Defoaming: Antifoams and mechanical methods

Current Opinion in Colloid & Interface Science 20 (2015) 81–91

Contents lists available at ScienceDirect

Current Opinion in Colloid & Interface Science journal homepage: www.elsevier.com/locate/cocis

Defoaming: Antifoams and mechanical methods Peter R. Garrett School of Chemical Engineering and Analytical Science, University of Manchester, Oxford Road, Manchester M13 9PL, UK

a r t i c l e

i n f o

Article history: Received 29 December 2014 Accepted 2 March 2015 Available online 10 March 2015 Keywords: Antifoam mechanism Particles Deactivation of oil/particle antifoams Dynamic effects Ultrasound

a b s t r a c t Recent progress in both the mode of action of antifoams and mechanical defoaming is reviewed. New insights concern the simulation of the orientation of particles in interfaces and films, the role of dynamic surface effects in antifoam action, antifoam action under micro-gravity, deactivation of oil/particle antifoams and antifoam action in hydrocarbon media. Progress in mechanical defoaming is mainly confined to new insights concerning the use of ultrasound. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction Defoaming is a surprisingly ubiquitous requirement. In the oil industry, for example, processes such as gas–crude oil separation and desulphurization of natural gas by bubbling through alkanolamine solutions usually require defoaming, which often involves the use of chemical antifoams. Other industrial processes such as the jet dying of textiles, radioactive waste treatment, the kraft pulp process and fermentation using oxygen bubble columns are also beset by unwanted foam. Again treatment with antifoams is often used but mechanical means are sometimes employed in the case of bubble columns. The control of flotation froth in mineral processing can be adversely affected if hydrophobed mineral particles too readily destabilize the froth. Formation of undesirable foam can also be an aspect of the use of certain formulated products such as waterborne latex paints, and detergents for use with machine washing of textiles and dishes. This problem usually necessitates the addition of antifoams to the relevant formulations. However some products, such as shampoos and hand dishwashing liquids, are designed to produce copious amounts of foam during use. These products must be formulated to minimize the antifoam effects of the triglyceride soils which are intrinsic to their application. There are even medical applications of defoaming. These include the use of antifoams to treat gastrointestinal gas, to eliminate any foam obscuring the view of colonoscopy cameras and in filters for removal of foam bubbles from the aspirated blood derived from surgery in order to permit recirculation. Defoaming is therefore clearly widely relevant in many contexts and does in fact present many interesting challenges to fundamental scientific understanding. This is however not always recognized

E-mail addresses: [email protected], [email protected].

http://dx.doi.org/10.1016/j.cocis.2015.03.007 1359-0294/© 2015 Elsevier Ltd. All rights reserved.

in academic circles. A recent foam conference for example, included only one paper explicitly concerning defoaming [1] out of a total of over 100 oral and poster presentations. Despite this seeming lack of recent interest much progress has been made in understanding the mode of action of antifoams over the past thirty years. A detailed account of that progress can be found in a recent monograph [2⁎⁎] which is concerned with defoaming by both antifoams and mechanical means. Briefer accounts of defoaming by antifoams have also been published recently by Denkov et al. [3⁎] and Karakashev and Grozdanova [4]. Another recent review by Owen [5] is distinct in that it ignores much of the progress made over the past twenty years. Earlier noteworthy reviews include those by Miller [6] and Denkov [7⁎⁎]. Use of antifoams always implies contamination of the system to be defoamed. Such contamination is, however, sometimes unacceptable. Examples include preparation of pharmaceuticals by fermentation and downstream processing in petrochemical plants where catalysts may be poisoned by antifoam residues. In these situations recourse is sometimes made to defoaming by mechanical means, utilizing for example either ultrasonic or various rotational devices. Progress in understanding the mode of action of such devices has however been rather limited. A complete review is to be found in the recent monograph [2⁎⁎]. Developments in the understanding of the mode of action of antifoams, which have been published over the past 5 years or so, are the main concern of this review. Earlier work, however, is also cited in order to provide relevant context. Antifoams are defined here as particles, oils or mixtures of oils and particles which reduce either the foamability or foam stability of the liquid in which they are dispersed. Some account of defoaming by mechanical means is also included where however a dearth of recent relevant publications means a review largely confined to ultrasonic defoaming.

82

P.R. Garrett / Current Opinion in Colloid & Interface Science 20 (2015) 81–91

2. Defoaming by particles 2.1. Hydrophobic particles; theory It is now well known that finely divided particles, which are sufficiently hydrophobic, can produce antifoam effects when dispersed in aqueous surfactant solution [2⁎⁎,3⁎,8]. Whether these effects are apparent is dependent upon the contact angle and the shape of the particles. In the case of spherical particles antifoam effects are only present if the contact angle (measured through the aqueous phase), θAWN ~90° [2⁎⁎, 3⁎,8]. As the film drains the particles bridge the film, and dewet (now often called the “bridging–dewetting mechanism” after Denkov [7⁎⁎]) leaving a hole in the foam film which subsequently expands to cause film rupture. The essentials of this mechanism have been verified experimentally in model experiments for spheres using high-speed cinematography [8]. Joshi et al. [9⁎] present interesting observations concerning the antifoam mechanism of hydrophobic fatty alcohol particles in aqueous surfactant solution. The fatty alcohols consist of a blend of C14 –C30 chain lengths. The particles appear to have a spherical or ellipsoidal geometry and should therefore require contact angles N 90° in any given aqueous medium for antifoam effect. However in the absence of surfactant the particles possessed advancing and receding contact angles of 93° and 95° respectively and in the relevant aqueous surfactant solution 85° and 71° respectively. The effect of these particles on the stability of the film formed between two colliding bubbles was observed by videomicroscopy. In these experiments the fatty alcohol particles were dispersed in an aqueous concentrated ethoxylated alcohol solution containing a thickener and preservative — the resultant concoction often being described commercially as an aqueous or surfactant antifoam. This concoction was then added to an aqueous solution of a second surfactant in order to achieve an antifoam effect for the latter. In this arrangement one bubble was allowed to equilibrate with the antifoam dispersion so that the fatty alcohol particles could attach to the relevant air–water surface. The second bubble was then grown towards the first to form a film between the two bubbles. The subsequent flow of liquid from the film produced a surface tension gradient and effectively removed the particles. However when that flow ceased, as the second bubble closely approached the first, a reverse flow occurred because the surface tension of the air–water surfaces in the film was now higher than that on the remaining surface of the first bubble. The resulting Marangoni effect dragged the antifoam particles back into the film which then rapidly ruptured by a bridging mechanism. Joshi et al. [9⁎] attribute the origin of the reverse flow to desorption of surfactant from the fatty alcohol particles. However spreading from the shorter chain length fatty acids in the blend used here could also contribute (the equilibrium spreading pressure of, for example octadecanol – a component of the blend used – will lower the surface tension of water at the relevant temperature to some 20 mN m−1 below that of the solutions of the second surfactant considered here [10]). Clearly a dynamic process is occurring which probably means that dynamic contact angles are relevant which could explain the apparent discrepancy between the measured receding angles and the requirement that θAWN ~ 90°. There are in fact two main difficulties with generalizing the condition θAWN ~ 90° to all foamability measurements. The first concerns measurement of the contact angle under conditions directly relevant for those actually existing during foam generation. As the work of Joshi et al. [9⁎] suggests contact angle hysteresis and dynamic effects due to the rate of surfactant transport relative to the rate of air–water surface generation can conspire to produce antifoam effects even though the receding contact angle b 90° [2⁎⁎]. However arguably a more significant factor concerns particle geometry. The presence of sharp edges and rugosities appears to confer antifoam effects at contact angles significantly b 90° [2⁎⁎]. Orthorhombic particles have received particular recent interest in this context following the work of Dippenaar [8] more than thirty

