You are on page 1of 86

Advances in Colloid and Interface Science 103 (2003) 219304

Mechanism of cationic surfactant adsorption at the solidaqueous interface


R. Atkina, V.S.J. Craigb, E.J. Wanlessc,*, S. Biggsd
a School of Chemistry, University of Bristol, Cantocks Close, Bristol BS8 1TS, UK Department of Applied Mathematics, Research School of Physical Sciences and Engineering, Australian National University, Canberra, ACT 0200, Australia c Discipline of Chemistry, School of Environmental and Life Sciences, The University of Newcastle, Callaghan, NSW 2308, Australia d School of Process, Environmental and Materials Engineering, University of Leeds, Leeds LS2 9JT, UK b

Abstract Until recently, the rapid time scales associated with the formation of an adsorbed surfactant layer at the solidaqueous interface has prevented accurate investigation of adsorption kinetics. This has led to the mechanism of surfactant adsorption being inferred from thermodynamic data. These explanations have been further hampered by a poor knowledge of the equilibrium adsorbed surfactant morphology, with the structure often misinterpreted as simple monolayers or bilayers, rather than the discrete surface aggregates that are present in many surfactantsubstrate systems. This review aims to link accepted equilibrium data with more recent kinetic and structural information in order to describe the adsorption process for ionic surfactants. Traditional equilibrium data, such as adsorption isotherms obtained from depletion approaches, and the most popular methods by which these data are interpreted are examined. This is followed by a description of the evidence for discrete aggregation on the substrate, and the morphology of these aggregates. Information gained using techniques such as atomic force microscopy, fluorescence quenching and neutron reflectivity is then reviewed. With this knowledge, the kinetic data obtained from relatively new techniques with high temporal resolution, such as ellipsometry and optical reflectometry, are examined. On this basis the likely mechanisms of adsorption are proposed. 2003 Elsevier Science B.V. All rights reserved.
Keywords: Surfactant aggregates; Adsorption to silica; Surfactant adsorption; Adsorption mechanism; Adsorption kinetics *Corresponding author. Tel.: q61-2-4921-8846; fax: q61-2-4921-5472. E-mail addresses: ewanless@mail.newcastle.edu.au (E.J. Wanless), rob.atkin@bristol.ac.uk (R. Atkin), vince.craig@anu.edu.au (V.S.J. Craig), s.r.biggs@leeds.ac.uk (S. Biggs). 0001-8686/03/$ - see front matter 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0001-8686(03)00002-2

220

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Contents
1. Introduction ............................................................................................ 2. General surfactant and substrate properties ..................................................... 2.1. Silica surface chemistry ....................................................................... 2.2. Surfactant properties ........................................................................... 3. Adsorption isotherms ................................................................................ 3.1. Introduction ...................................................................................... 3.2. Traditional analysis ............................................................................. 3.2.1. The two-step model ..................................................................... 3.2.2. The four-region model .................................................................. 3.2.3. Similarities between models ........................................................... 3.3. The influence of surfactant chain length .................................................. 3.4. The role of surface charge ................................................................... 3.4.1. Increases in surface charge with adsorption ........................................ 3.4.2. The common intersection point ....................................................... 3.4.3. Adsorption model based on the cip .................................................. 3.4.4. The influence of surface preparation ................................................ 3.4.5. Comparison of adsorption mechanisms on raw and acid washed silica ..... 3.4.6. Surface charge and gemini surfactant adsorption ................................. 3.4.6.1. The importance of the spacer group ........................................... 3.4.6.2. Gemini surfactant adsorption isotherms ...................................... 3.5. Evidence for discrete aggregation from adsorption isotherms ........................ 3.6. Calorimetry ...................................................................................... 3.6.1. The importance of surface water ..................................................... 3.6.2. Calorimetry and adsorption mechanism ............................................. 3.6.3. Interactions between the hydrocarbon tail and the surface ..................... 3.7. Summary of adsorption isotherms .......................................................... 4. Atomic force microscopy ........................................................................... 4.1. Introduction ...................................................................................... 4.2. The earliest images of surfactant aggregation: CTAB on graphite .................. 4.3. Graphite strongly orientates surfactant aggregates ...................................... 4.4. Adsorption studies on mica .................................................................. 4.4.1. Alkyltrimethylammonium halides on mica ......................................... 4.4.2. The influence of electrolyte on aggregate morphology .......................... 4.4.3. Gemini surfactants on mica ............................................................ 4.5. Adsorption studies on silica .................................................................. 4.5.1. The influence of electrolyte and counterion type ................................. 4.5.2. Adsorption kinetics measured by AFM ............................................. 4.5.3. The influence of counterion polarisability .......................................... 4.5.4. Gemini surfactant aggregates on silica .............................................. 4.6. Model hydrophobic substrates ............................................................... 4.7. Summary of AFM investigations ........................................................... 5. Fluorescence quenching experiments ............................................................. 5.1. Introduction ...................................................................................... 221 223 223 225 225 225 225 227 229 230 230 230 231 233 235 235 236 238 238 240 242 243 244 244 246 247 248 248 249 251 251 252 252 253 254 254 254 256 256 259 262 263 263

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

221

5.2. Time resolved fluorescence quenching .................................................... 5.3. Determination of aggregation numbers .................................................... 5.4. The earliest fluorescence probe studies .................................................... 5.5. Time resolved fluorescence quenching of adsorbed aggregates ...................... 5.6. Summary of fluorescence quenching experiments ...................................... 6. Reflectance techniques ............................................................................... 6.1. Introduction and underlying principles .................................................... 6.2. Neutron reflectivity and adsorbed layer structure ....................................... 6.2.1. Limitation of NR ......................................................................... 6.2.2. Contrast control ........................................................................... 6.2.3. NR studies of cationic surfactants on silica ........................................ 6.3. Ellipsometry, optical reflectometry and adsorption kinetics............................ 6.3.1. Dynamic aspects of surfactant adsorption .......................................... 6.3.2. Principles of optical techniques ....................................................... 6.3.3. Hydrodynamic considerations ......................................................... 6.3.4. Ellipsometric measurements of nonionic surfactant adsorption ................ 6.3.5. OR studies of nonionic surfactant adsorption ...................................... 6.3.6. Adsorption kinetics of CTAB on silica ............................................. 6.3.6.1. Ellipsometry ......................................................................... 6.3.6.2. Optical reflectometry .............................................................. 6.3.7. The role of micelles in adsorption ................................................... 6.3.8. The influence of electrolyte on adsorption ......................................... 6.3.9. The influence of co-ion type on CTAC adsorption .............................. 6.3.10. Effect of chain length on adsorption ............................................... 6.3.11. The slow adsorption region .......................................................... 6.3.12. Adsorption of gemini surfactants to silica by OR ............................... 6.3.13. Adsorption of ionic surfactants to a charged hydrophobic substrate ........ 6.4. Summary of reflectance observations ...................................................... 7. Summary ................................................................................................ 7.1. Mechanism of adsorption and the adsorption isotherm ................................ 7.1.1. The electrostatic concentration span ................................................. 7.1.2. The electrostatic and hydrophobic concentration span ........................... 7.1.3. The hydrophobic concentration span ................................................ 7.2. Adsorption kinetics and the adsorption isotherm......................................... References ..................................................................................................

