Particle-stabilised foams: an interfacial study€¦ · dilational elasticity E of the...

8
Particle-stabilised foams: an interfacial study Antonio Stocco, * a Wiebke Drenckhan, a Emanuelle Rio, a Dominique Langevin a and Bernard P. Binks b Received 20th January 2009, Accepted 19th March 2009 First published as an Advance Article on the web 24th April 2009 DOI: 10.1039/b901180c In an attempt to elucidate the remarkable stability of foams generated from dispersions of partially hydrophobic nanoparticles (fumed silica), we present investigations into the static and dilational properties of the gas–liquid interfaces of such dispersions. By relating the dynamic surface tension g(t) and the dilational elasticity E measured using an oscillating bubble device, we confirm that the Gibbs stability criterion E > g/2 against foam coarsening is fulfilled. We complement these studies using ellipsometry and Brewster angle microscopy, which provide evidence for a pronounced adsorption barrier for the particles and a network-like structure in the interface at sufficiently high concentrations. We observe this structure also in freely suspended films drawn from the same particle dispersions. 1. Introduction It is well known that Pickering emulsions, 1,2 i.e. emulsions stabilized solely by partially hydrophobic particles of nano- or micrometre size, are remarkably stable. More recently, it has been demonstrated also that foams can be stabilized by parti- cles. 3,4 Colloidal particles play, in particular, a key role in the stabilization of metallic foams 4 since the more classical foaming agents, such as surfactants or polymers, degrade at the temper- atures required for the foaming process. Several research groups are now working on aqueous foams stabilized solely by particles 3,5–10 or on individual particle-coated bubbles. 11,12 All the existing studies confirm that the particle layer at the gas–liquid interface forms a ‘‘colloidal armour’’ which inhibits the two main ageing mechanisms of the foams: bubble coalescence (film rupture) and coarsening (exchange of gas between bubbles due to differences in Laplace pressure) which are often completely stopped. This leads to super-stable foams (lifetimes of months!), provided that the foaming medium contains a sufficient amount of particles. 5,8–10 The shape and size of the stabilizing particles can vary significantly, ranging from micrometre sized particles 6 to nano- metric aggregates 10 or even rods. 7 Recent investigations revealed a strong correlation between the hydrophobicity of the particles and foam stability. 8,13 For example, in the case of the silica particles used in this article, a clear maximum in foamability is found at a particle hydrophobicity of 34%, expressed as the percentage of unreacted SiOH groups after silanisation. 8,10 Previous research indicates that a key physical parameter underlying the origin of foam stabilization by particles is the dilational elasticity E of the particle-coated bubble surfaces. 8 A general argument, provided by Cervantes-Martinez et al. 8 and based on an analysis by Gibbs, 14 proceeds as follows: foam coarsening occurs because the derivative of the bubble capillary pressure P with respect to the bubble radius R is negative (dP/dR ¼2g/R 2 < 0). This is generally the case as in most systems the surface tension g is independent of the bubble size. Solid particles at a gas–liquid interface, however, have such high desorption energies 5 that their number can be considered fixed, meaning that their concentration (and therefore the surface tension) varies with the interfacial area A (and hence the bubble size). This provides an interfacial, dilational elasticity E ¼ dg/ dln(A), and the derivative of the bubble capillary pressure can now be written as dP/dR ¼2g/R 2 +4E/R 2 . Hence, a bubble becomes stable against coarsening when E > g/2, which is called the Gibbs stability criterion. 14 Safouane et al. 13 and Zang et al. 15 measured g and E for particle monolayers obtained by spreading particles dispersed in alcohol on water surfaces and the measurements confirm the validity of this criterion. The behaviour of foams stabilized solely by silica particles was investigated based on two different aspects: the effect of the particle hydrophobicity and of the particle concentration on the foam- ability of the aqueous dispersions. 8,10,13 It seems likely that particles stabilize foams due to the surface elasticity E of the particle-coated bubble surfaces. Cervantes-Martinez et al. showed that by increasing the particle concentration from 0.1 to 0.7 wt.%, foam coarsening could be entirely prevented over the duration of the experiment, which was attributed to the effect of the surface elas- ticity. The bubble size of polydisperse foam was monitored by a multiple light scattering technique, and no significant change was detected during 10 h for dispersion concentration $0.7 wt%. Also, photographic and optical images of foam samples left in sealed containers displayed no remarkable variations in volume and macroscopic structure over several months. 8 The properties of adsorbed particle layers have been investi- gated both at liquid–liquid and gas–liquid interfaces (ref. 16,17 and references therein). In addition to steric and electrostatic interactions, colloidal particles at interfaces commonly experi- ence capillary-mediated interactions which are either induced by non-spherical colloid shapes, by chemical heterogeneities at the surface or by ‘‘direct’’ interactions. 18 A model using these capil- lary forces was shown recently to satisfactorily explain the rupture of particle clusters at the air–water interface upon application of shear. 19 To complement those investigations we present here studies of the dynamic surface tension g(t) and of a Laboratoire de Physique des Solides, Universit e Paris-Sud, F-91405 Orsay Cedex, France b Surfactant and Colloid Group, Department of Chemistry, University of Hull, Hull, UK, HU6 7RX This journal is ª The Royal Society of Chemistry 2009 Soft Matter , 2009, 5, 2215–2222 | 2215 PAPER www.rsc.org/softmatter | Soft Matter

Transcript of Particle-stabilised foams: an interfacial study€¦ · dilational elasticity E of the...

