were near the edge of the domain it also eliminated bubbles experiencing significant distortion due to edge wall effects or distortion due to the proximity of the foam surface. The nodes where 3 bubbles meet were identified by passing a 3 × 3 kernel over the images in order to identify points neighbored by more than 2 different bubbles. The films of each bubble of interest were measured by fitting a path along the edge between pairs of nodes that shared two bubbles in common; including a straight path for small bubbles and a curved path for larger bubbles in order to save computational time. Additional checks were performed for the particular case of large bubbles surrounded by small bubbles as problematic situations such as multiple contacts between pairs of bubbles could potentially arise. Bubbles undergoing coalescence or a topological rearrangement at the moment of image capture were discounted from the measurement as they caused blurred and uncertain edge locations. An example of the final output of the image processing method is illustrated in Fig. 5. The largest contribution to the uncertainty in the measurements of bubble size and film length arose from the watershed part of the process, which needs to assign a contact line between bubbles which are separated by Plateau borders of finite width. This had the greatest effect at the base of the foam where the liquid fraction tended to be higher.

ACS Paragon Plus Environment

10

Page 11 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

Figure 5. Output of the image processing method showing nodes, bubble center and film center overlaid on a grayscale image of the foam photograph.

While the distance between the wall panels of the Hele-Shaw cell can affect some features of the pseudo-2D foam (e.g. the drainage dynamics of the liquid), it does not affect the key topological features (e.g. size distribution of films) of the pseudo-2D dry foam that we present in this paper as long as the gap is small enough (which is the case in this work).

ACS Paragon Plus Environment

11

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 26

For each set of photographs that were taken in the experiment, the poly-dispersity of the bubbles (represented by the NSD) in each of the 3 (horizontal) and 6 (vertical) grid elements (Fig. 2) is calculated. As shown in Table 1, in each of the 3 (horizontal) and 6 (vertical) grid elements, the respective values of the poly-dispersity of the bubbles for the three different sets of photographs is averaged and the uncertainty is specified. The photographs shown in Fig.4 only represent one set of photographs.

1.05±0.62

1.14±0.87

0.83±0.33

0.91±0.31

0.39±0.25

0.92±0.23

0.78±0.18

0.28±0.11

0.66±0.11

0.54±0.13

0.22±0.09

0.49±0.06

0.23±0.09

0.22±0.08

0.26±0.12

0.17±0.05

0.16±0.03

0.16±0.02

Table 1. Mean bubble poly-dispersity of the three different sets of foams measured in the 3 (horizontal) by 6 (vertical) grid elements of the foam column as shown in Figure 2.

In this paper, the uncertainty intervals given correspond to the expanded uncertainty of the data providing a level of confidence of 99%.

3. RESULTS AND DISCUSSION 3.1 Data validation Figure 6 shows the distribution of the number of sides per 2D bubble in a simulation with NSD=0.7, which corresponds to a highly poly-dispersed foam. From the experimental results, the average number of sides of bubble is 5.966 ± 0.051 as measured with image analysis. According to Euler’s law, in an infinite (or periodic) foam this average must be 6 and thus the slight discrepancies from this value in the experimental measurements are due to the finite size of the imaged areas.

ACS Paragon Plus Environment

12

Page 13 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

Figure 6. Distribution of number of sides per 2D bubble in the simulation case at NSD=0.7

The Aboav–Weaire law29,30 mathematically formulates that bubbles with fewer sides tend to be surrounded by bubbles with larger numbers of sides, following:

Mn = 6 − a +

6a + µ 2 , n

(1)

where Mn is the average number of sides of neighboring bubbles surrounding a bubble with sides n,

µ2

is the second moment of the number of sides of the bubbles, and a is a coefficient. In Fig.7, the scattered symbols represent the raw simulation and experimental data of the average number of sides of neighboring bubbles, and the continuous curves show fits corresponding to Eq.1. The fitted curves have coefficient values from Eq. 1 of a = 1.190 ± 0.018 (based on the NSD=0.7 case) and

a = 1.171 ± 0.071 for the fitting with experimental data.