264 265 265 265 266 268 268 268 268 269 269 272 272 272 273 273 276 277 277 278 278 280 282 283 285 291 293 296 298 298 300 300 300 301 301

1. Introduction The adsorption of a solute at the solidaqueous interface results in an increase in the local concentration or surface concentration. When the interaction is favourable the local concentration will exceed the concentration of the bulk solution. This is commonly referred to as a surface excess. For simple solutes, adsorption behaviour is generally uncomplicated, and can be modelled accurately on the basis of the interactions between the adsorbing species and the surface of the substrate. This

222

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

type of adsorption is generally interpreted using the Langmuir isotherm, which adequately describes adsorption behaviour up to a monolayer level of coverage. The behaviour of amphiphilic surfactant molecules at such an interface is more complex, and has been of considerable scientific interest since the concept of a hemimicelle was first proposed by Gaudin and Fuerstenau w1x. This study showed that the level of adsorption of both cationic and anionic surfactants on quartz increased only slightly up to a certain critical concentration, but once this concentration was exceeded, the surface excess increased markedly, indicating a cooperative adsorption process. This increase was attributed to the formation of adsorbed aggregates, termed hemimicelles, and the concentration at which a rapid increase in the surface excess occurs became known as the hemimicelle concentration (hmc). The aim of this manuscript is to review contemporary studies of surfactant adsorption at the solidaqueous interface in order to develop the most likely adsorption mechanism. In classical chemistry, all but the simplest chemical reactions are a consequence of several steps, or elementary reactions. The reaction mechanism, which describes the process by which the reactants are converted into products, generally consists of a series of such elementary reactions. For a given chemical reaction, there may be several plausible reaction mechanisms. In order to determine which mechanism best describes what actually occurs, the theoretical rate laws for the proposed elementary steps and overall reaction are compared to experimentally determined reaction rates. That is, the reaction mechanism is elucidated by studying the reaction kinetics. Analogously, an understanding of the adsorption kinetics is important to understanding the mechanisms of adsorption. The fast kinetics associated with the formation of an adsorbed surfactant layer at the solidaqueous interface has, until recently, prevented accurate investigation. This has led to the mechanism of surfactant adsorption being inferred from thermodynamic data or, in the analogy of a classical chemical reaction, by considering only the products. These explanations have been further hampered by a poor knowledge of the equilibrium adsorbed surfactant structures (the products). In many cases, these have been misinterpreted as simple monolayers or bilayers, rather than the discrete surface aggregates that are present in many surfactantsubstrate systems. To summarise, experimental limitations associated with kinetic and structural measurement have, until recently, hindered any determination of a satisfactory mechanism of surfactant adsorption at the solidaqueous interface. Here we aim to link the equilibrium and kinetic information in order to describe the adsorption process. In order to accomplish this, we shall first examine traditional equilibrium data and the most popular methods by which these data are interpreted. This is followed by a description of the evidence for discrete aggregation on the substrate, and the morphology of the aggregates formed. With this knowledge, the kinetic data obtained from relatively new ellipsometric methods of investigation will be examined, and on this basis the likely mechanism or mechanisms of adsorption will be proposed. To this end, particular attention will be given to literature that is representative of the present understanding in the field. These results will be supplemented with

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

223

other relevant works where appropriate. Nonionic surfactant adsorption will only be examined when the results give insight into the nature of ionic surfactant adsorption. This generally occurs when the interactions between the hydrocarbon chains and a hydrophobic surface are considered, but also may be of interest when examining a specific area of investigation, such as the adsorption kinetics. The head-group charge present on ionic surfactants results in a more complicated adsorption process when compared to nonionic amphiphiles. Ionic surfactant adsorption is particularly sensitive to the interactions of counter- and co-ions with the charged groups of the surface. Adjustment of the solution pH may also affect several factors in the surfactantysubstrate system. These can include the level of dissociation of surface groups, the degree of counterion binding to micelles, and the overall ionic strength. If the affinity of co-ions for surface groups is sufficiently high, then the co-ions can compete for adsorption sites at the surface. All these factors have implications, not only for the surface excess, but also for the morphology of the surface aggregates formed. For all of these reasons, systematic studies are required to quantify the nature of the various factors that control surfactant adsorption. The structure of the adsorbed layer has been elucidated by innovative experimental techniques, such as atomic force microscopy (AFM), neutron reflectivity (NR), and fluorescence spectroscopy. Knowledge of the substrate structure allows the general features of adsorption phenomena to be equated with the morphology of the surfactant aggregate present on the substrate. The intermolecular energetics will influence the type of structure formed and the macroscopic properties of the interface, which in turn affects the suitability of the surfactant for practical applications. These include ore flotation, stabilisation of foams and emulsions, wetting control, and detergency, amongst others. 2. General surfactant and substrate properties 2.1. Silica surface chemistry A significant proportion of the available literature concerning surfactant adsorption at the solidaqueous interface concerns amorphous silica w14x. As a result, we will briefly discuss the chemistry associated with silica. Specific differences between the silica surface and other substrates will be discussed when applicable. Note that many of the features described here for silica are relevant to other mineral oxide interfaces. Silica is by far the major constituent of the earths crust and as a result, the chemistry associated with the silica surface has been widely studied w5x. Bulk silica consists of siloxane units joined together in a tetrahedral lattice. Several different functional groups can be present at the surface, depending on the preparation of the surface and, if in solution, the nature of that solution. Functional groups commonly associated with the silica surface are depicted schematically in Fig. 1. Like other mineral oxide surfaces, silica has a surface charge character that is defined by the relative concentrations of Hq and OHy (the potential determining

224

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 1. Schematic representation of the types of functional groups that occur on the silica surface. (a) Hydrated and (b) anhydrous silanol groups are associated with the hydroxylated surface whereas (c) siloxane-dehydrated groups occur mainly on the pyrogenic surface. Redrawn after Ref. w5x(a).

ions) in solution, as shown by the following equations. SiOHqHqmSiOHq 2


K1

(1) (2)

SiOHqOHymSiOyqH2O
K2

It is the relative magnitude of the equilibrium constants K1 and K2 in Eqs. (1) and (2) that determine the charge on the silica surface. The isoelectric point (iep) for silica occurs at approximately pH 2 w5x, and is somewhat dependent on the exact nature of the surface. The density of negative charges remains low until the solution pH reaches 6, but increases sharply between pH 6 and 11 w6x. When compared to other well-characterised mineral oxide surfaces, the charge vs. pH curve for silica is unusual w2,6,7x, and modeling studies indicate that the surface potential of silica as a function of pH is highly non-Nernstian w8x. While solution depletion studies use silica particles (which typically have a high sodium content) the silica surfaces used in reflective techniques and AFM studies are often produced from silicon wafers. High purity silicon wafers are readily available commercially. The simplest method for preparing oxide layers on the surface of a wafer is to bake the wafer at high temperature in an oxygen atmosphere. By controlling the length of time that the wafer is baked, the oxide film thickness can be easily controlled. This process produces wafers of pyrogenic silica. Hydroxylated silica surfaces are prepared by rehydrolysing the surface, either by soaking the wafers in water or treatment with basic solution. When analysing the silica surface charge, the structure of the oxide layer must be considered. Hydroxylated silica has a high density of hydroxyl groups (;4.5 OH nmy2) w5x that are in close proximity to one another. This leads to hydrogen bonding between the hydrogen of one hydroxyl group and the oxygen of the neighbouring group, as depicted in Fig. 1b. Consequently, the hydroxyl hydrogen atoms are strongly bound at normal pH levels, resulting in the hydroxylated silica having a low surface charge. Conversely, pyrogenic silica has a lower density of