  • PAPER www.rsc.org/softmatter | Soft Matter

    Particle-stabilised foams: an interfacial study

    Antonio Stocco,*a Wiebke Drenckhan,a Emanuelle Rio,a Dominique Langevina and Bernard P. Binksb

    Received 20th January 2009, Accepted 19th March 2009

    First published as an Advance Article on the web 24th April 2009

    DOI: 10.1039/b901180c

    In an attempt to elucidate the remarkable stability of foams generated from dispersions of partially

    hydrophobic nanoparticles (fumed silica), we present investigations into the static and dilational

    properties of the gas–liquid interfaces of such dispersions. By relating the dynamic surface tension g(t)

    and the dilational elasticity E measured using an oscillating bubble device, we confirm that the Gibbs

    stability criterion E > g/2 against foam coarsening is fulfilled. We complement these studies using

    ellipsometry and Brewster angle microscopy, which provide evidence for a pronounced adsorption

    barrier for the particles and a network-like structure in the interface at sufficiently high concentrations.

    We observe this structure also in freely suspended films drawn from the same particle dispersions.

    1. Introduction

    It is well known that Pickering emulsions,1,2 i.e. emulsions

    stabilized solely by partially hydrophobic particles of nano- or

    micrometre size, are remarkably stable. More recently, it has

    been demonstrated also that foams can be stabilized by parti-

    cles.3,4 Colloidal particles play, in particular, a key role in the

    stabilization of metallic foams4 since the more classical foaming

    agents, such as surfactants or polymers, degrade at the temper-

    atures required for the foaming process.

    Several research groups are now working on aqueous foams

    stabilized solely by particles3,5–10 or on individual particle-coated

    bubbles.11,12 All the existing studies confirm that the particle layer

    at the gas–liquid interface forms a ‘‘colloidal armour’’ which

    inhibits the two main ageing mechanisms of the foams: bubble

    coalescence (film rupture) and coarsening (exchange of gas

    between bubbles due to differences in Laplace pressure) which

    are often completely stopped. This leads to super-stable foams

    (lifetimes of months!), provided that the foaming medium

    contains a sufficient amount of particles.5,8–10

    The shape and size of the stabilizing particles can vary

    significantly, ranging from micrometre sized particles6 to nano-

    metric aggregates10 or even rods.7 Recent investigations revealed

    a strong correlation between the hydrophobicity of the particles

    and foam stability.8,13 For example, in the case of the silica

    particles used in this article, a clear maximum in foamability is

    found at a particle hydrophobicity of 34%, expressed as the

    percentage of unreacted SiOH groups after silanisation.8,10

    Previous research indicates that a key physical parameter

    underlying the origin of foam stabilization by particles is the

    dilational elasticity E of the particle-coated bubble surfaces.8 A

    general argument, provided by Cervantes-Martinez et al.8 and

    based on an analysis by Gibbs,14 proceeds as follows: foam

    coarsening occurs because the derivative of the bubble capillary

    pressure P with respect to the bubble radius R is negative

    aLaboratoire de Physique des Solides, Universit�e Paris-Sud, F-91405 OrsayCedex, FrancebSurfactant and Colloid Group, Department of Chemistry, University ofHull, Hull, UK, HU6 7RX

    This journal is ª The Royal Society of Chemistry 2009

    (dP/dR ¼ �2g/R2 < 0). This is generally the case as in mostsystems the surface tension g is independent of the bubble size.

    Solid particles at a gas–liquid interface, however, have such high

    desorption energies5 that their number can be considered fixed,

    meaning that their concentration (and therefore the surface

    tension) varies with the interfacial area A (and hence the bubble

    size). This provides an interfacial, dilational elasticity E ¼ dg/dln(A), and the derivative of the bubble capillary pressure can

    now be written as dP/dR ¼ �2g/R2 +4E/R2. Hence, a bubblebecomes stable against coarsening when E > g/2, which is called

    the Gibbs stability criterion.14 Safouane et al.13 and Zang et al.15

    measured g and E for particle monolayers obtained by spreading

    particles dispersed in alcohol on water surfaces and the

    measurements confirm the validity of this criterion.