years ago. Dippenaar [8] investigated the rupture of single aqueous films by orthorhombic hydrophobed galena particles using high-speed cinematography. He attributed his finding that aqueous film rupture by such particles occurs with contact angles of only (80 ± 8)° to the particle shape. He concluded that such particles could adopt two orientations at the air–water surface as that surface is pinned to edges as shown in Fig. 1a and b. According to Dippenaar [8] only the diagonal orientation can give rise to foam film rupture by a mechanism analogous to that for a sphere but in this case, for a particle of a square cross-section, contact angles must lie in the range 45° b θAW b 135°. A problem with this argument concerns the assumption that only two orientations of an orthorhombic particle at a planar air–water surface are possible. That is strictly only true if the aspect ratio ≫ 1. In the case of the diagonal orientation the air–water surface must satisfy the contact angle against the two particle end surfaces perpendicular to the plane of the paper as shown in Fig. 1c and d. If for example the air–water surface remote from the particle is planar then the capillary pressure across that surface must be zero. For equilibrium the capillary pressure in the segment of air–water surface against the particle ends must also be zero. That is only possible if it forms a catenoid element, which must also contact everywhere the perpendicular surface of the particle at the relevant contact angle. In the case of a particle with an aspect ratio close to unity the contribution of the surface energy of the two particle end surfaces, including the catenoid elements, to the total work of adhesion of the particle to the air–water surface must be significant. This factor therefore greatly complicates calculation of the relative work of adhesion and probabilities of the two orientations depicted in Fig. 1a and b in the case of aspect ratios close to unity. It also suggests the possibility that yet other orientations are possible. Calculating the relative work of adhesion for orthorhombic particles as a function of contact angles has been made tractable by use of an iterative surface energy minimization technique described by Morris et al. [11–13,14⁎⁎]. The technique makes use of the Surface Evolver software developed by Brakke [15]. A more detailed description of the use of this technique is given in the accompanying review by Morris et al. [16]. It permits calculation of the (surface) energy profile of a particle in an air–water surface or foam film as a function of the orientation, contact angle and shape. The profile reveals the presence of energy minima, consistent with stable orientations, where the depth of the minima gives an indication of their relative probabilities. This analysis reveals, for example, four possible orientations for an orthorhombic particle bridging a foam film [14⁎⁎]. These are designated vertical, horizontal, rotated and diagonal and are shown in Fig. 2. The relative stabilities of these orientations are determined by both the contact angle and the aspect ratio of the particle. This is exemplified in Table 1 for the aspect ratios corresponding to the only direct experimental observations of the rupture of aqueous films by orthorhombic particles. Careful examination of the cinematographic film frames of Dippenaar [8] reveals that the aspect ratio of the orthorhombic hydrophobed galena particle is about 1.4. With a contact angle in the range of 72–88° (i.e. 80 ± 8°) the results of Table 1 would suggest that either both horizontal and diagonal orientations are likely to co-exist or that the diagonal orientation alone is to be present. Dippenaar [8] suggests that his observations are in fact consistent with the former possibility. However the relevant cinematographic frames do suggest a reconfiguration to a rotated orientation immediately prior to rupture of the aqueous film. Morris and Cilliers [13] have recently repeated these observations of Dippenaar [8] with a similar hydrophobed galena particle of contact angle in the range of 70–90° (i.e. 80 ± 10°) but of aspect ratio unity. As expected from the calculations summarized in Table 1 this particle adopted a rotated orientation although with no apparent observations of either horizontal or vertical orientations. Morris et al. [11–13,14⁎⁎] have extended these surface energy minimization calculations to estimate the critical capillary pressure required to rupture an aqueous film containing a bridging orthorhombic particle. In particular Morris and Cilliers [13] have compared such calculations

P.R. Garrett / Current Opinion in Colloid & Interface Science 20 (2015) 81–91

83

air

water

a

film thins horizontal orientation

air water

b

film thins diagonal orientation;

catenoid airwater surface

aqueous phase peels off particle to form a hole

planar air-water surface hinging on an edge

air water

c

d

Fig. 1. Orientation of an orthorhombic particle in a foam film. (a) and (b) Horizontal orientation stabilizes a draining foam film whereas a diagonal orientation leads to foam film rupture provided θAW N θp = 45o according to Dippenaar [8] for the case of a particle with a square cross-section and an aspect ratio ≫ 1 (i.e. a 2D view). (c). Mechanical equilibrium requires that the capillary pressure over the whole air–water surface be zero which must mean a catenoid profile over the end surfaces of a diagonally oriented particle which become progressively more significant as the aspect ratio decreases. (d). Vertical slice through centre of particle where the contact angle is satisfied and the air–water surface is concave, the orthogonal radius of curvature must then be convex if the surface is to form a catenoid as shown in (c).

with experimental observations which supposedly repeat those of Dippenaar [8] but, as we have seen, using a hydrophobed galena particle of aspect ratio unity rather than 1.4. The calculations indicate that the particle, placed in a rotated orientation but bridging the film, maintains that orientation as capillary pressure is increased and the film thins until it ruptures. These calculations suggest that film rupture does not occur as a result of the film liquid peeling off the particle surface as indicated in Fig. 1b and apparently revealed for an orthorhombic particle by the cinematographic observations of Dippenaar [8]. Morris and Cilliers [13] rather suggest the particles induce rupture directly in foam films where “its highly distorted shape around the particle is what forces the opposite sides (of the film) together”. Unfortunately the experimental observations were not able to convincingly verify this conclusion. Moreover strict comparison with Dippenaar [8] was not possible because of selection of an aspect ratio of unity rather than 1.4. The surface energy minimization technique used in these calculations of the critical capillary pressure is not able to account for the positive contribution to the disjoining forces, which will generally facilitate

stabilization aqueous films of solutions of surface active materials against rupture. Application of this technique to the actual process of film rupture by a bridging particle for comparison with the experiments of Morris and Cilliers [13] and Dippenaar [8] is nevertheless justified because the aqueous films in that context consisted of pure water — no surfactant at all was present and therefore a positive contribution to the disjoining force was absent. However the presence of significant positive disjoining forces due to the presence of surfactant is likely to suppress the process of foam film rupture by the indirect influence of a bridging particle on the neighbouring film. After all it is that property of surfactants which justifies the addition of antifoams at all! It probably means that in those circumstances foam film rupture would have to occur by the mechanism originally suggested by Dippenaar [8] where the film liquid peels off the particle until the two three phase contact lines become coincident as shown in Fig. 1b. However as described by Garrett [2⁎⁎] application of that mechanism to particles of complex geometry requires knowledge of their orientations at air–water surfaces and in foam films. The surface energy minimization technique may

84

P.R. Garrett / Current Opinion in Colloid & Interface Science 20 (2015) 81–91

b

a

d

c

Fig. 2. Possible stable orientations of an orthorhombic particle bridging a thin liquid film (of thickness ≪ dimensions of the particle). (a). Horizontal (H), (b). Diagonal (D), (c), Rotated (R), (d). Vertical (V). Orientations calculated by a surface energy minimisation technique using the Surface Evolver software [14⁎⁎,15]. The catenoid surface suggested by Fig. 1c is clearly revealed in the diagonal orientation (b). Reprinted from Morris, G., Neethling, S.J., Cilliers, J.J. J. Colloid Interface Sci. 361, p373; copyright 2011, with permission from Elsevier.

therefore potentially yield valuable insights in that context especially for particles of complex crystal habit. Such considerations are important not only for rupture by particles of air–water–air foam films but also for the rupture by particles of oil–water–air pseudoemulsion films which represents an important aspect of the mode of action of commercial antifoams for application in aqueous surfactant solution [2**]. 2.2. Effect of precipitation of crystalline and mesophase particles on foam Addition of counterions to aqueous solutions of anionic surfactants can result in partial substitution of monomers or micelles by precipitates of either crystalline or mesophase particles. This change in physical state of the relevant solution usually produces a marked reduction in foamability. For example calcium soaps exhibit extremely low solubilities so that they may be precipitated from aqueous mixed solutions of sodium soaps and other surfactants. The relevant precipitates are crystalline and can act as antifoams for solutions of the latter. Finite contact angles suggest that they function by a bridging–dewetting mechanism Table 1 Stable orientations of an orthorhombic particle in a thin aqueous film as a function of contact angle and aspect ratio calculated by surface energy minimization. H = Horizontal, D = Diagonal, R = Rotated, and V = Vertical (see Fig. 2). From Morris et al. [14⁎⁎]. Aspect ratio