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

225

hydroxyl groups (;0.7 OH nmy2) w5x and a higher net charge than hydroxylated silica. The presence of numerous siloxane-dehydrated groups (Fig. 1c) will render the pyrogenic surface partially hydrophobic. 2.2. Surfactant properties Table 1 lists the name, structural formula, the most commonly used abbreviation and cmc of the surfactants covered in this review. In some cases, particularly for the alkyltrimethylammonium bromide surfactants, more than one acronym is used to refer to the same surfactant. In this review, the abbreviation used in the paper under consideration will be used so that the text corresponds to the reproduced figure. 3. Adsorption isotherms 3.1. Introduction Adsorption isotherms are traditionally determined by solution depletion methods w9x. Depletion experiments are accomplished by mixing a surfactant solution with a given mass of adsorbate of known surface area. After equilibration, the surface excess is determined by the change in the solution surfactant concentration. In order to facilitate measurement of solution concentrations, surfactants containing spectroscopically active groups are often, but not always, employed. A series of experiments conducted at appropriate surfactant concentrations allows the adsorption isotherm to be resolved. 3.2. Traditional analysis Much of the literature concerning adsorption isotherms predates in situ methods of probing the adsorbed layer morphology w9x. As a consequence, models proposed to reconcile the features of the isotherm, particularly the saturation surface excess, often describe simple monolayers and bilayers. This is in stark contrast to more recent data that in many cases suggests discrete surface aggregation. However, this does not discount isotherm analysis in developing an understanding of the adsorption process, particularly below the critical surface aggregation concentration (csac). In this pre-aggregation region of the isotherm, even the most recent experimental methods yield only inconclusive indications of adsorbed layer structure. Adsorption isotherms can provide particularly useful information concerning the electrostatic interactions that occur at low surfactant concentrations and also probe the manner in which the surface charge adapts as the solution conditions and surface excess are altered. In this section we will examine two of the more durable explanations for adsorption at a charged interface: the two-step and four-region adsorption models. Detailed attention will also be given to the influence of chain length, surface charge effects, the relevance of the common intersection point (cip) between isotherms

226 R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Table 1 Characteristics of surfactants reviewed Surfactant name Cetyltrimethylammonium bromide Cetyltrimethylammonium chloride Cetylpyridinium bromide Cetylpyridinium chloride Tetradecyltrimethylammonium bromide Dodecyltrimethylammonium bromide Dodecylpyridinium chloride Sodium dodecylsulphate Didodecyldimethylammonium bromide Benzyldimethyloctylammonium bromide Benzyldimethyldodecylammonium bromide Ethyl-a,v-bis (dodecyldimethylammonium bromide) Propyl-a,v-bis dodecyldimethylammonium bromide) Butyl-a,v-bis (dodecyldimethylammonium bromide) Hexyl-a,v-bis (dodecyldimethylammonium bromide) Octyl-a,v-bis (dodecyldimethylammonium bromide) Decyl-a,v-bis (dodecyldimethylammonium bromide) Dodecyl-a,v-bis (dodecyldimethylammonium bromide) Methyl groups (CH3) are abbreviated to Me. Structural formula C16H33NqMe3Bry C16H33NqMe3Cly C16H33 Nq (C2H2)2CHBry C16H33 Nq (C2H2)2CHCly C14H29NqMe3Bry C12H25NqMe3Bry C12H25 Nq (C2H2)2CHCly y C12H25SO4 Naq (C12H25)2NqMe2Bry C8H17NqCH2C6H5Me2Bry C12H25NqCH2C6H5Me2Bry C2H4(C12H25NqMe2 Bry)2 C3H6(C12H25NqMe2 Bry)2 C4H8(C12H25NqMe2 Bry)2 C6H12(C12H25NqMe2 Bry)2 C8H16(C12H25NqMe2 Bry)2 C10H20(C12H25NqMe2 Bry)2 C12H24(C12H25NqMe2 Bry)2 Acronymyabbreviation CTAB or HTAB CTAC CPBr CPC MTAB, C14TAB or TTAB DTAB or C12TAB DPC SDS DDAB BDOAB BDDAB 12-2-12 12-3-12 12-4-12 12-6-12 12-8-12 12-10-12 12-12-12 cmc (mM) 0.9 1.1 0.7 0.8 3.6 15.3 14.7 8.1 0.05 5.6 0.84 0.9 1.09 1.01 0.83 0.63 0.37 cmc (mM) 10 mM salt 0.15 0.3 0.1 0.15 2.1 11 10.5 6.5

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

227

Fig. 2. Adsorption isotherms of DTAq (s) and Bry (h) ions measured on particulate silica at pH 8. Note the adsorption of bromide ions only occurs once the second increase has commenced for the DTAq ions. The solid and dashed lines were drawn by hand to guide the eye. The solution cmc is indicated by a dashed vertical line. Reproduced from Ref. w10x.

measured at different salt concentrations and the consequences of different methods of substrate preparation. In general, adsorption isotherms are interpreted by discerning changes in the rate of increase in the surface excess with concentration. This allows the isotherm to be divided into regions, and the most likely conformation of adsorbed surfactant in each region ascertained. The most common approaches for this type of interpretation are the two-step and four-region adsorption isotherms. At first glance these models may appear to be fundamentally different, but in actual fact they have much in common. Both models divide the isotherm into four sections, and there is good agreement regarding the orientation of surfactant adsorbed at the interface in most regions. The primary difference between the models pertains to the region in which hemimicellar aggregation is initiated. The four-region model predicts that hemimicelle formation takes place in the second region, whilst the two-step model has hemimicelle formation occurring at higher solution concentrations, in the third region. In more recent studies, isotherm data are often combined with other information allowing more precise determination of the nature of adsorption. The surface charge, zeta potential, counterion concentration, solution pH and solution conductivity have been monitored with surface excess. As we shall see below, studies that combine techniques allow the adsorption mechanism to be commented upon with much greater certainty. 3.2.1. The two-step model When expressed on a linear scale, adsorption isotherms typically display two plateau regions w10x, and a sharp increase in the surface excess near the cmc. For a classical example see Fig. 2. Many descriptions of two-step isotherms are available in the literature for a wide variety of surfactantsubstrate combinations w1116x. In works published before the application of the AFM to imaging of adsorbed surfactant layers in 1994 w17x,

228

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 3. The two-step model for cationic surfactants adsorbed to silica. (a) The general shape of the adsorption isotherm. The x axis indicates residual surfactant concentration and the y axis indicates adsorption density. (b) The proposed model of adsorption. Adapted from Ref. w15x.

the shape of the isotherm is often interpreted as being indicative of a monolayer on hydrophobic surfaces and a bilayer on hydrophilic surfaces e.g. Fig. 2. A notable exception is the work of Gao et al. w15x. In their study of the adsorption of alkylpyridinium halides to silica, they determined two plateau regions in the adsorption isotherm. The plateau regions were at low surfactant concentrations (prehmc) and the saturation level plateau observed above the cmc. This led to the proposal of a two-step model for adsorption as shown in Fig. 3. The regions suggested were a low surface excess region (I), a first plateau region (II), a hydrophobic interaction region (III), and a second plateau (IV). It was suggested that in region (I) the surfactant is adsorbing via electrostatic interactions with the silica substrate. The surface excess is determined mainly by the surface charge. Adsorption is sparse, so interactions between adsorbed surfactant molecules are negligible. In region (II), the substrate surface charge has been neutralised. However, the solution activity of the surfactant is not sufficient to lead