    The behaviour of foams stabilized solely by silica particles was

    investigated based on two different aspects: the effect of the particle

    hydrophobicity and of the particle concentration on the foam-

    ability of the aqueous dispersions.8,10,13 It seems likely that particles

    stabilize foams due to the surface elasticity E of the particle-coated

    bubble surfaces. Cervantes-Martinez et al. showed that by

    increasing the particle concentration from 0.1 to 0.7 wt.%, foam

    coarsening could be entirely prevented over the duration of the

    experiment, which was attributed to the effect of the surface elas-

    ticity. The bubble size of polydisperse foam was monitored by

    a multiple light scattering technique, and no significant change was

    detected during 10 h for dispersion concentration $0.7 wt%. Also,

    photographic and optical images of foam samples left in sealed

    containers displayed no remarkable variations in volume and

    macroscopic structure over several months.8

    The properties of adsorbed particle layers have been investi-

    gated both at liquid–liquid and gas–liquid interfaces (ref. 16,17

    and references therein). In addition to steric and electrostatic

    interactions, colloidal particles at interfaces commonly experi-

    ence capillary-mediated interactions which are either induced by

    non-spherical colloid shapes, by chemical heterogeneities at the

    surface or by ‘‘direct’’ interactions.18 A model using these capil-

    lary forces was shown recently to satisfactorily explain the

    rupture of particle clusters at the air–water interface upon

    application of shear.19 To complement those investigations we

    present here studies of the dynamic surface tension g(t) and of

    Soft Matter, 2009, 5, 2215–2222 | 2215

  • the dilational elasticity E, both of which are measured directly on

    the gas–liquid interfaces of aqueous particle dispersions. Thus,

    we emulate more realistic conditions occurring during foam

    generation.

    We used dispersions of fumed silica nanoparticles with inter-

    mediate hydrophobicity (34% SiOH). Among different % SiOH

    contents, those dispersions provide the maximum volume of foam

    after production, i.e. maximum foamability.8,10 We present

    measurements of the dynamic surface tension g(t) (change of

    surface tension with time) and of the dilational elastic modulus E

    conducted using an oscillating bubble device, for a range of

    particle concentrations previously investigated in a foam stability

    study.8,10 We complement these studies by investigations of the

    nature of the interfacial particle layer using ellipsometry, Brewster

    angle microscopy and observations of free-standing liquid films.

    Fig. 1 (a) Zimm plot obtained using static light scattering measurements

    for aqueous dispersions of silica nanoparticles at different concentrations

    (given). K is an optical constant, Rq is the Rayleigh ratio and a is an

    arbitrary constant. (b) Field autocorrelation function fm(q,s) measured atthe scattering angle q ¼ 90� for systems in (a): 0.1 (,), 0.3 (B), 0.5 (O),0.7 (P) wt.%. Solid lines are cumulant fits.

    2. Materials and methods

    2.1 Materials

    Particles. Fumed silica nanoparticles were kindly provided by

    Wacker-Chemie (Germany). The particles investigated in this

    work were chemically coated with a short-chain silane reagent

    (dichlorodimethylsilane) by the manufacturer. The hydrophobic

    character of the particles is expressed by the percentage of

    surface silanol groups SiOH. We used a 34% grade throughout

    our experiments which corresponds to the particle hydropho-

    bicity which provides maximum foamability.10 The primary

    particles are quasi-spherical of approximately 20 nm diameter,

    but aggregate into clusters over 200 nm in size.20

    Particle dispersions. We prepared aqueous dispersions of

    particles at 1 wt.% concentration by a stepwise procedure using

    doubly distilled and deionized Milli-Q water (pH z 5.8), silicaparticles and a small amount of ethanol (

  • the dispersion was negligible. Using the same apparatus, we also

    performed oscillating bubble experiments in order to obtain the

    dilational elastic modulus E ¼ dg/dlnA. We imposed sinusoidalvariations of the bubble volume DV¼ 0.5–1 mL and we measuredthe corresponding changes of surface tension by image analysis.

    The frequency of the oscillation was swept from 0.1 to 1 Hz.

    Ellipsometry and Brewster angle microscopy (BAM). We used

    an imaging ellipsometer (Nanofilm, Germany) working with

    green laser light (l ¼ 532 nm) in order to gain information onparticle organization in the interfacial region. The aqueous

    dispersions were placed in a large dish (diameter¼ 8 cm, depth¼3 cm) and the laser beam was directed at the surface in the center

    of the cell where the meniscus effect is negligible. A multiple

    angle of incidence (MAI), fixed compensator (¼ �45�) and 4-zone averaging nulling scheme was adopted.24 The ellipsometric

    parameters (J,D), which are related to the ratio r of the complex

    reflection coefficients by r ¼ rp/rs ¼ tanJ exp(iD), deviatesignificantly from the values of the bare water surface since the

    larger refractive index of silica particles (nSi ¼ 1.475) providesa good optical contrast.25 The ellipsometric parameters J and D

    were measured around the Brewster angle, scanning the incident

    angle by steps of 0.1�. From the fits of the latter quantities, the

    refractive index profile along the interfacial region could be

    resolved and information on layer thickness and particle

    concentration at the water surface could be extracted.

    Brewster angle microscopy was also implemented in the same

    setup. By this method, we could gain information on the texture

    and organization of particles in the interfacial plane, which

    served to complement our ellipsometric data with visual infor-

    mation. To check the accuracy of our measurement protocols, we

    always compared our data with the bare air–water interface.