1.0 1.36

Contact angle, θAW/degrees 45

60

65

70

75

80

90

H/V H/V

H/V H/V

H/V H/D/V

H/R/V H/D

H/R/V H/D

R D

R D

[2⁎⁎]. Similarly addition of excess counterion to solutions of sodium dodecyl sulphate lowers the Krafft temperature to produce a crystalline precipitate. That precipitate also exhibits a finite contact angle [17] against the saturated solution and appears to function by a bridging– dewetting mechanism as an antifoam [2⁎⁎,18]. In contrast to single chain surfactant molecules, exemplified by carboxylate soaps and SDS, addition of counterions to double chain molecules tends to precipitate mesophase dispersions – often as multi-walled vesicles of lamellar phase. The precipitates formed upon addition of Ca2 + ions to aqueous solutions of sodium alkyl benzene sulphonate [19,20] represent an example of this behaviour. Decreases in foamability of the solution accompany formation of the precipitate. The residual foam is however stable. The defoaming effect accompanying addition of Ca2+ ions to solutions of sodium alkyl benzene sulphonate can find practical application in detergency by the elimination of residual foam during rinsing with hard water. Removal of the lamellar phase precipitate by filtration does not however restore the foamability [19,20] which means that the precipitate does not function as an antifoam. This is probably a consequence of the structure of the multi-walled vesicles which are formed upon precipitation of lamellar phase. The outer layer of the vesicle formed in an aqueous environment has to be covered with surfactant head groups if it is composed of bilayers. That surface is therefore hydrophilic and is extremely unlikely to posses the contact angle of 90° required for antifoam action by a sphere — indeed it is likely to be nearzero. Measurements of the dynamic surface tensions indicate that the reduction of foamability simply accompanies replacement of labile

P.R. Garrett / Current Opinion in Colloid & Interface Science 20 (2015) 81–91

micelles with non-labile mesophase particles which are larger (with smaller diffusion coefficients and slower breakdown kinetics) [2⁎⁎]. This is illustrated by a comparison of the effect of addition of Ca2 + ions on the foam and dynamic surface tension behaviour of a micellar solution of sodium dodecyl 4-phenyl sulphonate shown in Fig. 3. The figure reveals a marked increase in dynamic surface tensions with the onset of precipitation of the mesophase. In turn this suggests slow transport of surfactant to air–water surfaces, which will contribute to low adsorption levels, low positive contributions to disjoining forces and high capillary pressures in Plateau borders. These factors will all conspire to produce lower foam film stability during foam generation. However after foam generation has ceased surfactant can slowly adsorb to ameliorate these factors ensuring the stability of any foam which survives the process of generation. All of this suggests the generalisation that precipitation of crystalline or amorphous crystalline particles by addition of counterions to aqueous solutions of anionic surfactants may produce antifoam effects whereas precipitation of mesophase is unlikely to do so. However precipitation in either case means that labile entities are replaced with non-labile entities. In turn this means that the rate of transport of surfactant to the rapidly expanding surfaces present during foam formation will be reduced. This factor will always contribute to diminished foamabilities. In the case of crystalline precipitates it will even enhance

Volume air in foam/cm3

Clear micellar region

Lamellar phase precipitate region

100 80 60 40 20 0 -4

-3

-2 2+

Log10 ([Ca ]/M)

a

the likely antifoam effects. We should however emphasize that these generalisations are drawn from relatively few examples. There is clearly a need to extend these observations to other systems and include more detailed characterization of the nature of the precipitates. 3. Antifoam effects by neat oils 3.1. Oil drops in aqueous foam films; definitions and theory There is some confusion, even in recent literature, about the exact criteria for effective defoaming action by liquid antifoams. Here we give an abridged outline using an antifoam oil–water combination as an example — with some important exceptions the same basic criteria apply to other fluid–fluid systems. Full details are to be found elsewhere (see for example reference [2⁎⁎]). The first requirement for effective antifoam action is that the solubility limit of the antifoam liquid in the foaming liquid should be exceeded (there are however some exceptions to this generalisation — see for example Section 3.3 and [2⁎⁎]). The second requirement is that the oil be dispersed as drops, sufficiently small and numerous to ensure presence in foam films. The third requirement is that the oil drops spontaneously emerge into the air–water surface. This means that the entry coefficient should be positive — a thermodynamic criterion. However the aqueous “pseudoemulsion” film separating an oil drop from the air is usually metastable and must first be ruptured by for example, particles of a suitable wettability, if emergence of the drop is to occur [2⁎⁎,3⁎]. Once the oil has emerged into the air–water surface there are essentially three different phenomena which can occur. In the simplest case the oil can form lenses without in any way contaminating the air– water surface — so-called partial wetting (using the nomenclature of Brochard-Wyatt et al. [21]). Alternatively the oil may spread to contaminate the air–water surface to form an unstable film which subsequently breaks up to leave oil lenses in equilibrium with an air–water surface contaminated with oil — so-called pseudo-partial wetting [21]. Finally the oil may spread to form a duplex film with no lenses. This film is so thick that the oil–air and oil–water surface tensions of the film have the values of the bulk phase — so-called complete wetting [21]. In each of these three cases the equilibrium spreading pressure, Se, must satisfy Eq. (1) where we can only have [22] e

e

−2σ OW ≤ S ≤0

Precipitation boundary

65 55 45 35 25



b

-3.5

-3

-2.5

ð1Þ

where σeOW is the oil–water surface tension. Eq. (1) means that (at equilibrium) Se can only be either zero or negative. However in the case of both pseudo-partial and complete wetting the oil must first spread over the air–water surface before equilibrium is attained. In these cases a transient “initial” positive spreading pressure prevails, defined by non-equilibrium values of the relevant surface tensions. An emerged oil drop, for which Eq. (1) is satisfied, may actually bridge across both sides of an aqueous film. It was shown more than thirty years ago [23] that if such an oil bridge forms in a plane parallel foam film then it will be unstable if the angle θ* formed between the air–water and oil–water surfaces is N π/2. That angle, θ*, is determined by Neumann's triangle of surface forces so that we must have

75

-4

85

-2

Log10 ([Ca2+]/M)

Fig. 3. Effect of precipitation of lamellar phase by addition of Ca2+ to a solution of 2 mM sodium 4-phenyl benzene sulphonate in 17 mM NaCl at 25 °C (a). Foamability by Bartsch method (cylinder shaking by hand — see [2⁎⁎] for details). (b). Dynamic surface tensions at surface ages; ▲, 0.1 s; ■, 1 s; ♦, 10 s. [19,20].

π=2 b θ ≤π

ð2Þ

where θ* = π represents complete wetting of the aqueous film by a duplex oil film. It follows from Neumann's triangle that we may rewrite the Eq. (2) as e