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

229

Fig. 4. The four-region or reverse orientation model of adsorption. Proposed adsorption isotherm and surfactant aggregates on solid substrates. Adapted from Ref. w19x.

to any form of aggregation at the interface thus surfactants are still adsorbed as monomers. The abrupt increase in adsorption at the hmc denotes the onset of region (III). In this region, the solution surfactant concentration is sufficient to lead to hydrophobic interactions between monomers. The monomers electrostatically adsorbed in region (II) are thought to act as anchors (or nucleation sites) for the formation of hemimicelles. In this article, a hemimicelle was defined as a spherical structure with surfactant head-groups facing both towards the substrate and into solution w15x. In more recent times this type of structure has been redefined as an admicelle. In region (III) the admicellar structure was not necessarily fully formed, allowing for further adsorption. Region (IV) occurred above the cmc, with the formation of fully formed aggregates and saturation levels of surface coverage. 3.2.2. The four-region model Whilst this type of two-step analysis adequately explains many of the common features of adsorption isotherms, it is not the only method of evaluation available. Somasundaran and Fuerstenau proposed the four-region or reverse orientation model for interpretation of surfactant adsorption isotherms when plotted on a loglog scale w18x. This method has been shown to be particularly successful for modeling adsorption behaviour on alumina and rutile w4,18x. The primary advantage of using a loglog plot is that it amplifies the features of the isotherm at low surface excess values. The general form of isotherms plotted in this manner, and the morphology of adsorbed structures associated with each region are depicted schematically in Fig. 4. In region I of the isotherm, surfactant monomers are electrostatically adsorbed to the substrate, with head-groups in contact with the surface. Hydrocarbon tail-groups may interact with any hydrophobic regions of the substrate. Region II involves strong lateral interaction between adsorbed monomers, resulting in the formation of primary aggregates. Using techniques such as Raman spectroscopy, fluorescence spectroscopy, electron spin resonance and contact angle measurement, Somasundaran et al. w1922x have shown that the surfactants are adsorbed with head-groups facing towards the surface while the hydrocarbon tail-groups protrude into solution. This creates hydrophobic patches on the surface. In the four-region model, this type of aggregate is known as a hemimicelle. Increases in the surface excess in region III

230

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

are thought to result from growth of the structures formed in region II, without any increase in the number of surface aggregates. The presence of head-groups facing into solution renders the surface hydrophilic once more. The transition between region II and region III is thought to be due to neutralisation of the surface charge. Finally, in region IV, the surface morphology is assumed to be a fully formed bilayer. Further increases in the solution surfactant concentration do not lead to any further increases in the surface excess. 3.2.3. Similarities between models Clearly, these traditional types of analysis have a good deal in common. The most obvious difference between the models is a lack of hydrophobic interaction in the second region for the two-step model. Interestingly, in light of recent data, the structures proposed in the four-step model below charge neutralisation, and the admicellar structures predicted by the two-step model above charge neutralisation, may yet prove to be correct. 3.3. The influence of surfactant chain length As the hydrocarbon chain length of a surfactant molecule is increased, the monomer is essentially rendered more hydrophobic. That is, an increased number of clathrate bound water molecules are required to solubilise successively longer tail-groups, which lowers the overall entropy of the system. As a result, surfactants with longer hydrocarbon chains have a much greater driving force for aggregation, and this dramatically reduces the solution cmc, cf. Table 1. Chain length is also of critical importance in determining the adsorption behaviour of a surfactant. Fig. 5 shows that increasing the chain length by four methylene groups, from C12 to C16 (i.e. DPC to CPC) lowers the concentration at which the features of the isotherm occur by approximately an order of magnitude, in line with the reduction in solution cmc w23x. The shifting of the isotherm to lower concentrations for longer chained surfactants is a result of the increased hydrophobicity imparted by longer tail-groups. At the solidaqueous interface, hydrophobic interactions may exist between the surfactant and the surface, and also laterally between adsorbed surfactants. Some evidence for this is apparent in Fig. 5. As the increase in surface excess in regions II and III of the isotherm is dependent on lateral hydrophobic interactions, it would be expected that the surface excess should increase more rapidly with concentration for the surfactant bearing the longer hydrocarbon chain. This is indeed what is observed in Fig. 5b, as the slopes of regions II and III are clearly steeper for the C16 surfactant. In region IV the saturation surface excess is clearly greater for CPC than for DPC, but whether this was due to an increased level of hydrophobic interaction, or a change in the structure of the aggregate formed at the interface, could not be ascertained from this study. 3.4. The role of surface charge A major limitation of the solution depletion method for studying surfactant adsorption was observed by Goloub et al. w2x, who argued that the silica surface

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

231

Fig. 5. Adsorption isotherms for CPC and DPC with 0.001 M KCl at pH 9. (a) Presents the data on a loglog scale and the four-regions of the isotherm are indicated. (b) Shows the data on a linear-log scale. 1997 ACS. Reproduced with permission from Ref. w23x.

charge varies not only with pH, but also with surfactant adsorption. Ionisation of surface groups will alter the pH of the solution. This means that without careful pH control, pH changes may occur not only from isotherm to isotherm, but also along an isotherm. However, most studies report only the initial pH. 3.4.1. Increases in surface charge with adsorption In order to overcome this difficulty, Goloub et al. conducted a systematic study of the variation of surface charge with surfactant adsorption w2x. The solution pH was adjusted throughout equilibration of the surfactant and substrate until no further changes in pH were observed. The results obtained (an example of which is shown in Fig. 6) give valuable insight into the mechanism of the adsorption process. Fig. 6 shows that at both low electrolyte and low surfactant concentrations the adsorption and surface charge isotherms are practically identical. The greatest increase in surface charge occurs within this initial region of the isotherm and this effect is more pronounced with increased pH. The correlation between surface charge and adsorption at low concentrations suggests that whenever a surfactant is adsorbed to the surface a proton is displaced, which indicates that the surfactant head-group is in close proximity to the surface. The surfactant concentration at which the substrate surface charge is neutralised is denoted as the charge compen-

232

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 6. Adsorption of DPC and surface charge of silica at 0.001 M KCl as a function of the DPC concentration at (a) pH 7 and (b) pH 9. To facilitate the comparison, the surface charge is expressed as G0ss0yF, were F is the Faraday constant. In this way surface charge and the adsorbed amount of surfactant, Gs, are both expressed in micromole per square metre. 1996 ACS. Reproduced with permission from Ref. w2x.