    Individual films. We observed qualitatively the behaviour of

    freely suspended horizontal films formed from the same disper-

    sions using a home-built apparatus in which we placed 5 mL

    drops between two metal barriers of 2 cm width. One barrier was

    moved horizontally by a micrometric screw; hence a film could be

    formed between the barriers and videos of light reflected from the

    film were recorded by means of a camera placed below the film.

    The system was closed to the atmosphere and saturated in

    vapour, to minimize evaporation effects.

    All measurements in this article were conducted at room

    temperature, being 20 � 2 �C.

    3. Results and discussion

    We performed a range of interfacial studies at the air–water

    interface of particle dispersions of different concentrations. In the

    first part, we combine measurements of the dilational elasticity E

    and of the dynamic surface tension g(t) and correlate these results

    to foam stability. In the second part, we describe the adsorption

    and organization mechanisms of the particles at the water surface.

    Fig. 2 Dynamic air–water surface tension measured in the rising bubble

    experiments for different silica nanoparticle concentrations: 0.1 (,), 0.3

    (B), 0.5 (O), 1 (P) wt.%.

    3.1 Purely elastic stabilization against foam coarsening

    Non-chemically modified silica particles have commonly

    a hydrophilic character; hence the surface tension of their

    dispersions is generally independent of the particle concentration

    This journal is ª The Royal Society of Chemistry 2009

    as no significant particle adsorption to the interface takes place.26

    Even if adsorption of particles takes place, no significant change

    of g is reported when particle interactions are of purely steric

    nature.27 This is different to the kind of particles investigated

    here, which adsorb to the interface due to their partial hydro-

    phobicity and whose interactions are governed by repulsive

    electrostatic forces (see also light scattering results, Section 2.1)

    and additionally by attractive capillary forces. We expect the

    latter to be generated by surface chemical inhomogeneities or by

    the irregular shape of the particle aggregates on whose surfaces

    the wetting angle condition must be fulfilled locally, leading to

    a curvature of the liquid surface around the particle and hence to

    particle interactions.18

    Here we have measured changes of the air–water surface

    tension g with time t (‘‘dynamic surface tension’’) for a range

    of particle concentrations. Some representative examples are

    displayed in Fig. 2. We note that the measurement is started

    (t ¼ t0 ¼ 0) as soon as the droplet/bubble is formed in themeasurement device (refer to Section 2.2), which takes about 1 s.

    It is evident from the variation of the observed surface tension

    values at t0 that during this period a significant, concentration-

    dependent amount of particles is already adsorbed at the gas–

    liquid interface. This is probably due to the additional energy

    input resulting from the droplet/bubble generation.

    At concentrations lower than 0.1 wt.%, no appreciable

    changes of g were observed during a time of up to 104 s. Between

    0.1 and 1 wt.%, however, we measured a slow relaxation of the

    surface tension with an overall decrease of approximately

    5 mN m�1. At the highest concentration studied (1 wt.%), the

    surface tension value decreases to around 50 mN m�1, which is

    close to the predicted value given by Binks and Clint.28 Thus, the

    adsorption of particles can be detected by the change of g with

    time, where slow kinetics is usually observed. In fact, as we shall

    illustrate in Section 3.2, nanoparticle adsorption is not only

    diffusion-controlled but also displays kinetics with an energy

    barrier arising from steric and/or electrostatic interactions, as

    was demonstrated by Kutuzov et al. for nanoparticles with a size

  • Fig. 4 (a) Dilational elastic modulus E as a function of particle

    concentration c. (b) Variation of E with the surface tension g measured at

    t ¼ 104 s (reported in Fig. 2) for four different particle concentrations(given). The solid line shows E ¼ g/2.

    for infinite waiting times. The systematic deviation between

    surface tensions measured with pendant drops and rising bubbles

    (Fig. 3) may be due to two effects. Firstly, particle depletion may

    come into play due to the finite amount of particles available in

    the small pendant drop (between 5 and 10 mL). Secondly, grav-

    itational effects may lead to a more efficient transport of particles

    to the interface in the case of the rising bubble. Nonetheless,

    measurements obtained by both methods confirm a significant

    decrease of g between 0.1 and 1 wt.%.

    In order to evaluate the surface elastic properties of the particle-

    laden interfaces, we carried out oscillating droplet and bubble

    experiments. To obtain quantitative information we preferred the

    oscillating bubble over the pendant drop configuration, because

    in the latter, after adsorption times of 104 s, the shape of the drop

    tended to become non-Laplacian with a corrugated surface. This

    phenomenon was seldom observed in the rising bubble configu-

    ration. A similar phenomenon was reported for irreversibly

    adsorbed mixed polyelectrolyte-surfactant layers, where the

    surface can be deformed by thickness gradients that can relax

    towards uniformity, at extremely long times.30

    The dilational elastic modulus E¼ dg/dlnA was measured (aftert z 104 s) with oscillating bubble experiments performed at thefrequency n ¼ 1 Hz, as a function of the particle concentration c.The results are presented in Fig. 4(a). The absolute value of E

    increases markedly with increasing particle concentration c,

    reaching a value of approximately 40 mN m�1 at c ¼ 0.7 wt.%. Atlower frequencies, we observed non-elastic phenomena in the form