e

0 b B ≤ 2σ OW σ AW

ð3Þ

86

P.R. Garrett / Current Opinion in Colloid & Interface Science 20 (2015) 81–91

where B is the bridging coefficient defined by e

2

e

2

e 2

B ¼ σ AW þ σ OW − σ AO

ð4Þ

where σeAW and σeAO are respectively the equilibrium air–water and air– oil surface tensions. Here we should note that if the oil exhibits a positive initial spreading coefficient, then mechanical equilibrium does not exist between the three surface tensions while that transient condition prevails and Neumann's triangle is violated. The bridging coefficient is then meaningless with values higher than the upper limit of Eq. (3). This restriction has not always been recognized [7⁎⁎,24–27]. There is much experimental evidence that the antifoam action of an oil in an aqueous surfactant solution is largely confined to situations where the bridging coefficient is positive [2⁎⁎,3⁎]. It should be stressed also that Eqs. (2) and (3) account for antifoam effects by oils exhibiting all types of possible wetting behaviour — complete wetting, pseudopartial wetting and partial wetting. Positive initial spreading coefficients then simply imply a transient process of rapid supply of oil lenses or duplex film to foam films so that unstable liquid bridges can be formed. Denkov et al. [3⁎,7⁎⁎] have proposed that unstable bridging drops will lead to foam film rupture by either a bridging–stretching mechanism or by a bridging–dewetting mechanism. In the bridging–stretching mechanism the drop is stretched until the two oil–air surfaces touch whereupon a hole forms and the drop is to torn apart to rupture the film. By contrast the bridging–dewetting mechanism is supposed to occur if the aqueous film peels off the drop in a manner similar to that for a solid spherical hydrophobic particle [3⁎]. It would seem possible that this mechanism could prevail if the viscosity of the oil is high enough. Recently Denkov et al. [3⁎] claim to have observed bridging– dewetting in the case of non-spreading mineral oils. The results are however as yet unpublished. Theoretical challenges here concern the exact nature of the process by which a stretching oil drop ruptures a film — what are the consequences for the drop, does it disproportionate into several smaller drops as the foam film ruptures? What is the exact role, not only of capillarity, but oil viscosity in the process and is the bridging–dewetting mechanism likely to prevail at sufficiently high viscosities? There is also some evidence that the magnitude of the bridging coefficient (or θ*) can determine relative antifoam effectiveness even when the Eq. (3) is satisfied [28] for which theory has no convincing explanation. A recent study by Gao et al. [29] attempts to simulate the bridging– stretching process in the case of an oil drop in a foam film formed from an aqueous surfactant solution by recourse to a “steered molecular dynamics” approach. Essentially this approach supplements classical molecular dynamics with force probes to simulate processes which are therefore only partially accounted for at a molecular level. In reality the relevant forces are capillary and disjoining pressures which are “mimicked” by arbitrary forces in this simulation, the molecular origins of which are not revealed. In consequence it is not possible to link this simulation to quantitative experimental observation. Also the simulation produces only a two-dimensional view of the process little better than the schematic diagrams to be found in the literature [2⁎⁎,3⁎]. None of the issues listed in the previous paragraph have therefore been addressed (or even can be addressed by this kind of approach?). Perhaps a more fruitful approach would be the use of the surface energy minimization as exemplified by that described in Section 2.1 for the case of solid particles at interfaces. 3.2. Comparison of the criteria for antifoam action by neat oils in aqueous and non-aqueous liquids Realization of antifoam effects in aqueous solutions of surface active compounds by liquids which satisfy that Eq. (3) is inhibited by the metastability of oil–water–air pseudoemulsion films. That metastability largely derives from the electrostatic and steric contributions to the relevant disjoining pressure isotherms due to adsorbed surfactant. It can

however be overcome in the case of drops trapped in Plateau borders by the high capillary pressures attained as foam drains, producing slow declines in the overall stability of foam [2⁎⁎]. Whether the resulting foam film rupture, occurring in or at the edge of the Plateau border, is determined by the bridging coefficient is however not yet established [2⁎⁎]. More commonly, however, rupture of pseudoemulsion films is achieved in the case of commercial antifoams by adding hydrophobed particles to the oil. Those particles supposedly rupture the pseudoemulsion films by a bridging mechanism similar to that by which hydrophobic particles rupture aqueous foam films (for full accounts of this see for example [2⁎⁎,3⁎,7⁎⁎]). Other possibilities are sometimes claimed to exist however. Shu et al. [27] for example show that addition of acid to an aqueous solution of sodium oleate precipitates oleic acid drops and causes reduction of foamability and foam stability. They claim that this is due to a bridging antifoam effect by the oleic acid drops. The relevant pseudoemulsion film is supposed intrinsically unstable at low pH due to low levels of adsorption of charged species at the oleic acid–water interface. The resulting absence of electrostatic effects would therefore imply that particles are unnecessary in this case. However Shu et al. [27] have not shown that removal of the oleic acid drops causes an increase in foamability and therefore actual antifoam behaviour in this case is not proven. The effect could be due entirely to transport effects deriving from the change in concentration and nature of the remaining dissolved surface active species after the addition of the acid. We should note also that Shu et al. [27] present positive values of both spreading and bridging coefficients in violation of Neumann's triangle so that the quoted bridging coefficients are meaningless. It has recently been shown that the defoaming of crude oil by suitable neat liquid antifoams appears to involve a bridging mechanism characterized by negative values of spreading coefficients and positive values of bridging coefficients [30⁎]. Thus the available evidence [30⁎] suggests that the criteria represented by Eqs. (1)–(3) are still relevant for this non-aqueous system (making the substitutions of σeGL for σeAW; σeDL for σeOW; σeGD for σeAO where σeGL, σeDL and σeGD are the gas–crude oil, antifoam–crude oil and gas–antifoam surface tensions respectively). We therefore have for the formation of unstable bridging configurations by a neat antifoam liquid in this non-aqueous context e

e

0 b B ≤ 2σ DL σ GL :

ð5Þ

However despite this overall similarity the differences in physical properties between water and hydrocarbon liquids produce some important differences in the criteria for selection of antifoam liquids. The dielectric constant of hydrocarbons is nearly two orders of magnitude less than that of water which means that electrostatic effects derived from charge separation of any solutes are essentially absent as shown by Shearer & Akers [31] in the case of lubrication oils. Moreover adsorption of solutes at the air–hydrocarbon surface is expected to be small [2⁎⁎,32]. In turn all this means that pseudoemulsion film stability is likely to be low in this context and that therefore formulation of antifoams with particles is unnecessary, as revealed by normal commercial practice. On the other hand the solubility of typical antifoam liquids such as polydimethylsiloxanes and perfluoropolymers in hydrocarbons is usually several orders of magnitude higher than in water (with polydimethylsiloxane solubilities in crude oil of b10−3 g dm−3 and in water of b10−6 g dm−3 [2⁎⁎]). Even polydimethylsiloxanes, when present at concentrations below their solubility limit act as pro-foamers for hydrocarbons [30⁎,31–33]. Effective foam control is then largely determined by solubility of such compounds. The study by Mansur and coworkers [30⁎,33] of the antifoam behaviour of various alkoxylated polydimethylsiloxanes in crude oil is illustrative. Here the solubilities in the crude oil are reduced by incorporation of the alkoxy groups in the polydimethylsiloxane chains. The most weight effective of these molecules is the least soluble in the crude oil — differences in antifoam effectiveness at equivalent concentrations above the solubility limits are

P.R. Garrett / Current Opinion in Colloid & Interface Science 20 (2015) 81–91

not apparent with similar values of σeGD, positive values of bridging coefficients and negative spreading coefficients [30⁎]. 3.3. “Cloud point” antifoams Ethoxylated and propoxylated nonionic surfactants and polymers exhibit high solubilities in water at low temperatures and partial miscibility at higher temperatures. Above a critical temperature, micellar solutions undergo a phase change to form two conjugate micellar solutions — one dilute and the other relatively concentrated. This process is accompanied by the onset of turbidity where the cloud point is the temperature at which droplets of the concentrated conjugate phase separate at a given overall surfactant concentration. Those droplets readily act as antifoams for solutions of the dilute phase by a bridging mechanism apparently characterized by intrinsically unstable pseudoemulsion films (see [2⁎⁎] for a review of this phenomenon). Addition of ethoxylated and propoxylated nonionic compounds to aqueous solutions of other surface active compounds can also produce antifoam effects at temperatures above the cloud point. However the cloud point may be increased or even eliminated in such systems as a result of the formation of mixed micelles. Nevertheless compounds such as polyethoxy–polypropoxy block copolymers are often described as “cloud point” antifoams. Despite this nomenclature Nemeth et al. [34] have shown that such compounds can exhibit antifoam effects when added to aqueous solutions of a protein–bovine serum albumin at low concentrations so that the cloud phase is absent. In such solutions the polymers simply displace the protein from the air–water surface to produce low foamability much as would be the case in the absence of the protein. Marinova et al. [35] recently report similar observations with this type of polymer but with a different protein–sodium caseinate. However these observations include some curious features. Firstly increasing the concentration of two of the polymers studied, up to concentrations where cloud phase is present, is reported to actually increase the foam height. Since the activity of the polymer will be at a maximum at phase separation it is difficult to see how that could decrease any antifoam effect concerning displacement of protein from the air–water surface. Another interesting feature concerns the effect of addition of Ca2+ to the chosen system but with the polymer concentration high enough for cloud phase drops to be present. This apparently produces a precipitate of CaCO3, particles of which seem to be hydrophobed so that they adhere to the cloud phase drops. The presence of such particles appears to be essential for the cloud point antifoam effect in this case because in the absence of Ca2 + there is apparently little such effect. This then appears to represent an analogue of the well-known oil-particle antifoam synergy. In turn this suggests that the pseudoemulsion films present in these systems are intrinsically relatively stable in contrast to those formed in the case of aqueous dispersions of cloud phase in the absence of any additional surface active species. A detailed study of these issues would appear to be justified in order to provide a firm basis for the optimisation of these cloud point antifoams which find wide application in dish washing formulation, sugar beet processing, and fermentation. 4. Mixed Oil-particle antifoams for aqueous solutions 4.1. Mode of action It is now well-known that oils which exhibit positive bridging coefficients in aqueous surfactant solutions are relatively ineffective as antifoams unless mixed with suitable hydrophobic particulate promoters [2⁎⁎]. The role of the latter concerns reduction of the critical capillary pressure required to overcome the disjoining forces in the oil–water– air pseudoemulsion films to permit film rupture by bridging oil drops [2⁎⁎,3⁎,7⁎⁎]. A detailed review of the present level of understanding is given in reference [2⁎⁎]. However there remain serious challenges to both experiment and theory, particularly concerning the role of particle