sation point (ccp). Increasing the surfactant concentration above the ccp had little effect on the surface charge even though the surface excess continues to increase. This suggests adsorption of a second layer on top of the electrostatically adsorbed layer, with surfactant head-groups facing into solution. This interesting result was expanded upon by examining a plot of the surface charge vs. the surface excess of the surfactant, reproduced in Fig. 7, which shows that at low surfactant concentrations the surface excess is less than the native surface charge. In view of this it is somewhat surprising that the surface charge began to increase as soon as surfactant adsorption commenced. That is, rather than surfactant monomers first adsorbing to existing charged sites on the substrate, then creating additional charges, the adsorption of surfactant molecules causes nearby hydroxyl groups to immediately become more acidic, inducing further surface ionisation. As the charge neutralisation point was approached (intersection with the line of unit slope), the gradient of the surface charge isotherm was close to unity, which shows

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

233

Fig. 7. Surface charge of silica as a function of the surface excess of DPC measured at 0.001 M KCl and pH 7 and 9, and the line of unit gradient. 1996 ACS. Reproduced with permission from Ref. w2x.

that the number of surfactant molecules adsorbing and the number of surface sites created are nearly equal. Somewhat different results were obtained at high electrolyte concentrations. In 100 mM KCl, the initial surface charge was much higher, and the increase in surface charge on surfactant adsorption was much decreased. The surfactant ions were competing for charged sites on the substrate with the potassium ions of the electrolyte; hence adsorption did not reach measurable levels until much higher solution surfactant concentrations compared with the low electrolyte case. However, at high electrolyte concentrations, the Coulombic repulsions between the monomer head-groups was greatly decreased. As a consequence, once adsorption is commenced the isotherm increased steeply. Other authors have reported similar results w16x. The implications of high electrolyte concentration for the adsorbed morphology will be elucidated below in the discussion of AFM imaging. 3.4.2. The common intersection point De Keizer et al. w24x showed that the cip between adsorption isotherms measured at different electrolyte concentrations was a useful method of analysing adsorption isotherms. Further examples of the cip effect are provided in the work of Goloub and Koopal w23x. An example of the cip for DPC at pH 7 at two salt concentrations is presented in Fig. 8. At the surfactant concentration at which the cip occurs, added electrolyte has no effect on the surface excess. This condition may not hold in the case of longer chain surfactants that adsorb strongly at low concentrations. In the case of DPC, however, this observation allowed the effect of electrolyte on the adsorption process above and below the cip to be commented upon. It was suggested that, providing

234

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 8. Isotherms for DPC adsorption on silica at pH 7 for two salt concentrations on (a) a loglog and (b) a linear-log scale. 1997 ACS. Reproduced with permission from Ref. w23x.

there are no specific interactions between the electrolyte and the substrate, the cip corresponds to the iep of the substrate. In its simplest form, the cip represents the point where the orientation of surfactants adsorbing at the surface changes from heads facing towards the substrate to heads facing towards solution, forming bilayered aggregates. In order to test the validity of the cip as a means of identifying the iep of the adsorbent, the variation in electrophoretic mobility in the presence of surfactant was also investigated. This result is reproduced and shown in Fig. 9 for two pH values and salt concentrations w23x. The iep results for DPC and its C16 analogue CPC both correlate with the cip. These data are directly comparable to the surface charge isotherm already discussed in Section 3.4.1. Not only did the cip correspond to the iep, but also to the ccp, cf. Fig. 6. The fact that these three points occurred at the same bulk surfactant concentration showed that there is little or no specific adsorption of the electrolyte to the substrate i.e. no adsorption beyond the level dictated by Coulombic attraction. Thus, the cip represents the point where the electrostatic contribution to the adsorption process changes from attractive to repulsive.

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

235

Fig. 9. Electrophoretic mobilities of silica particles as a function of DPC concentration at two salt concentrations and (a) pH 7 and (b) pH 9. 1997 ACS. Reproduced with permission from Ref. w23x.

3.4.3. Adsorption model based on the cip It has been postulated that, below the cip where adsorption is primarily electrostatically driven, adsorption is decreased with increasing electrolyte concentration because of competition between electrolyte co-ions and the surfactant monomers. As with the other models described, surfactant adsorption in this region is orientated with head-groups towards the surface due to electrostatic attraction. Above the cip, hydrophobic interactions between surfactant tail-groups provide the driving force for further adsorption. As electrolyte reduces the Coulombic repulsions between the surfactant head-groups, increasing the electrolyte concentration above the cip enhances surfactant adsorption. This can be observed in Fig. 9. 3.4.4. The influence of surface preparation It has been shown that the solution conditions are not the only factors that influence the surface charge of the substrate. The effect of different methods of surface preparation was investigated by Chorro et al. w25x, who found that acid treatment of the substrate prior to adsorption could reduce the maximum surface excess by almost 50%. In this depletion study, differences in the adsorption isotherms on raw and HCl washed particulate silica, designated SiNa and SiH, respectively,

236

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

were investigated for DTAB and the gemini surfactant 12-2-12. For the moment we shall direct our attention towards the general implications of this study. A more thorough treatment of gemini surfactant adsorption and surface charge will then be undertaken in Section 3.4.6. This study was performed under free conditions i.e. no attempt was made to control the solution conditions. Thus, the ionic strength of the major species in the solution (surfactant ions, counterions, Hq , Naq and OHy) regulated the behaviour of the system. This may appear to be precisely the type of study that Goloub et al. w2x had described as flawed due to a lack of control over solution conditions. However, in this case, as the level of surfactant adsorption increased, the properties of the supernatant (pH, electrophoretic mobility, conductivity and counterion concentration) were closely monitored. This allowed the effect of surfactant adsorption on the system as a whole to be monitored and is therefore a valuable means of investigation. Upon immersion of the silica and surfactant in solution, the presence of sodium ions was noted. It was shown that the sodium content for acid washed silica (7 ppm) was considerably less than that of raw silica (67 ppm). Thus, the HCl washing technique was particularly successful for the removal of sodium from the surface and the majority of residual sodium ions were strongly surface bound. These ions were released during equilibration, and not during washing, due to the much longer time period of the equilibration. The adsorption isotherms for DTAB and 12-2-12 are reproduced in Fig. 10, with the concentration axis presented as a function of the cmc. The cmc of DTAB in the supernatant is 12.8 and 11.8 mM for the SiH and SiNa systems, respectively. For 12-2-12 the cmc is significantly reduced at 0.8 and 0.4 mM for the SiH and SiNa systems. All of these cmc values are less than that of the corresponding surfactant in pure water, reflecting the contribution of the released sodium ions to the solution ionic strength. The most startling difference between the two substrates is that, for both surfactants, almost double the surface excess was obtained on raw silica as opposed to the acid treated silica. These differences were attributed to the different charging properties of the surface. When sodium ions were released from the surface, this resulted in the formation of ionised sites i.e. the raw silica surface is considerably more charged than the acid washed substrate. The released sodium ions also raised the ionic strength of both systems, but obviously this effect is much greater for the raw silica system. The marked difference in plateau surface excess values reflects not only the effect of increased ionic strength, but also the importance of the number of initial surface charges on the adsorption process. This work highlights the significant effects that changes in surface chemistry induced by surfactant adsorption can have on adsorption behaviour and highlights a major weakness in many depletion studies. 3.4.5. Comparison of adsorption mechanisms on raw and acid washed silica Surfactant monomers initially adsorb to pre-existing charged sites electrostatically, and these act as nucleation points for further surfactant adsorption. The acidity of nearby hydroxyl groups increases, releasing Hq ions into solution i.e. adsorption

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

237

Fig. 10. The adsorption of DTAB and 12-2-12 on (a) SiNa and on (b) SiH reproduced from Ref. w25x.