    of non-smooth profiles of surface tension and surface area with

    time. It seemed that at low frequencies particles had enough time to

    re-arrange during the slow sinusoidal change of the bubble area.30

    Combining the results from Fig. 3 and Fig. 4(a), we show in

    Fig. 4(b) the dilational elastic modulus E as a function of the

    surface tension g in the range of particle concentrations c of

    interest. For comparison, Fig. 4(b) also displays the line E ¼ g/2corresponding to the Gibbs stability criterion. From this, it

    follows that only dispersions with particle concentrations above

    0.5 wt.% can generate foams with significantly reduced coars-

    ening. This result is in perfect agreement with the foam stability

    investigations reported by Cervantes-Martinez et al.8 for the

    same particle dispersions and now allows us to predict when

    foams will be stable or not against coarsening. Indeed, the

    original work by Cervantes-Martinez et al.8 was based on esti-

    mations of the particle concentration at bubble surfaces and

    Fig. 3 Surface tension values g obtained at t z 104 s as a function ofsilica nanoparticle concentration c measured using the pendant drop (-)

    and the rising bubble (B) technique.

    2218 | Soft Matter, 2009, 5, 2215–2222

    comparison with independent measurements made with spread

    layers.

    Moreover, the oscillating bubble experiments indicate the exis-

    tence of an adsorption barrier: the surface tension values obtained

    for a static bubble after long waiting times can be reached much

    more rapidly by oscillating the bubble and hence by providing an

    additional energy input facilitating particle adsorption (see Fig. 5).

    Such effects are known from similar systems. For example, anionic

    latex particles do not adsorb spontaneously at the air–water inter-

    face because the interface itself appears anionic and hence repul-

    sive.31 Similarly, the silica particles under investigation, being

    decorated on the surface by SiO�groups, are also anionic (see also

    Section 2.1). Thus, an energy barrier resulting from electrostatic

    interactions between the particles and the air–water surface, and

    between the particles at the surface and in bulk, may hinder

    adsorption to the interface. Additionally, steric interaction between

    Fig. 5 In dynamic surface tension measurements in the rising bubble

    configuration, the surface tension g decreases more rapidly when the

    bubble is oscillated (here n ¼ 0.1 Hz and c ¼ 0.1 wt.%).

    This journal is ª The Royal Society of Chemistry 2009

  • particles at the surface and in bulk tend to slow the adsorption at

    high surface concentrations.29

    3.2 Organization of nanoparticles at the air–water interface

    In this part we discuss and provide experimental evidence of the

    adsorption mechanism and of the arrangement of partially

    hydrophobic fumed silica particles at the air–water interface of

    particle dispersions. We performed ellipsometric measurements

    in order to gain information on the kinetics of the adsorption, the

    layer thickness and the surface concentration G of particles.

    We investigate the influence of the method of preparation of the

    dispersions on the organization of particles at the interface. At

    the outset, for comparison with previous findings,13,15 we compare

    the case of adsorbed particle layers with those of spread layers.

    3.2.1 Spread layers. We performed ellipsometric scans of high

    accuracy around the Brewster angle (¼ arctan(nwater/nair)¼ 53.12�)on a spread layer of particles. We obtained such a monolayer by

    spreading a known amount of particles dispersed in isopropyl

    alcohol of c¼ 0.1 wt.% onto a pure water surface of fixed area. Westudied two spread layers with a surface concentration of G¼ 20 mgm�2 and G¼ 40 mg m�2 which served as reference concentrations tobe compared with the adsorbed particle layers.

    From the angular dependence of D and J,24 as displayed in

    Fig. 6, we can simultaneously extract information on the thick-

    ness d and on the refractive index nl of the particle layer (see inset

    in Fig. 6).32 From data fitting we obtain d ¼ 162 � 5 nm and nl ¼1.3403 � 0.0005 for G ¼ 20 mg m�2. For twice this surfaceconcentration (G ¼ 40 mg m�2), we find a similar layer thickness

    Fig. 6 Results of ellipsometric scans performed around the Brewster

    angle on a spread layer of silica nanoparticles at the air–water surface

    with surface concentrations G ¼ 20 mg m�2 (,) and G ¼ 40 mg m�2 (B).The solid lines represent the fits. Inset: sketch of the stratified layer model.

    This journal is ª The Royal Society of Chemistry 2009

    (d ¼ 175 � 1 nm), whilst the refractive index increases tonl ¼ 1.354 � 0.0002. From the latter data we could evaluate thesurface concentration ‘seen’ by ellipsometry using the relation

    Gelli ¼ (nl � nwater)/(vn/vc)$d, where vn/vc is the differentialrefractive index increment. Calculating for the two reference

    surface concentrations, Gelli changed from 17.6 mg m�2 to 56.9

    mg m�2. Those values are in a reasonable agreement with the

    actual surface concentrations of 20 and 40 mg m�2. Thus, for

    spread layers, in the range of concentration studied, it seems that

    the particle aggregates remain trapped at the water surface. It is

    worth noting that for nwater ¼ 1.333 < nl < nSi ¼ 1.475, nosatisfying fit can be obtained assuming the presence of non-

    aggregated particles of 20 nm diameter. In fact, if d ¼ 20 nm,around the Brewster angle, D would change from 180� to 360�

    instead of from 180� to 0�.