87

geometry, contact angle and even particle size. Here we should note that even the orientation of particles of simple geometry at fluid–fluid surfaces has been shown by surface energy minimization to be surprisingly complex (see Section 2). 4.2. Transient dynamic effects Another aspect of antifoam function concerns the likelihood that the rate of surfactant transport to the rapidly expanding air–water surfaces formed during foam generation is not fast enough to maintain equilibrium. This factor can give rise to high dynamic surface tensions during foam generation and can therefore produce positive transient bridging coefficients which become negative as surface tensions decline and trend towards equilibrium values after foam generation ceases. Similarly high dynamic surface tensions imply low dynamic adsorption levels and therefore diminished contributions to the magnitude of disjoining forces. In turn this could reduce entry barriers and facilitate emergence of antifoam drops into the air–water surface. Approach to equilibrium may subsequently lead to enhanced positive disjoining forces, failure of drops to emerge into air–water surfaces and ineffective antifoam action. Examples are known where the dynamic effect on the bridging coefficient appears to dominate. Li Ran [20] for example has recently reported negative bridging coefficients using equilibrium surface tensions for a tristearin/triolein antifoam in sodium dodecyl 4-phenyl sulphonate solutions – antifoam effects are only apparent under dynamic conditions – the foam becomes stable after foam generation ceases. An obvious difficulty with these dynamic phenomenon concerns the actual state of the air–water surfaces during foam generation. Calculation of bridging coefficients, for example, requires knowledge of the air–water surface tension under dynamic conditions. Although dynamic surface tensions can be readily measured as a function of surface age, in general, we have no clear knowledge of the actual effective surface age of the air–water surfaces during foam generation. Estimates have however been made, Garrett and Moore [28] have for example set the surface age during foam generation by Ross Miles technique as 0.5–1 s. Denkov et al. [3⁎] estimate that surface ages during typical foam generation as 0.1–1 s with a characteristic time for Ross Miles of 0.1–0.2 s. Differences in characteristic time with different foam generation techniques mean that antifoam effectiveness can even vary depending upon the technique as shown by Marinova et al. [1]. It would seem that enhanced dynamic antifoam effects are to be expected in decreasing order; hand-shaking cylinders (the Bartsch method) N RossMiles N bubble columns (see [2⁎⁎] for a description of these methodologies). The key challenge here is of course direct measurement of the surface tension of the upper air–water surface of foam during generation for each of these techniques. Transient antifoam effects accompanying foam generation have also been shown to be apparent when foam is generated under microgravity. This behaviour has been observed during a comparison of foam behaviour under micro-gravity in the International Space Station with that under earth gravity [36⁎]. This study included the effect of a polydimethylsiloxane/silica antifoam on foam generation from aqueous sodium dodecyl sulphate solutions using the Bartsch method [2⁎⁎]. Addition of this antifoam produced foam which totally collapsed in 2 s under Earth gravity using the selected conditions. That any foam generated rapidly collapsed under gravity implies that this antifoam is effective under both dynamic and the near equilibrium conditions prevailing after foam generation has ceased. In contrast significant amounts of stable foam were generated under micro-gravity. Not surprisingly Caps et al. [36⁎] attribute the presence of stable foam under microgravity to the absence of gravity-driven drainage. Gas volume fractions of this residual foam were so low that foam films would not exist (i.e. with volume fractions of air ≤ ~ 0.72 [2⁎⁎]) — inter-bubble distances would be so thick that antifoam drops would not be able to form bridging configurations leading to foam collapse. However the

88

P.R. Garrett / Current Opinion in Colloid & Interface Science 20 (2015) 81–91

total volume of foam generated in the presence of the antifoam was significantly less than that generated under gravity in the absence of antifoam. Clearly some foam films must have been formed during shaking of the relevant vessels to permit transient antifoam action to occur under micro-gravity despite the absence of gravity driven drainage.

Z

4.3. Antifoam deactivation

V F ðt Þ ¼ V G

That polydimethylsiloxane (PDMS)/hydrophobed silica antifoams gradually deactivate during the processes of foam generation (and even during emulsification) is well-known [2⁎⁎]. It has been shown by Denkov and co-workers [3⁎] that this deactivation appears to be driven by disproportionation of antifoam oil drops containing particles into a population of inactive particle free drops and particleenriched drops and agglomerates, where the latter can assume sizes of the order of mms. A detailed review of this issue is to be found in [2⁎⁎]. It has been shown that the process of disproportionation cannot simply concern a random distribution of particles across drops [2⁎⁎] – that could in any case not account for the formation of large silica rich agglomerates. Some process of coalescence is clearly implicated – indeed the presence of hydrophobed silica could facilitate that process [2⁎⁎]. It has been suggested [2⁎⁎] that continuous drop splitting and coalescence could produce the disproportionation where the higher viscosity of particle-rich drops inhibits splitting and facilitates coalescence. That drop splitting is also important is implied by overall decreasing rates of deactivation with increasing viscosity of the PDMS [37]. Perhaps a rule-based simulation could be based on these observations to provide a clear explanation for the deactivation process. An additional complication however concerns the behaviour of other oil/particle antifoams. There is some evidence for example that hydrocarbon/hydrophobed silica antifoams also show deactivation when dispersions are subject to continuous aeration as do mixtures of PDMS with ethylene distearamide [2⁎⁎]. On the other hand there is that with (organic liquid/organic particle)-based antifoams deactivation can be extremely slow (or even absent) [2⁎⁎]. For example antifoam mixtures of triolein with tristearin particles reveal little or no deactivation after 2h of continuous aeration in solutions of sodium alkyl benzene sulphonate [19]. If then all these various oil/particle antifoams are subject to the same splitting and coalescence processes then we can have radically different outcomes. It seems likely that those differences derive from the ease with which intractable particle-rich entities can be formed. This in turn may concern the viscosity of the oil, the rheology of the oil/particle mixtures as a function of particle volume fraction and concentration, degree of aggregation and even particle geometry. Systematic experimental studies of the behaviour of different antifoam types with respect to rates of deactivation and characterized with respect to this list of variables would seem to have merit especially if combined with attempts to develop a rule-based simulation of the deactivation process. Experimental investigations of the effect of antifoam concentration on the rate of deactivation are extremely rare. One such investigation is reported by Pelton [38] for a PDMS/silica antifoam with foam generated by sparging. A theoretical treatment of deactivation which affords comparison with these experiments has been published recently [2⁎⁎]. Somewhat simplistically we can represent the deactivation process as the sum of a first order splitting phenomenon and a second order coalescing phenomenon so that the rate of loss, dc(t)/dt, of effective antifoam entities is given by −

dcÞt 2 ¼ k1 cðt Þ þ k2 cðt Þ dt

concentration may be used. Thus Garrett [2⁎⁎] has shown that a simple exponential can represent the antifoam concentration dependence of foam behaviour in the case of sparging in the presence of polydimethylsiloxane/silica antifoam at short times where deactivation is negligible and concentration is therefore known. We can therefore write