that leads to ionisation of surface groups is detectable by a decrease in the solution pH. This was observed for all surfactantsubstrate combinations but was more obvious for the acid washed silica due to the increased number of protons associated with this substrate. Thus, when a free system is under investigation, the equilibrium concentration of sodium ions in solution relates to the initial surface charge, while changes in the Hq concentration are a measure of the number of surface charge sites that are induced by surfactant adsorption. The acid washed surface in Fig. 10b is obviously chemically different from the silica used in the studies described previously. Nonetheless, the results obtained are interesting, particularly when contrasted with the results for raw silica. For the DTABySiH system, the pH of the supernatant was decreasing gently up to the point where the surface excess reached 25% of the maximum value i.e. the end of the first pseudo-plateau in the isotherm. Electrophoretic data indicated that the pzc was reached at this surface excess and the adsorption of bromide ions at the surface was shown to be very low. These data suggest that up to the end of the first pseudo-

238

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

plateau, the DTAB was adsorbed head-group towards the substrate due to electrostatic interactions. In the region of rapid increase in surface excess, the fraction of surface bound bromide ions increased. A rapid decrease in the supernatant pH accompanied this bromide ion adsorption, but only until the solution Hq concentration reached 60% of the maximum value. This suggests that above this surface excess, surfactant was adsorbed with head-groups facing towards solution. Increases in the electrophoretic mobility of the particles supported a change in the orientation of the adsorbing surfactant at this surface excess. It was postulated that surface bound aggregates were the most likely adsorbed morphology to account for these observations. These aggregates could be loosely packed initially, but as saturation levels of coverage were approached, aggregates would become tightly packed. The major difference between the acid washed and raw silica surface was in the ionisation of surface groups. For the raw silica surface, the solution pH reached a plateau level at approximately 25% coverage of the surface i.e. the concentration where surfactant is adsorbing exclusively with head-groups facing into solution is reached much earlier for the raw silica system. Thus, the model for adsorption that the data on raw silica implies shows good agreement with that described by Goloub and Koopal w23x. 3.4.6. Surface charge and gemini surfactant adsorption Gemini surfactants w26x are a relatively new genre of amphiphilic molecules, first appearing in the literature in 1974 w27x. They have recently become a topic of revived scientific interest, due in part to their effectiveness in the modification of interfacial properties, but also because their unusual molecular geometries lead to interesting aggregation structures. A gemini surfactant consists of two identical surfactant molecules joined by an alkyl spacer group. The spacer group can be flexible or rigid w28x, hydrophilic or hydrophobic w29x and generally connects the two surfactant moieties at, or near, the head-group. The attachment of the spacer group increases the hydrophobicity of the dimeric surfactant relative to the constituent monomeric units. As a consequence, the cmc of the gemini can be up to 100 times lower than that of the monomer units w30x. For simplicity, shorthand nomenclature of gemini surfactants is often employed, based on the number of carbon atoms in the surfactant chain and the spacer group, and is best illustrated by example. For alkanediyl-a,v-bis (dodecyldimethylammonium bromide) dimeric surfactants with the alkanediyl spacer groups C2H4, or C8H20, the corresponding surfactants are referred to as 12-2-12, and 12-8-12, respectively. The molecular structure of a typical gemini surfactant is depicted in Fig. 11. All discussion is limited to this family (12-s-12) of gemini surfactants. 3.4.6.1. The importance of the spacer group. The properties of gemini surfactants are greatly dependent on the length of the spacer group. The spacer group controls the separation between the two head-groups and may be greater or less than the average separation of the corresponding monomers in an aggregate. This changes

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

239

Fig. 11. The structural formula of a typical gemini surfactant. The molecule represented is 12-4-12. Counterions have not been included.

the mobility and the packing geometry of the gemini within a micelle, whether in solution or at an interface. Danino et al. w31x have demonstrated that the structure of the micelles formed in solution varies significantly with spacer length. For quaternary ammonium surfactants of the form 12-s-12, spacer groups of length less than or equal to five methylene groups dictate that the head-groups are in close proximity. As the Bjerrum length in water at 25 8C is equal to 0.7 nm w32x, for gemini surfactants with short spacer lengths, charge condensation must occur, and the effective charge of the surfactant is less than 2. The resultant monomer geometry leads to aggregates of lower curvature than that of the corresponding monomer. For s values between 6 and 10, the distance between head-groups induced by the spacer is similar to that of the monomer in a micellar aggregate, and similar structures result. For s values greater than 14, it is suggested that the spacer adopts a looped conformation within the aggregate, thus acting like additional hydrocarbon chains. The structure formed in this case is similar to those of dimeric surfactants. More specifically, for ss2 worm-like micelles result, ss3 gives rise to extended micelles, while for s greater than 4 essentially spherical micelles are formed w31,3335x. The effect of variation in the length of the spacer group has been extensively investigated at the solutionair interface. Perhaps not surprisingly, it has been demonstrated that the surface area occupied per surfactant molecule increases with the size of the spacer for s between 3 and 10 w36x. The behaviour of gemini surfactants at the solidliquid interface has been shown to follow similar trends. The first adsorption isotherms for a gemini surfactant at the solidaqueous interface were determined by Esumi et al. The adsorption of 12-2-12 was investigated

240

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

at the silicaaqueous interface w37x, the laponite clayaqueous interface w38x and the titaniumaqueous interface w39x. It was shown that the amount of 12-2-12 adsorbed at the silicaaqueous interface, was lower than that of the monomeric analogue, DTAB. However, while some of the macroscopic properties of the adsorbed aggregates were elucidated, no consideration was given to the adsorption mechanism. This initial study motivated subsequent research w25,40,41x. A systematic investigation of the effect of spacer length on the adsorption isotherm, and the mechanism of adsorption was initially undertaken w40x. A subsequent paper dealt with the effects of a variation in surface preparation, which has already been discussed in Section 3.4.5 w25x. We shall review the effect of spacer length and mechanism of adsorption first, before returning to the effects of acid washing the substrate prior to adsorption of gemini surfactants. 3.4.6.2. Gemini surfactant adsorption isotherms. Adsorption isotherms for 12-2-12, 12-4-12, 12-6-12 and 12-10-12 on acid washed silica are presented in Fig. 12 w40x. As the size of the spacer group increased the maximum surface excess of surfactant was decreased. Corresponding results have been reported at the airwater interface w36x, and have been attributed to increasing head-group area. Similarly, at the silica aqueous interface it is likely that the adsorbed morphology alters as the spacer length is varied. This will be demonstrated in Section 4.5.4. Interestingly, although Esumi et al. w37x did not pre-treat their surface to remove sodium ions, the maximum surface excess attained for 12-2-12 was similar in both studies. This suggests that all of the surface bound sodium ions were exchanged by the surfactant. Electrophoretic data showed that the silica was substantially negatively charged at the beginning of adsorption, and that the amount of surfactant adsorbed at the point of zero charge was the same irrespective of the spacer length and corresponded to an area occupied per surfactant of 25 nm2. This equates to an average distance between monomers of 5 nm. As this value is much larger than the length of a fully extended spacer group (;1.4 nm for 12-10-12) each surfactant can only neutralise one surface charge site. The observed zero net charge implied that the second headgroup, which is not surface bound, had a bromide ion associated with it. Alternatively, the close proximity of the unattached head-group may have resulted in the formation of a charged site at the surface, which was then associated with the headgroup. A third possibility is that the number of unassociated surfactant head-groups matched the number of free charged sites on the surface, thereby achieving electroneutrality. All of the gemini surfactant adsorption isotherms exhibited a plateau after the point of zero charge was reached. As can be seen in Fig. 12, this plateau was much narrower for 12-10-12 than for the other surfactants. Surprisingly, the surface excess of gemini surfactant required for charge neutralisation was 5 times less than that for DTAB cf. Fig. 10. Recall that both electrostatic and hydrophobic interactions were involved in the first step of the adsorption process for DTAB. This result suggests that the surface charge is indeed redistributed once gemini surfactants with short spacers were adsorbed to the surface. Thus, for spacer lengths less than or equal to