    3.2.2 Adsorbed layers. We performed the same ellipsometric

    scans on the surface of aqueous particle dispersions of concen-

    trations between 0.1 and 0.7 wt.%. The dispersions were prepared

    and sonicated one day before use. We cleaned the surface of the

    dispersions (and of water) several times by means of a pump

    before performing the first ellipsometric scan. Then we waited at

    least 1 h to allow the interface to equilibrate before performing

    further measurements.

    Fig. 7 shows J and D as a function of the incident angle 4

    for the bare air–water interface and for a particle dispersion at

    Fig. 7 Ellipsometric scans performed around the Brewster angle at the

    bare air–water interface (x) compared to those conducted at the surface of

    an aqueous silica dispersion with a particle concentration of c ¼ 0.1 wt.%after adsorption for one day (,). > corresponds to measurements

    obtained with the same dispersion after 10 min of strong sonication. The

    dotted line is the simulated step-like profile of the air–water surface and

    the solid line is a film model fit giving a layer thickness of d¼ 149 nm witha refractive index nl ¼ 1.348.

    Soft Matter, 2009, 5, 2215–2222 | 2219

  • Fig. 8 Brewster angle microscopy images of the air–water surface for

    silica nanoparticle concentrations c of (a) 0.1 wt.% and (b) 0.7 wt.% taken

    after one day of aging and 10 min of sonication. The bright areas contain

    significantly more particles. At low concentrations (a), one observes

    isolated structures which slowly and freely move within a homogeneous

    layer of particles. At sufficiently high particle concentrations (b),

    a network-like structure of particles appears. The scale bar is 20 mm.

    c ¼ 0.1 wt.% after 1 day. Both surfaces give nearly identicalsignals. The same holds for measurements in the whole concen-

    tration range between 0.1 and 0.7 wt.%. Even after 1 day, ellip-

    sometric scans did not show any change of the interfacial profile,

    thus implying that no significant amount of particles had been

    adsorbed at the interface. Using Brewster angle microscopy, we

    observed islands of particles floating at the water surface, but

    their area fraction was not high enough to contribute to the

    ellipsometric signal.

    These results are in apparent contradiction with those

    obtained using spread particle monolayers. The main difference

    between the two cases is that in the latter all particles are placed

    directly at the surface where they remain due to large desorption

    energies. In the case of aqueous dispersions, however, the

    particles need to overcome two obstacles: firstly, they need to

    overcome energy barriers present at the interface, as discussed at

    the end of Section 3.1. Secondly, they need to diffuse against

    gravity from the bulk to the surface (silica particles are signifi-

    cantly heavier than water with a density of r ¼ 2.2 g cm�3). Thelatter effect may also explain why we see spontaneous particle

    adsorption in the case of bubbles or drops (Section 3.1), where

    particle sedimentation helps adsorption due to the different

    geometry of the air–water interface. To see this, let us evaluate

    the relationship between particle diffusion and sedimentation for

    our system. The mean square displacement for a diffusing

    particle is given by ¼ 6Dt, with ¼ 2.5 � 1012 m2s�1, as evaluated by dynamic light scattering (Section 2.1. and ref.

    22). The displacement of a particle due to gravity can be written

    as L¼ vt, where the Stokes velocity is given by v¼mg/(6phRh)¼3.4 � 10�8 m s�1 (mg is the gravitational force and Rh ¼ 85 � 10nm, the hydrodynamic radius (Section 2.1)). The two different

    lengths are equal ( ¼ L2) for t ¼ 1.2 � 104 s, meaningthat gravitational effects become important at these time scales,

    which corresponds well to what we see in Fig. 3.

    To test this hypothesis, we strongly sonicated (10 min, Section

    3.2.3) the dispersions after one day (at which no significant

    particle adsorption had taken place). Performing ellipsometric

    scans immediately after sonication reveals significant changes of

    the interfacial profile, i.e. significant particle adsorption, which

    can be seen thanks to the changes of J and D displayed in Fig. 7

    (>) for the case of a dispersion concentration c ¼ 0.1 wt.%.From fitting this profile we obtain a layer thickness of d ¼ 149 �2 nm with a refractive index of nl ¼ 1.3477 � 0.0007.

    3.2.3 Particle networks at air–water surfaces. Comparing

    Fig. 6 and 7, we noticed that the profile of the interface generated

    by sonication-induced adsorption for c ¼ 0.1 wt.% (d ¼ 149 � 2nm, nl¼ 1.3477� 0.0007) differs significantly from the one of thespread layer for G ¼ 40 mg m�2 (d ¼ 175 � 1 nm, nl ¼ 1.3539 �0.0002). This difference is even more evident from Brewster angle

    microscopy images. Whilst spread monolayers tend to be very

    homogeneous and show compact packing at a surface concen-

    tration of G ¼ 60 mg m�2,13,15 we find very inhomogeneousparticle distributions at the interface for adsorbed layers.