ACS Paragon Plus Environment

13

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 26

Figure 7. Average number of sides per neighboring bubble as either predicted by the computer simulation or measured in experiment, and fits according to the Aboav-Weaire law. The simulation data is shifted horizontally by +0.1 sides per bubble.

The fits derived from the Aboav-Weaire law are in close agreement with both the structural simulation and the experimental results.

3.2 Film size related to bubble area ratio According to what was formulated previously1,31, the film size in either the 2D or 3D case has a strong influence on the film drainage rate and hence the likelihood of coalescence of bubbles and thus also the overall foam stability. Therefore a rigorous mathematical formulation for the film size as a function of

ACS Paragon Plus Environment

14

Page 15 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

bubble size is necessary in order to extend the model of Tong et al. (2011)1 and model the stability of bubbles and foam with population balance methods. The film size is predicted as a function of bubble size because the bubble size is the main independent parameter within the population balance modelling. In the work of Tong and Neethling (2010) 6, the film size was found to be most strongly dependent on the size of the smaller bubble, rather than the bigger bubble to which it is attached, with a roughly linear dependency. There was also a secondary dependency on the ratio of the size of the bigger bubble to that of the smaller bubble, with this dependency being virtually independent of either the specific sizes of the bubbles involved or the underlying bubble size distribution. In this paper a similar technique is used, with the film length being normalized by the size (equivalent circular diameter) of the smaller bubble to which it is attached. The equivalent circular diameter is defined as the diameter of a circle that has the same area as that of the bubble of interest. The dependency on the relative size of the bubbles to which the film is attached is assessed in terms of the bubble area ratio, which is defined as the ratio of the area of the bigger bubble to that of the smaller bubble. In a random foam for any given bubble area ratio, there will always be a distribution of possible normalised film sizes, and therefore the dependency on the bubble area ratio must be analysed in terms of shape of this distribution. In order to do this the films from all the simulations are divided into a number of different narrow intervals based on their bubble area ratios. Fig.8 shows the cumulative probability distribution of the normalized film size for a number of these narrow bubble area ratio intervals. Separate curves are shown for each of the levels of underlying polydispersity. It is clear from this figure that while the shape and position of this distribution depends on the bubble area ratio, there is virtually no dependency on the level of the underlying poly-dispersity. It is worth noting that there was less data available for film sizes with a large bubble area ratio due to the

ACS Paragon Plus Environment

15

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 26

low frequency of larger bubbles occurring in the foam. For example, at the level of NSD=0.6 of the simulation results, there were only 355 films available to analyse for an area ratio interval of 8.0 to 10.0 while there were 1329 films in the area ratio interval of 2.4 to 2.6 and 2422 films in the area ratio interval of 1.0 to 1.1. For simulations of coalescence where the level of poly-dispersity may vary both spatially and temporally, this result is very useful as it implies that the probability distribution for the film size involved in a coalescence event depends virtually only on the size of the two bubbles involved and not on the size distribution of the other bubbles in the vicinity.

Figure 8. Cumulative probability distribution of the normalised film size for foams of different underlying polydispersities, grouped into bins of bubble area ratio from computer simulation results. Solid curves represent the bubble area ratio of 1.0 to 1.1, dotted curves represent 2.4 to 2.6 and the dash dot curves represent 8.0 to 10.0.