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

241

Fig. 12. Adsorption isotherms of 12-2-12 (circles); 12-4-12 (triangles); 12-6-12 (squares); and 12-1012 (diamonds) on silica at 25 8C. The surface excess is expressed in micromole of surfactant per gram of silica. The concentration scale has been normalised by the appropriate surfactant cmc. Solid lines are guides for the eye. The surface area per gram was determined by the BET method to be 29 m2. Reproduced from Ref. w40x.

6, the adsorption mechanism operating in the first step is different to that of the monomeric analogue. Once the gemini is adsorbed at the surface, the second charged head-group is brought into close proximity with the surface, an effect which becomes more pronounced as the spacer length is decreased. The acidity of nearby silanol groups is increased, raising the likelihood of local ionisation. As a decrease in the solution pH was not noted at this time, it was postulated that the charged sites were created by desorption of surface bound sodium ions that were not removed during the washing process. Increasing the spacer length allows the second head-group to be positioned further from the surface. In this case, the second head-group is more likely to be neutralised by a bromide ion. Thus, gemini surfactants with long spacer groups behave more like their monomeric analogue, with an overlap between the first and second steps in the adsorption process. The decreased size of the first plateau for 12-10-12 suggested that it was acting more like a monomeric surfactant than its counterparts with shorter spacer lengths. The second step in the adsorption process is thought to be due to lateral hydrophobic interactions. As the solution surfactant concentration is increased, interactions between electrostatically adsorbed monomers and the adsorbing surfactants are more likely, which leads to aggregate formation. It was expected that the concentration at which the surface excess rises appreciably should be lowest for the surfactant with the lowest solution cmc. This is indeed what was observed, with the steeply rising region of the isotherm occurring in the order 12-10-12-12-2-1212-6-12f12-4-12, cf. Table 1. A drop in the pH of the supernatant accompanied this step, which showed that the high positive charge density of the surfactant

242

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 13. Physicochemical characteristics of 12-2-12 adsorption on SiH: the degree of surface coverage u; the conductivity of the equilibrated supernatant, the pH, and the degree of bromide ion association to the surface b are plotted against the concentration of surfactant in the equilibrated supernatant. Reproduced from Ref. w25x.

aggregates induced the formation of surface charge sites. Recent optical reflectometry (OR) studies reported somewhat different findings as will be discussed in Section 6.3.12. We now return to the adsorption of 12-2-12 on acid washed and raw silica. As the study described above was for acid washed silica, the mechanism for the SiHy 12-2-12 system has already been outlined above. Variation in the physicochemical properties of the SiHy12-2-12 system with solution surfactant concentration can be seen in Fig. 13. It is important to note that along with the decrease in solution pH that accompanies the sharp increase in surface excess, the degree of bromide ion adsorption reaches a plateau at approximately 0.5 mM. This value is similar to that obtained for 12-2-12 micelles in solution w42x, and is indicative of an adsorption process that was hydrophobically driven. 12-2-12 was proposed w25x to have the same mechanism of adsorption on raw silica. The quantitative differences in the adsorption isotherms are thought to be due mostly to the greater capacity of the raw silica surface to release sodium ions, thereby increasing the ionic strength of the supernatant. The pzc for the SiNay122-12 system was significantly higher than that of the SiH surface, and corresponded to a smaller area per adsorbed molecule (only 2.7 nm2). However, once the initial charged surface sites were neutralised, there was no further decrease in the pH of the supernatant. This shows that the second adsorption step had already begun and aggregation was occurring at the surface. 3.5. Evidence for discrete aggregation from adsorption isotherms Adsorption isotherms yield significant information concerning the nature of the interactions between the surface and the surfactant, particularly in the initial stages

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

243

of adsorption. However, it is difficult to infer adsorbed structures from this type of data, and the traditional monolayerybilayer interpretation has been the favoured model until recently. Despite this, some indirect evidence for admicelle structures may be inferred from adsorption isotherms. When one compares isotherms obtained with different surfactants, counterions, and degrees of surface modification, it becomes difficult to rationalise the results simply in terms of mono- and bi-layer aggregation. Furthermore the surface excess values obtained usually do not correlate with the levels that would be expected for complete monolayers or bilayers. This has been rationalised as being due to patchy coverage, which of course in itself infers a more discrete structure. Some of the most persuasive isotherm data for the presence of interfacial aggregates is that obtained with the salicylate ion. Several studies have shown that the surface excess of surfactants with pyridinium-based head-groups on silica depends strongly on the counterion. Leimbach et al. w43x first demonstrated for tetradecylpyridinium that a sixfold increase in the plateau surface excess occurs when the counterion is changed from the weakly binding chloride ion to the strongly binding salicylate ion. Although this result may be deceptive due to the reported specific interactions between the salicylate ion and pyridinium head-groups w44 47x, it is nonetheless very difficult to explain this type of increase only in terms of mono- and bi-layer aggregation. However, it is relatively easy to envisage a situation of increased aggregate growth around electrostatically bound surfactants, given the well-known increase in solution aggregation number as the degree of counterion binding is increased. Further support for this aggregation model can be found in isotherm data obtained on hydrophobically modified silica. Leimbach and Rupprecht w48x covalently attached a low concentration of octadecyl groups to a silica surface, thereby creating anchor sites for surface aggregation. The concentration of the hydrophobic groups was such that only 7% of surface hydroxyl sites were occupied. Thus the surface was negatively charged and hydrophobic. The adsorption of the anionic surfactant SDS to the treated and untreated silica was investigated. On the untreated silica, SDS did not adsorb to detectable levels. However, on the modified silica a one step isotherm was obtained both with and without added electrolyte. The saturation adsorption density was 0.4 and 1.4 mmol my2 for the no added electrolyte and 0.1 mM NaCl system, respectively. The increase in adsorption density in electrolyte was justified on the basis of decreased repulsions between adsorbed surfactant headgroups, and the surfactant and the substrate, which carried the same charge. This model provides convincing evidence for electrostatically adsorbed surfactant monomers acting as nucleation sites for further adsorption. This hypothesis will be further probed in Section 5. 3.6. Calorimetry Adsorption isotherms can be complemented by measurement of the heat of adsorption (calorimetry w49x). This allows the energetics of adsorption to be monitored throughout an adsorption isotherm. Much useful information has been