ð6Þ

where k1 and k2 are rate constants and c(t) is the concentration of active antifoam at time t. Integration of Eq. (6) yields c(t) as a function of time. An empirical relationship between the known foam volume and

t 0

expðm  cðt ÞÞdt

ð7Þ

where m is an empirical constant, VG is the volumetric gas flow rate and VF(t) is the foam volume at time t in the presence of deactivating antifoam of residual antifoam concentration c(t). Integrating Eq. (6) and substituting for c(t) in Eq. (7) produces an integral equation for the time dependence of VF(t) containing three unknown constants — m, k1 and k2. This equation can be fitted to a plot of VF(t) against time for a given initial antifoam concentration, c(t = 0), to obtain values of those constants. The fit is shown in Fig. 4. A test of theory is then the prediction of the time dependence of VF(t) for different initial antifoam concentrations which are also presented in the figure. Semi-quantitative agreement is found — that agreement however represents a significant improvement over earlier attempts [38] to interpret these results. 5. Mechanical defoaming 5.1. Ultrasonics That ultrasound may represent a method of defoaming has been known for more than sixty years. Initial demonstrations of defoaming utilized sirens which proved impractical. Successful application required the development of piezoelectric ultrasonic generators in the late 1970s. A review of the history of defoaming by ultrasound is to be found in [2⁎⁎]. Some context is provided if we consider the general features of the interaction of sound with foam in the usual case where the ultrasonic generator is placed above the foam, separated by an air gap, as shown in Fig. 5. Firstly we note that the velocity of sound in foam is significantly less than that in either air or water [39] where we also remember that the velocity of sound in water is more than three times faster than in air. This means that the wavelength of the sound pressure wave decreases as it passes into foam from the air as shown in the figure. Sound is also attenuated when passing through foam. Thus the rate of

3

Foam volume, V(t)/ cm 1000 900 calculated 800 700 600 500 400 300 200 100 0 0

500

fit

1000 1500 Time/s

2000

2500

Fig. 4. Deactivation of a hydrophobed silica/polydimethylsiloxane antifoam emulsion dispersed in 5 g dm−3 of a dishwashing liquid [38]. Foam generated by passing nitrogen through a porous frit at 7 cm3 s−1. Antifoam concentrations, ■, 0.1 g dm−3; △, 0.3 g dm−3; and ♦, 0.5 g dm−3. Fit to results at 0.5 g dm−3 yielded m = −9.65 g−1 dm3; k1 = 1.3 × 10−6 s−1; k2 = 1.36 × 10−3 g−1 dm3 s−1 in Eqs. (6) and (7). These values were used to calculate the plots shown for 0.1 and 0.3 g dm−3. ([2⁎⁎], © 2013 from The Science of Defoaming: Theory, Experiment and Applications by P.R.Garrett. Reproduced by permission of Taylor and Francis Group, LLC, a division of Informa plc).

P.R. Garrett / Current Opinion in Colloid & Interface Science 20 (2015) 81–91

source

air

foam

liquid

Fig. 5. Schematic showing propagation of sound waves through foam in a bubble column. Continuous lines represent progressive waves and broken lines represent reflected waves. (after Komarov, S.V. et al. [43]. Originally published in ISIJ International).

delivery of acoustic energy per unit area, the acoustic intensity IH, after passing through a foam layer of height H is given by [40] IH ¼ expð−2αH Þ IA

ð8Þ

where IA is the acoustic intensity transmitted through the air–foam interface and α is the attenuation coefficient. The attenuation coefficient increases with increasing frequency [40]. The acoustic intensity is proportional to the square of the amplitude of the pressure fluctuations caused by the sound wave. The amplitude of that wave is therefore shown in Fig. 5 to decline as it passes through the foam. However the acoustic intensity arriving at a given interface is partially transmitted and partially reflected at that interface. In general then [41] IA 4Z 1 Z 2 ¼ Ii ðZ 1 þ Z 2 Þ2

ð9Þ

where Ii is the incident intensity at the interface. Z1 and Z2 are the specific acoustic impendances of mediums 1 and 2 respectively where the specific impedance of a medium is the product of the density and the velocity of sound in that medium. Here we note that if Z1 = Z2 the transmission is total but if there is a marked imbalance between the acoustic impendances so that either Z1 ≫ Z2 or Z1 ≪ Z2 then transmission is minimal. In the case of foam of gas volume fraction 0.999 at the top of the foam column, for example, the acoustic transmission is almost 94% at the air–foam interface [2⁎⁎]. However in the case of a very wet foam, with a gas volume fraction at the Kugelschaum limit of ~ 0.72 at the

89

top of the foam column [2⁎⁎], only about 10% of the acoustic energy is transmitted into the foam. The imbalance is even more marked at the water–foam interface. The high velocity of sound in water means that essentially no acoustic energy will be transmitted across that interface regardless of whether the transmission is from water to foam or foam to water as shown in Fig. 5. Winterburn and Martin [42] have in fact shown that placing the ultrasound generator in the aqueous phase is ineffective in causing defoaming. Although defoaming by ultrasound can be achieved there have apparently been few attempts to understand or even optimize the process. However some early insight is provided by the work of Komarov et al. [43]. This study in part concerned the effect of low frequency sound (using a loud speaker and therefore not in the ultrasound range) on the steady state foam height obtained by sparging a solution of a surfactant in aqueous mixtures of different proportions of glycerin and water. For a given frequency defoaming became apparent only at a threshold acoustic intensity (as measured by the acoustic pressure). Increase in frequency however increased the threshold value of the acoustic intensity. Increasing the viscosity of the solution decreased the threshold slightly. All of this tends to suggest an optimal frequency where the acoustic energy required for a given extent of defoaming is minimal. Rodriguez et al. [44] have more recently made a study of the effect of acoustic intensity on defoaming a fixed volume of foam sparged from an aqueous solution of “soap”. This utilized only one ultrasonic frequency of 25.8 kHz which was generated from a piezoelectric device. As found by Komarov [43] there appears to be a threshold acoustic intensity below which defoaming does not occur. Beyond the threshold defoaming increases with increase in the total amount of acoustic energy — i.e. the higher the acoustic intensity and the longer it is applied above the threshold the greater the volume of foam destroyed. Usefully Rodriquez et al. [44] include gas volume fractions and estimates of bubble sizes in their study. The wavelength of the ultrasound can be calculated from the frequency and knowledge of the velocity of sound as a function of gas volume fraction in a foam (see for example Kann [39]). We find then that in the case of foam with bubble diameters of 0.2–2 mm the acoustic wavelength was ~4 mm and in the case of bubble diameters of 0.2–10 mm the wavelength was ~ 6 mm. These findings hint at the possibility that defoaming by ultrasound concerns a resonance with the mesostructure of the foam. The nature of any quantitative relationship between the bubble diameters and acoustic wavelength would however require detailed knowledge of the bubble size distributions so that at least mean bubble diameters can be calculated. A recent paper by Ben Salem et al. [45⁎⁎] represents a significant development in the understanding of the propagation of ultrasound in aqueous foams. In this work the ultrasound frequency was again maintained constant but at acoustic intensities below the threshold for defoaming. The bubble size of the foam was measured by image analysis and allowed to vary by the process of diffusional disproportionation. The change in amplitude of the ultrasound signal of constant incident acoustic intensity across a sample of foam of known thickness was used as a measure of the attenuation of acoustic intensity. The phase difference yielded the transit time across the foam and therefore the speed of sound in that context. The relative amplitude varied as a function of time as the bubble sizes increased due to diffusional disproportionation. A plot of relative amplitude against time yielded a pronounced minimum at a “transition time”, implying a maximum rate of energy absorption by the foam. A separate plot of the evolution of bubble size against time meant that this transition time could be related to a “critical” bubble size. Ben Salem et al. [45⁎⁎] used these techniques to compare the transition times and critical bubble sizes of two radically different foams — a shaving foam in air and a dilute SDS solution in C6F6. Despite radically different extents of acoustic attenuation and rates of disproportionation the critical bubble sizes of each of these foams were in the same range, i.e. 50 ± 20 μm. Ben Salem et al. [45⁎⁎] then argue that at the critical radius the “bubbles become resonant, hence they pump a maximum of energy from the incident signal”. We should