    Examples of typical Brewster angle microscopy images for

    dispersion concentrations of c ¼ 0.1 and 0.7 wt.% are displayedin Fig. 8, clearly indicating pronounced interfacial textures after

    sonication for adsorbed monolayers. At low concentrations

    (Fig. 8(a)), sonication leads to the formation of some solid

    2220 | Soft Matter, 2009, 5, 2215–2222

    structures which slowly and freely move within a homogeneous

    layer of particles. This observation may also explain the scattered

    ellipsometric data obtained after sonication (Fig. 7). At suffi-

    ciently high particle concentrations (Fig. 8(b)), a completely

    different scenario is observed: a network-like structure of parti-

    cles appears. We did not observe changes of these textures within

    24 h by Brewster angle microscopy.

    Between the two concentrations depicted in Fig. 8, i.e. 0.1

    and 0.7 wt.%, foams dramatically change their stability (Section

    3.1 and Fig. 4(b)).8,10 Thus, the observation of the different

    kinds of structures shown in Fig. 8 could confirm the suggestion

    made by Kostakis et al.,33 who related foam stabilization to the

    formation of weak gel networks of particles between gas–liquid

    interfaces separating the bubbles. At this stage, however, it is

    not clear to us what is the origin of the attractive forces between

    particles enabling the formation of the observed networks. They

    could be capillary-mediated interactions induced by the

    complex aggregate geometry.18 Additionally, SiOH groups on

    adjacent particles can interact to form siloxane bonds (Si–O–Si)

    which allow particles to aggregate. Being partially hydrophobic,

    This journal is ª The Royal Society of Chemistry 2009

  • Fig. 9 Images of a freely suspended, horizontal film drawn from silica nanoparticle dispersions with (a) low c ¼ 0.05 wt.% and (b) high c ¼ 0.5 wt.%particle concentration. Time between images is 30 s. The scale bar is 1 mm.

    their charge is reduced compared with the equivalent fully

    hydrophilic silica.

    3.2.4 Particle networks in thin films. The formation of

    network-like particle structures, as described in the previous

    section, can also be observed in freely suspended individual thin

    films (Section 2.2). Fig. 9 shows the development of two exam-

    ples of films drawn from dispersions of two different concen-

    trations (Fig. 9(a): c ¼ 0.05 wt.% and Fig. 9(b): c ¼ 0.5 wt.%). Ineach picture, the grey top and bottom parts correspond to the

    metal barriers. In the central part, a meniscus (black coloured)

    surrounds the thin film region (see sketch in Fig. 9(a)). Colours

    and light intensity result from light interference at the two gas–

    liquid interfaces and therefore provide information on film

    thickness and thickness changes. At low concentrations

    (Fig. 9(a)), fairly homogenous thin films are formed whose

    thickness fluctuates in time accompanied by rapid colour

    changes, and decreases overall with time to cause film rupture

    after a few minutes.

    At sufficiently high particle concentrations (Fig. 9(b)), the

    scenario is quite different: as observed in the previous Section

    (Fig. 8(b)), the particles form network-like structures and lead to

    the formation of very thick films which are significantly more

    stable with lifetimes of up to 1 h.

    4. Conclusions

    We have presented investigations into the static and dynamic

    properties of air–water interfaces of aqueous dispersions of

    partially hydrophobic silica nanoparticles (34% SiOH coverage).

    Our measurements strongly support the argument that the

    stability of particle-stabilised, aqueous foams can be predicted by

    the Gibbs elasticity criterion E > g/2, which relates the surface

    tension g and the dilational elastic modulus E of the particle

    covered interfaces. Both g and E depend on the bulk particle

    concentration of the dispersion, making this therefore a key

    parameter for the control of foam stability. Our predictions are

    in perfect agreement with the results of Cervantes-Martinez

    et al., which were based on estimations of the particle concen-

    tration at the surface of bubbles.

    This journal is ª The Royal Society of Chemistry 2009

    We find that particle adsorption is inhibited by a pronounced

    energy barrier29 which can be overcome using strong sonication

    of the particle dispersions. The precise nature of this adsorption

    barrier is not yet clear to us, but it might explain why the

    generation of stable particle foams requires high energy tech-

    niques, such as turbulent mixing.

    Finally, we show that at sufficiently high particle concentra-

    tions and after sufficient energy input, particles form network-

    like structures at the surface of aqueous dispersions and in thick

    horizontal free-standing films. Similar structures were seen

    recently in electron microscopy experiments.34 The formation of

    such structures requires the presence of attractive particle inter-

    actions in the interface, whereas we find repulsive particle

    interactions in the bulk from our light scattering data (B2 > 0).

    The nature of these interfacial attractive interactions is still

    unclear.18

    Acknowledgements

    The authors would like to thank Pawel Pieranski for discussions

    and for kindly sharing his freely suspended film equipment. We

    thank Liliane L�eger for lending us her ellipsometer, Eric Ras-

    paud for providing the light scattering equipment and Wacker-

    Chemie (Burghausen) for providing the silica particles. A. Stocco

    was financed by SOCON (European contract RTN 2004-

    512331).