ACS Paragon Plus Environment

16

Page 17 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

The data from all the simulations at different levels of underlying poly-dispersities can be combined since the cumulative distribution of the film size is virtually independent of the underlying polydispersity of the foam and is only significantly affected by the bubble area ratio. The combined data from the simulations of these specific bubble area ratios can then be compared to the equivalent experimental data (Fig.9a). The experimental and simulation data are in good agreement with one another. The biggest discrepancy is at the largest bubble area ratio, but this is also the experimental data for which the sample size is the smallest. For example, with regard to the situation where the bubble area ratio is 8 to 10, the sample size of the experimental data includes 26 films, whereas for bubble area ratios of 2.4 to 2.6, the sample size of the experimental data is bigger with 151 films. In the experiments, the Plateau borders have a small but finite size and therefore there is a minimum size for the films before they undergo a topological (T1) transformation. Therefore in comparison to the experimental results, the simulation results tend to over predict the number of small films (e.g. normalized film size between 0.3 and 0.6). Both the experimental and simulation results show a systematic stretching of the distribution as the bubble area ratio changes, with the average normalised size of the film between two bubbles increasing as the ratio of the bubble sizes increases. The relationship between the normalized film size of a bubble and the bubble area ratio can be explained using a simple model similar to the one derived previously for the 3D case 6, but now applied in 2D and with a minor update:

Ab 1 / 2 ) Lf As , =K Ab 1 / 2 Ds 1+ ( ) As (

(2)

where Lf is the film length, Ds is the equivalent circular diameter of the smaller bubble that is attached to the film, Ab is the area of the bigger bubble and As the area of the smaller bubble that is attached to the

ACS Paragon Plus Environment

17

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 26

film. The constant value of K=1.1497±0.0043 can be determined by analysing the simulation results from the NSD=0 case, as in a mono-dispersed foam the relation As = Ab holds. The derivation of Eq.2 can be found in the Supporting Information for Publication. The mean size of films of mono-dispersed foam is

Lf Ds

=

K , 2

(3)

because Ab=As holds. The ratio, P, of Eq.3 to Eq.2 is:

Ab 1 / 2 ) As . P= Ab 1 / 2 2( ) As 1+ (

(4)

The factor P gives the correlation between the mean size of films between bubbles of the same size and the mean size of films between bubbles of different size. It is also a general case for the distribution of film size at any poly-dispersity, derived from stretching the film size distribution of the mono-dispersed foam along the axis of the normalized film size according to the ratio of bubble size between neighboring bubbles as:

l fP = l fM / P ,

(5)

where l fP is the size distribution of films of the poly-dispersed foam and l fM is the size distribution of films of the mono-dispersed foam. In order to further prove this correlation, as shown in Fig.9b, the original curves of Fig.9a were stretched along the axis of the normalized film size using the factor of P:

l fP _ stretched = l fP * P ,

(6)

ACS Paragon Plus Environment

18

Page 19 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

where l fP _ stretched is the stretched size distribution of the films of the poly-dispersed foam. In this figure, all the stretched curves (of simulation results and experimental results) almost overlap one another regardless of the different levels of bubble area ratio. They also all overlap with the distribution of film size of the mono-dispersed foam. This means the mathematical formulations suggested in Eqs.2-5 are valid within the large range of bubble area ratios that have been considered. This suggests that the distribution of the film size of a poly-dispersed foam can be predicted by stretching the distribution of the film size of the mono-dispersed foam using the factor given in Eq.4 and the specific value of the bubble area ratio.

Figure 9. Cumulative probability distribution of the (a) original normalized film size and (b) stretched normalized films size according to Eq.6 for different values of bubble area ratio, based on simulation and experimental data.

Moreover, if the normalized film size is averaged over each bin of bubble area ratio, the results from the structural simulation, the experiments, and the analytical solution (Eq.2) can all be directly compared as shown in Fig.10. This shows that the mean normalized average film size over each bin of bubble area ratio is similar across the three different methods. Particularly, the analytical correlation of

ACS Paragon Plus Environment

19

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 26

Eq.2 agrees very well with the related experimental results. Although the structural simulation results tend to underpredict the film size, the uncertainty intervals of the majority of the data points overlap to some extent throughout the entire range of bubble area ratio that was analysed.

Figure 10. Normalized film size averaged over each bin of bubble area ratio for three different methods.