90

P.R. Garrett / Current Opinion in Colloid & Interface Science 20 (2015) 81–91

however note that the critical bubble radius is about an order of magnitude less than the relevant acoustic wavelengths — this leaves open the question of what energy absorption process could scale in such a manner. Perhaps it concerns the number and size of foam films rather than that of the bubbles [46]. Nevertheless it would appear surprising if the critical bubble radius is not also some function of the acoustic frequency. These findings of Ben Salem et al. [45⁎⁎] may be relevant for defoaming by ultrasound where the latter may also involve resonance but at higher incident acoustic intensities above the threshold leading to foam film destruction of bubbles of the critical size. This could suggest a possible speculative model for defoaming by ultrasound. Thus defoaming becomes maximal if the bubble sizes are at the critical size range for the resonant frequency. Increasing the intensity at that frequency will produce progressively more effective defoaming. However increasing the frequency above the resonant frequency for a foam of a given bubble size will mean less efficient energy transfer into the foam, needing higher threshold acoustic intensities for defoaming as shown by Komarov [43]. If realistic all this could imply that optimal defoaming requires multiple frequency generators tuned to the bubble sizes in the polydisperse foams of practical interest. It does not however address the nature of the actual process of foam collapse during the application of ultrasound. 5.2. Use of rotary devices Various rotary devices have been suggested for the control of foam, especially in sparged unstirred and stirred bubble columns [2⁎⁎]. These devices are exemplified by spinning blades, rotor-stators, turbines, and discs, placed at the top of these bubble columns. There appears to be little consensus about the mode of action of such devices, being variously attributed to either centrifugal or shear forces [2⁎⁎]. However the absence of any significant recent published literature on this matter places the subject outside the scope of this review — an account of earlier work is to be found in reference [2⁎⁎]. 5.3. Orifice defoamer Liu et al. [47] describe a defoaming device consisting of a collection of perforated plates. Defoaming is induced by simply passing foam through the orifices in the plates. However the orifices must have smaller diameters than those of the bubbles in the foam. This arrangement means that large deformations of bubbles occur as they pass through the orifices. Foam films are thereby stretched until they rupture. This in turn causes bubble coalescence and collapse, both of which contribute to defoaming. Observations of defoaming with this device were however confined to aqueous submicellar sodium dodecyl sulphate (SDS) solutions of concentrations ≤~3.5 × 10−3 M. A measure of the resistance to stretching of a foam film is provided by the Gibbs elasticity. Lucassen [48] has presented plots of Gibbs elasticity, εG, for submicellar solutions of SDS which show a maximum at ~ 2.5 × 10− 3 M for film thicknesses in the region of 1–2 μm. At concentrations of SDS lower than that maximum the gradient of Gibbs elasticity with respect to concentration, dεG/dc, is positive. This means that if a foam film is stretched so that both the thickness and the intra-lamellar surfactant concentration are reduced then the Gibbs elasticity also decreases [2**,48]. Lucassen [48] argues that the film will then be dynamically unstable. This conclusion appears to be consistent with the observations of Liu et al. [47] who report decreasing defoaming effects with increasing SDS concentration up to ~3.5 × 10−3 M which is close to the maximum experimental value quoted by Lucassen [48]. Any discrepancy could be due to the role of dynamic factors in the function of the device used by Liu et al. [47]. However if the apparent consistency with the calculations of Lucassen [48] represents the reason for the defoaming effect then increasing the surfactant concentration so that dεG/dc b 0 (under the relevant dynamic conditions) will tend to eliminate that effect.

6. Summarizing remarks We conclude by summarizing the significant recent developments and the remaining challenges in the study of defoaming. Use of surface energy minimization clearly represents an example of such a development, representing a useful technique for predicting the orientations of particles at surfaces. It may also represent a means of simulating other phenomena of importance in the study of defoaming such as the bridging–stretching process and the effect of bridging drops in Plateau borders. Limitations include exclusion of disjoining forces in thin films and, of necessity, viscous dissipation. Foam generation is a dynamic phenomenon where air–water surfaces may depart significantly from equilibrium if the rate of surfactant transport to surfaces is slow relative to the rate of formation of those surfaces. More evidence has been obtained to show that this can lead to transient antifoam effects which only occur during foam generation. In turn this can mean that the effectiveness of antifoams can vary according the method of foam generation. A key and very difficult challenge here is the characterization of the air–water surface during actual foam generation. Such transient effects concerning antifoam action are also apparent under micro-gravity but are then due to low gas volume fractions and the absence of liquid drainage after foam generation has ceased and do not concern transport to air–water surfaces. Precipitation of anionic surfactant from aqueous solution as mesophase particles can be induced by the addition of, for example, excess counterions. Such precipitation is accompanied by declines in foamability due to the slow transport of the surfactant present in the mesophase particles to the air–water surface. Similar behaviour is apparent in the case of surfactants at temperatures below the relevant Krafft temperature where the precipitate is then crystalline. However the hydrophobic nature of such precipitates of surfactant means that antifoam effects are likely to be superimposed upon the transient effects due to slow surfactant transport to air–water surfaces. These generalisations are, however, based upon a relatively small number of examples. Extension of modern insights into the mode of action of antifoams in aqueous systems to the study of non-aqueous systems has been essentially absent until recently. In the case of a hydrocarbon based liquid like crude oil it has recently been shown that antifoam action by liquid alkoxylated polydimethylsiloxane antifoams occurs with positive bridging coefficients and negative spreading coefficients. The main differences with respect to aqueous systems concern (1) the absence of factors which stabilize pseudoemulsion films so that particles are not a necessary components of the antifoams and (2) the relatively high solubility of these PDMS derivatives in the foaming liquid (i.e. crude oil). Mixtures of PDMS and hydrophobed silica are known to be effective antifoams for aqueous systems. However these antifoams deactivate in use. The process involves disproportionation of the dispersed antifoam into inactive oil drops without particles and particle rich drops and agglomerates. A simple theory based upon an assumption that this process involves both first order drop splitting and second order coalescence is in semi-quantitative agreement with experimental observations of deactivation. However there is evidence that deactivation is not a common feature of all types of oil/particle antifoams. There is a need therefore to study this issue in order to improve overall understanding of deactivation so that more effective antifoams may be developed. Recent accounts of defoaming by mechanical means are largely confined to the use of ultrasonics. There are indications that defoaming by application of ultrasound is a resonant phenomenon where an optimum frequency is tuned to a critical foam dimension which unsurprisingly appears to be some function of bubble size. More effective application of ultrasound in this context would appear to require detailed study of this issue.