    References

    1 S. U. Pickering, J. Chem. Soc, 1907, 91, 2001.2 C. W. Ramsden, Proc. Roy. Soc. A, 1903, 72, 156.3 B. P. Binks and R. Murakami, Nat. Mater., 2006, 5, 865.4 J. Banhart, Adv. Eng. Mater., 2006, 8, 781.5 Z. Du, M. P. Bilbao-Montoya, B. P. Binks, E. Dickinson, R. Ettelaie

    and B. S. Murray, Langmuir, 2003, 19, 3106.6 S. Fujii, A. J. Ryan and S. P. Armes, J. Am. Chem. Soc., 2006, 128,

    7882.7 R. G. Alargova, D. S. Warhadpande, V. N. Paunov and O. D. Velev,

    Langmuir, 2004, 20, 10371.8 A. Cervantes-Martinez, E. Rio, G. Delon, A. Saint-Jalmes,

    D. Langevin and B. P. Binks, Soft Matter, 2008, 4, 1531.

    Soft Matter, 2009, 5, 2215–2222 | 2221

  • 9 U. T. Gonzenbach, A. R. Studart, E. Tervoort and L. J. Gauckler,Angew. Chem., Int. Ed., 2006, 45, 3526.

    10 B. P. Binks and T. S. Horozov, Angew. Chem., Int. Ed., 2005, 117, 3788.11 A. B. Subramaniam, M. Abkarian, L. Mahadevan and H. A. Stone,

    Langmuir, 2006, 22, 10204.12 M. Abkarian, A. B. Subramaniam, S.-H. Kim, R. J. Larsen,

    S.-M. Yang and H. A. Stone, Phys. Rev. Lett., 2007, 99, 188301.13 M. Safouane, D. Langevin and B. P. Binks, Langmuir, 2007, 23,

    11546.14 J. W. Gibbs, The Scientific Papers of J. Willard Gibbs, Ox Bow Press,

    Woodbridge, 1993.15 D. Y. Zang, E. Rio, D. Langevin, B. Wei and B. P. Binks, in

    preparation.16 B. P. Binks, Phys. Chem. Chem. Phys., 2007, 9, 6298.17 B. P. Binks, Curr. Opin. Colloid Interface Sci., 2002, 7, 21.18 D. Frydel, S. Dietrich and M. Oettel, Phys. Rev. Lett., 2007, 99,

    118302.19 N. D. Vassileva, D. van den Ende, F. Mugele and J. Mellema,

    Langmuir, 2007, 23, 2352.20 T. S. Horozov, B. P. Binks, R. Aveyard and J. H. Clint, Colloids Surf.,

    A, 2006, 282–283, 377.21 P. N. Pusey, Neutrons, X-Rays and Light: Scattering Methods Applied

    to Soft Condensed Matter, ed. P. Linder and Th. Zemb, Elsevier,Amsterdam, 2002.

    2222 | Soft Matter, 2009, 5, 2215–2222

    22 B. J. Berne and R. Pecora, Dynamic light scattering, DoverPublications, Mineola, 2000.

    23 J. Benjamins, A. Cagna and E. H. Lucassen-Reynders, Colloids Surf.,A, 1996, 114, 245.

    24 R. M. A. Azzam and N. M. Bazhara, Ellipsometry and polarized light,Elsevier, Amsterdam, 1977.

    25 B. N. Khlebtsov, V. A. Khanadeev and N. G. Khlebtsov, Langmuir,2008, 24, 8964.

    26 F. Ravera, E. Santini, G. Loglio, M. Ferrari and L. Liggieri, J. Phys.Chem. B, 2006, 110, 19543.

    27 E. Vignati, R. Piazza and T. Lockhart, Langmuir, 2003, 19, 6650.28 B. P. Binks and J. H. Clint, Langmuir, 2002, 18, 1270.29 S. Kutuzov, J. He, R. Tangirala, T. Emrick, T. P. Russell and

    A. Boker, Phys. Chem. Chem. Phys., 2007, 9, 6351.30 H. Ritacco, A. Cagna and D. Langevin, Colloids Surf., A, 2006, 282,

    203.31 S. L. Kettlewell, A. Schmid, S. Fujii, D. Dupin and S. P. Armes,

    Langmuir, 2007, 23, 11381.32 J. L. Keddie, Curr. Opin. Colloid Interface Sci., 2001, 6, 102.33 T. Kostakis, R. Ettelaie and B. S. Murray, Langmuir, 2006, 22,

    1273.34 M. Antoni, Formulation and Characterization of Pickering Emulsions,

    Eufoam 2008, conference proceeding, Noordwijk, The Netherlands,2008.

    This journal is ª The Royal Society of Chemistry 2009

    Particle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial study

    Particle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial study

    Particle-stabilised foams: an interfacial studyParticle-stabilised foams: an interfacial study