4. CONCLUSION This work has shown that the film size distribution in dry 2D foams is independent of the polydispersity of the foam; instead a strong function of the size of the two bubbles to which the film is attached, particularly that of the smaller bubble. This finding is highly relevant as it can inform the numerical modelling of unstable, coalescing foams and in turn advance our understanding of these systems.

ACS Paragon Plus Environment

20

Page 21 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

A mathematical expression was formulated that describes the relationship between the normalized film size and the bubble area ratio. The resulting analytical correlation allows, for the first time, the prediction of film size distribution in poly-dispersed 2D foams by stretching the known distribution of film size in a mono-dispersed system. Excellent agreement was shown between structural simulations, experimental data and the analytical solution derived here for the distribution of the normalized film size. We expect these results to be instrumental in developing appropriate coalescence models for pseudo-2D foams. Many features of 2D foams have been described in publications, including the number of bubble sides 11,15, the size of Plateau borders 14 , the liquid content 20,21 and the coarsening of bubbles 13,16. This work quantitatively formulates, for the first time, the size of films of poly-dispersed 2D dry foams. The model developed here enables the prediction of film size using only the size of local bubbles. The underlying poly-dispersity or bubble size distribution of the foam is not needed as long as the foam is in the dry limit.

AUTHOR INFORMATION Corresponding Author * E-mail: [email protected] Author Contributions § Mingming Tong and Katie Cole contributed equally.

ACS Paragon Plus Environment

21

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 26

ACKNOWLEDGMENTS The financial support of Rio Tinto, through the Rio Tinto Centre for Advanced Mineral Recovery, is gratefully acknowledged. This work was supported by the Claude Leon Foundation Postdoctoral Fellowship in South Africa.

References (1) Tong, M.; Cole, K; Neethling, S.J. Drainage and stability of 2D foams: Foam behaviour in vertical Hele-Shaw cells. Colloids Surf., A. 2011, 382, 42–49. (2) Kraynik, A M. Structure of random monodisperse foam, Phys. Rev. E. 2003, 67, 031403. (3) Kraynik, A. M.; Reinelt, D. A.; Swol F. Structure of random bidisperse foam. Colloids Surf., A. 2005, 263, 11–7. (4) Kraynik, A. M.; Reinelt, D. A.; Swol. F. Structure of random foam. Phys. Rev. Lett. 2004, 93, 208301 (5) Kraynik, A. M. The structure of random foam. Adv. Eng. Mater. 2006, 8, 900–906. (6) Tong, M.; Neethling, S.J. The size of films in dry foams, J. Phys.: Condens. Matter. 2010, 22, 155109. (7) Newhall, K. A.; Pontani, L. L.; Jorjadze, I.; Hilgenfeldt S.; Brujic, J. Size-Topology Relations in Packings of Grains, Emulsions, Foams, and Biological Cells. Phys. Rev. Lett. 2012, 108, 268001. (8) Monnereau, C.; Prunet-Foch, B.; Vignes-Adler, M. Topology of slightly polydisperse real foams. Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys., 2001, 63, 061402. (9) Matzke, E. B. The three-dimensional shape of bubbles in foam—an analysis of the role of surface forces in three-dimensional cell shape determination. Am. J. Bot. 1946, 33, 58–80.