P.R. Garrett / Current Opinion in Colloid & Interface Science 20 (2015) 81–91

Acknowledgement The author is grateful to Dr. Gareth Morris (Imperial College) for supplying the images shown in Fig. 2. References [1] Marinova K, Tcholakova S, Denkov N. Role of foam dynamics for antifoam activity. EUFOAM; 2014[to be published]. [2⁎⁎] Garrett PR. The science of defoaming, theory, experiment and applications. Surfactant Sci. Series, vol. 155. New York: Taylor and Francis; 2013.This is the only comprehensive monograph dealing with the subject of defoaming including an account of the mode of action of antifoams and featuring some material not published elsewhere. The book also includes a review of mechanical defoaming together with detailed accounts of applications ranging from detergent products to gas–oil separation in crude oil production. [3⁎] Denkov N, Marinova K, Tcholakova S. Mechanistic understanding of the modes of action of foam control agents. Adv Colloid Interface Sci 2014;206:57–67.A good review of the essentials of antifoam action. [4] Karakashev SI, Grozdanova MV. Foams and antifoams. Adv Colloid Interface Sci 2012;176–177:1–17. [5] Owen MJ. Foam control. Advances in Silicones and Silicone-Modified Materials, vol. 1051. ACS Symposium series; 2010. p. 269–86. [6] Miller CA. Antifoaming in aqueous foams. Curr Opin Colloid Interface Sci 2008;13: 177–82. [7⁎⁎] Denkov ND. Mechanisms of foam destruction by oil-based antifoams. Langmuir 2004;20:9463–505.A thorough review of the work of Denkov and co-workers concerning the mode of action of oil-based antifoams in aqueous solutions. [8] Dippenaar A. The destabilization of froth by solids. 1. The mechanism of film rupture. Int J Miner Process 1982;9:1–14. [9⁎] Joshi K, Baumann A, Jeelani SAK, Blickenstorfer C, Naegeli I, Windhab EJ. Mechanism of bubble coalescence induced by surfactant covered antifoam particles. J Colloid Interface Sci 2009;339:446–53.Presents an interesting new approach to study of antifoam phenomena. [10] Brooks JH, Alexander AE. The spreading behaviour and crystalline phases of fatty alcohols. Part IV. Equilibrium spreading pressure. In: La Mer VK, editor. Retardation of Evaporation by Monolayers: Transport Processes. Academic Press; 1962. p. 259–69. [11] Morris GDM, Neethling S, Cilliers JJ. The effects of hydrophobicity and orientation of cubic particles on the stability of thin films. Miner Eng 2010;23:979–84. [12] Morris GDM, Neethling S, Cilliers JJ. A model for investigating the behaviour of non-spherical particles at interfaces. J Colloid Interface Sci 2011;354:380–5. [13] Morris GDM, Cilliers JJ. Behaviour of a galena particle in a thin film, revisiting Dippenaar. Int J Miner Process 2014;131:1–6. [14⁎⁎] Morris GDM, Neethling S, Cilliers JJ. An investigation of the stable orientations of orthorhombic particles in a thin film and their effect on its critical failure pressure. J Colloid Interface Sci 2011;361:370–80.Illustrates the use of energy minimisation simulation to reveal the orientations of particles in surfaces and foam films to discover orientations otherwise unrealized. This illustrates a technique which has some promise in predicting the potential of particles of various geometries as foam and pseudoemulsion film breakers. [15] Brakke K. The surface evolver. Exp Math 1992;1:141–65. [16] Morris GDM, Hadler K, Cilliers JJ. Particles in thin liquid films and at interfaces. Curr Opin Colloid Int Sci 2015;20:98–104. [17] Luangpiron N, Dechabumphen N, Saiwan C, Scamehorn JF. Contact angle of surfactant solutions on precipitated surfaces. J Surfactant Deterg 2001;4(4):367–73. [18] Peck TG. The Mechanisms of Foam Breakdown by Oils and Particles. (PhD Thesis) University of Hull; 1994. [19] Ran L, Jones SA, Embley B, Tong MM, Garrett PR, Cox SJ, et al. Characterisation, modification and mathematical modeling of sudsing. Colloids Surf A Physicochem Eng Asp 2011;382:50–7. [20] Ran L. Foaming of Anionic Surfactant Solutions in the Presence of Calcium Ions and Triglyceride-Based Antifoams. (PhD Thesis) University of Manchester; 2011. [21] Brochard-Wyatt F, di Maeglio J, Quere D, de Gennes P. Spreading on nonvolatile liquids in a continuous picture. Langmuir 1991;7:335–8. [22] Rowlinson JS, Widom B. Molecular Theory of Capillarity. Oxford: Oxford University Press; 1982.

91

[23] Garrett PR. Preliminary considerations concerning the stability of a liquid heterogeneity in a plane-parallel liquid film. J Colloid Interface Sci 1980;76(2):587–90. [24] Arnaudov L, Denkov ND, Surcheva I, Durbut P, Broze G, Mehreteab A. Effect of oily additives on foamability and foam stability. I. Role of interfacial properties. Langmuir 2001;17:6999–7010. [25] Basheva ES, Stoyanov S, Denkov ND, Kasuga K, Satoh N, Tsujii K. Foam boosting by amphiphilic molecules in the presence of silicone oil. Langmuir 2001;17:969–79. [26] Zhang H, Miller CA, Garrett PR, Raney KH. Mechanism of defoaming by oils and calcium soap in aqueous systems. J Colloid Interface Sci 2003;263:633–44. [27] Shu X, Meng Y, Wan L, Li G, Yang M, Jin W. pH-Responsive aqueous foams of oleic acid/oleate solution. J Dispers Sci Technol 2014;35:293–300. [28] Garrett PR, Moore PR. Foam and dynamic surface properties of micellar alkyl benzene sulphonates. J Colloid Interface Sci 1993;159:214–25. [29] Gao F, Yan H, Wang Q, Yuan S. Mechanism of foam destruction by antifoams: a molecular dynamics study. Phys Chem Chem Phys 2014;16:17231–7. [30⁎] Fraga AK, Santos RF, Mansur CRE. Evaluation of the efficiency of silicone polyether additives as antifoams in crude oil. J Appl Polym Sci 2012;124:4149–56.For arguably the first time the modern concept of the bridging coefficient is successfully applied to the interpretation of antifoam effects in a non-aqueous system. [31] Shearer LT, Akers WW. Foaming in lube oils. J Phys Chem 1958;62:1269–70. [32] Binks BP, Davies CA, Fletcher PDI, Sharp EL. Non-aqueous foams in lubricating oil systems. Colloids Surf A Physicochem Eng Asp 2010;360:198–204. [33] Rezende DA, Bittencourt RR, Mansur CRE. Evaluation of the efficiency of polyether-based antifoams for crude oil. J Petrol Sci Eng 2011;76:172–7. [34] Nemeth Z, Racz G, Koczo K. Antifoaming action of polyoxyethylene– polyoxypropylene–poloxyethylene-type triblock copolymers on BSA foams. Colloids Surf A Physicochem Eng Asp 1997;127:151–62. [35] Marinova KG, Dimitrova LM, Marinov RY, Denkov N, Kingma A. Impact of the surfactant structure on the foaming/defoaming performance of nonionic block copolymers in Na caseinate solutions. Bulg J Phys 2012;39:53–64. [36⁎] Caps H, Vandewalle N, Saint-Jalmes A, Saulnier L, Yazhgur P, Rio E, et al. How foams unstable on Earth behave in microgravity? Colloids Surf A Physicochem Eng Asp 2014;457:392–6.Reveals something of the potential for study of antifoam action under conditions where gravity driven drainage is absent permitting variation of gas volume fractions without that complication. [37] Koczo K, Koczone J, Wasan DT. Mechanisms for antifoaming action in aqueous systems by hydrophobic particles and insoluble liquids. J Colloid Interface Sci 1994;166:225–38. [38] Pelton R. A model of foam growth in the presence of antifoam emulsion. Chem Eng Sci 1996;51(19):4437–42. [39] Kann KB. Sound waves in foams. Colloids Surf A Physicochem Eng Asp 2005;263: 315–9. [40] Komarov SV, Kuwabara M, Sano M. Control of foam height using sound waves. ISIJ Int 1999;39(12):1207–16. [41] Rayleigh JWS. The Theory of Sound, vol. II. New York: Dover Publications; 1945. p. 69 [Chp. 13]. [42] Winterburn JB, Martin PJ. Mechanisms of ultrasound foam interactions. Asia Pac J Chem Eng 2009;4:184–90. [43] Komarov SV, Kuabara M, Sano M. Suppression of slag foaming by a sound wave. Ultrason Sonochem 2000;7:193–9. [44] Rodriguez G, Riera E, Gallego-Juarez JA, Acosta VM, Pinto P, Martinez I, et al. Experimental study of defoaming by air-borne power ultrasonic technology. Phys Procedia 2010;3:135–9. [45⁎⁎] Ben Salem I, Guillermic R, Sample C, Leroy V, Saint-Jalmes A, Dollet B. Propagation of ultrasound in aqueous foams: bubble size dependence and resonance effects. Soft Matter 2013;9:1194–202.Clear evidence of resonance between maximum sonic energy absorption at a given frequency and a critical bubble size is presented. This may be applicable to defoaming and point the way to a more effective approach than hitherto. [46] Pierre J, Dollet B, Leroy V. Resonant acoustic propagation and negative density in liquid foams. Phys Rev Lett 2014;112:148307. [47] Liu Y, Wu Z, Zhao B, Li L, Li R. Enhancing defoaming using the foam breaker with perforated plates for promoting the application of foam fractionation. Sep Purif Technol 2013;120:12–9. [48] Lucassen J. Dynamic properties of free liquid films and foams. In: LucassenReynders EH, editor. Anionic Surfactants, Physical Chemistry of Surfactant Action. Surfactant Sci. SeriesNew York: Marcel Dekker; 1981. p. 217 [Chp 6].