ACS Paragon Plus Environment

22

Page 23 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

(10) Cheng, H. C. & Lemlich, R. Errors in the measurement of bubble-size distribution in foam. Ind. Eng. Chem. Fundam., 1983, 22, 105-109. (11) Fortes, M. A.; Teixeira, P. C. Bubble size–topology correlations in two-dimensional foams derived from surface energy minimization. J. Phys. A: Math. Gen. 2003, 36, 5161–5173. (12) Cox, S. J.; Graner, F. and Vaz, M. F. Screening in dry two-dimensional foams. Soft Matter. 2008, 4, 1871–1878. (13) Roth, A. E.; Jones, C. D.; Durian, D. J. Coarsening of Two Dimensional Foam on a Dome, Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys. 2012, 86, 021402. (14) Gay, C.; Rognon, P.; Reinelt, D. and Molino, F. Rapid Plateau border size variations expected in three simple experiments on 2D liquid foams. Eur. Phys. J. E. 2011, 34, 2-11. (15) Durand, M.; Kraynik, A.M.; Van Swol, F.; Kafer, J.; Quilliet C., Cox, S.J.; Talebi, S.A.; Graner, F. Statistical Mechanics of Two-Dimensional Shuffled Foams, part II: Geometry - Topology Correlation in Small or Large Disorder Limits. Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys. 2014, 89, 062309. (16) Uplat, J.; Bossa, B. and VILLERMAUX, E. On two-dimensional foam ageing, Journal of Fluid Mechanics, 2011, 673, 147–179. (17) J. Marchalot, J. Lambert, I. Cantat, P. Tabeling and M.-C. Jullien, 2D foam coarsening in a microfluidic system, EPL, 2008, 83, 64006 (18) Saulnier, L.; Drenckhan, W.; Larré, P.-E., Anglade, C.; Langevin, D.; Janiaud, E.; Rio, E. In situ measurement of the permeability of foam films using quasi-two-dimensional foams. Colloids and Surfaces A: Physicochem. Eng. Aspects. 2015, 473, 32–39.

ACS Paragon Plus Environment

23

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 26

(19) Salem, I. B.; Cantat, I. and Dollet, B. Response of a two-dimensional liquid foam to air injection: swelling rate, fingering and fracture. J. Fluid Mech. 2013, 714, 258-282. (20) Yazhgur, P.; Honorez, C.; Drenckhan, W.; Langevin, Dominique. and Salonen, A. Electrical conductivity of quasi-two-dimensional foams, Physical Review E. 2015, 91, 042301. (21) Yazhgur, P. Flows in foams: The role of particles, interfaces and slowing down in microgravity. Soft Condensed Matter, Universite Paris-Saclay, 2015. (22) Cole, K., Brito-Parada, P., Morrison, A., Govender, I., Buffler, A., Hadler, K., Cilliers, J.J. Using positron emission tomography to determine liquid content in overflowing foam. Chem. Eng. Res. Des., 2015, 94, 721-725. (23) Mancini, M.; Guène, E. M.; Lambert, J.; Delannay R. Using Surface Evolver to measure pressures

and energies of real 2D foams submitted to quasi-static deformations. Colloids and Surfaces A: Physicochem. Eng. Aspects. 2015, 468, 193–200. (24) Cole, K.E., Bubble size, coalescence and particle motion in flowing foams. PhD thesis. 2010, Imperial College London. (25) Osei-Bonsu, K.; Shokri, N.; Grassia, P. Fundamental investigation of foam flow in a liquid-filled Hele-Shaw cell. J. Colloid Interface Sci., 2016, 462, 288-296 (26) Hutzler, S.; Cox, S. J.; Wang, G. Foam drainage in two dimensions. Coll. Surf. A, 2005, 263, 178183. (27) Cox, S. J.; Janiaud, E. On the structure of quasi-two-dimensional foams. Philos. Mag. Lett. 2008, 88, 693-701.

ACS Paragon Plus Environment

24

Page 25 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

(28) Schneider, C.A., Rasband, W.S., Eliceiri, K.W. NIH Image to ImageJ: 25 years of image analysis Nat. Methods. 2012, 9, 671-675. (29) Aboav, D.A. The arrangement of cells in a net. Metallography. 1980, 13, 43–58. (30) Aboav, D.A., The arrangement of cells in a net. II. Metallography. 1983, 16, 265–273. (31) Grassia, P.; Neethling S.J.; Cervantes, C.; Lee, H.T. The growth, drainage and bursting of foams. Colloids Surf., A. 2006, 274, 110-124.

ACS Paragon Plus Environment

25

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 26

For Table of Contents Only

ACS Paragon Plus Environment

26