244

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

obtained by calorimetry concerning the mechanism of adsorption, without actually revealing a definitive surface structure. Comparative measurements can allow the nature of the exchange of cations, the conformation of adsorbed water and the orientation of the surfactant with respect to the surface to be commented upon, amongst other issues. There are several methodologies by which heats of adsorption may be obtained w5052x. Flow calorimetry w52x is one technique that illustrates the fundamentals associated with the measurement of adsorption energetics, and is basically an extension of ion exchange chromatography. The surfactant solution is passed through a column containing the powdered form of the substrate, for example silica or alumina. This column is contained within a microcalorimetry chamber, and a constant temperature is achieved by means of a feedback loop that controls the power supply to a heating coil. Initially the column is equilibrated under flow of pure solvent; then a known quantity of surfactant is passed into the column. The magnitude of the energy required to maintain the temperature of the column is monitored, and a plot of heat flow vs. time deduced. The net heat transfer is simply the area under this curve. As this process is isothermal, this area directly equates to the heat of adsorption, which can be converted to the molar enthalpy of adsorption by dividing by the molar surface excess. 3.6.1. The importance of surface water The role of water in the energetics of adsorption cannot be underestimated. In order to adsorb to a substrate, an incoming surfactant may need to displace water of hydration at the solid surface. The influence of the solid on the arrangement of the adjacent water molecules will depend upon the properties of that surface. For example, sodium cations specifically bound at the substrate attract free water molecules. This leads to a local ordering of the water molecules at the interface. The ability of small metal ions in bulk solution to induce structure in nearby water molecules has been recognised for some time w53x. The likelihood of an analogous effect at the solidliquid interface has also been discussed w54x. Thus, endothermic contributions to the heat of adsorption will depend on the co-ion concentration both at the surface and in the bulk. Other specific interactions between water and silica may limit the ability of the surfactant to adsorb to the surface other than by Coulombic interactions. 3.6.2. Calorimetry and adsorption mechanism Fig. 14a shows the adsorption isotherms for DTAq and TTAq ions on silica at pH 8.3 w55,56x. It is worth noting that the TTAq isotherm was shifted to the left relative to DTAq, but the general shape of the isotherm was similar. This shift was due to the longer tail-group of TTAq. A longer hydrocarbon chain provides a greater driving force for aggregation (evidenced by a lower solution cmc cf. Table 1), and results in the entire isotherm being compressed relative to DTAq. Both surfactants reach their saturation surface excess slightly below their respective solution cmc values. The corresponding heats of adsorption are presented in Fig. 14b.

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

245

Fig. 14. The adsorption of DTAq (squares) and TTAq (circles) ions measured on silica at 25 8C and pH 8.2. (a) The adsorption isotherms; and (b) the heats of adsorption. The solid lines were drawn by hand to aid the eye. 1994 ACS. Reproduced with permission from Ref. w55x.

The change in enthalpy during the adsorption process was dependent on the properties of the bulk and adsorbed phases. Fig. 14b shows that the enthalpy initially decreased as the surface excess increases. This was due to the displacement of surface cations and water molecules from successively more strongly bound sites. As the surface excess increased two important effects came into play. Firstly, the rate of ion exchange was reduced, as there were fewer exchangeable ions present on the surface. Secondly, strong lateral interactions between the tail-groups of the adsorbed surfactants led to a perpendicular orientation of the hydrocarbon chains relative to the surface. These two effects led to the observed minimum, then subsequent increase in the heat of adsorption. The energetic state of interfacial water molecules was less affected and the overall heat of adsorption eventually became endothermic. The region of monotonic increase of enthalpy with the degree of surface coverage corresponds well with the sharply increasing region of the adsorption isotherm. Adsorption in this region is entropically driven and is dominated by intermolecular interactions, similar to those that lead to micellisation in the bulk. Bulk micellisation is also an endothermic process for most surfactants, with the driving force derived from the entropy gained upon aggregation w57x. This entropy increase is acquired from the release of clathrate bound water molecules associated with the tail-groups into the bulk solution upon micellisation. In some cases it has been shown that the heat of adsorption at moderate to high concentrations is remarkably similar to that of bulk micellisation w58x. Moreover, at high surface excess values, the temperature variation in the adsorption enthalpy mirrors that of the micellisation enthalpy w59x. These results have been used to argue for the presence of surface bound micelles.

246

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 15. The influence of the hydrophobic tail length on the adsorption energetics for benzyldimethylalkylammonium bromides onto precipitated silica from aqueous solutions at 298 K with an initial pH of 8.3: the differential molar enthalpies of displacement against the quantity of adsorption in a limited adsorption range. 1996 ACS. Reproduced with permission from Ref. w60x.

3.6.3. Interactions between the hydrocarbon tail and the surface Zajac et al. w60x investigated the effect of surfactant chain length on the enthalpy of adsorption at low surface excess values. Specific interactions between the surfactant tail-groups and the substrate brought about the desorption of structured interfacial water. This provided a significant endothermic contribution due to the heat of adsorption. The heats of adsorption of benzyltrimethylammonium bromide (BTMAB), benzyldimethyloctylammonium bromide (BDOAB) and benzyldimethyldodecylammonium bromide (BDDAB) are reproduced in Fig. 15. For surface excess values of up to 20 mmol gy1, enthalpies of adsorption of the C8 tailed BDOAB and its head-group BTMAB were negative and indistinguishable. This suggested that the short C8 alkyl chain did not interact with the silica surface and the most likely orientation for the tail-group was perpendicular to the substrate. Conversely, the C12 BDDAB had a positive enthalpy of adsorption throughout the same range of surface excess values. It was suggested that interactions between the longer C12 tail-group and the substrate led to disruption of the structured interfacial water. This de-wetted the silica and made a significant contribution to the energetics of the adsorption process. These interactions were made possible by the additional conformations available to a C12 tail-group over a C8 chain. Thus, at low surface coverage values, the C12 tail-group is oriented parallel to the surface to some degree. It would be expected that the surfactant tail-group would interact more strongly with a hydrophobic graphite substrate. The recent calorimetric study of Kiraly and Findenegg w59x used heat of adsorption data to determine whether the most likely conformation of C12TAB adsorbed to graphite was the classical reorientation model

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

247

(where the adsorbed surfactant molecules are reoriented from horizontal to a vertical position, accompanied by further adsorption from solution), or that of an interfacial aggregate (formation of hemicylindrical admicelles templated by an epitaxially bound surfactant monolayer). The calorimetric evidence, based on the displacement of water from the interface, showed that the adsorption process had two distinct phases. The first phase, in which the surfactant molecules are horizontally adsorbed as a monolayer, was strongly exothermic and, surprisingly, appeared to be independent of the ambient temperature (in the range 288318 K) and the surface coverage. The second phase was less exothermic than the first phase and weakly dependent on the level of surface coverage. The second stage was, however, inversely dependent on temperature. This important result strongly suggests a high degree of intermolecular cooperativity between neighbouring adsorbate molecules, which was extremely difficult to reconcile on the basis of the reorientation model. On this basis, the authors concluded that the most concordant aggregate morphology was of hemicylindrical aggregates, as suggested by AFM imaging studies. These AFM studies, and more detail of the structure of the adsorbed surfactant layer at the graphite water interface is discussed below in Section 4.3. 3.7. Summary of adsorption isotherms The study of adsorption isotherms by depletion methods continues to be an effective means of studying surfactant adsorption at the most fundamental level, however, future efforts must in all cases consider the possibility that changes in the surface chemistry during surfactant adsorption will influence the bulk concentration of various species and this in turn will influence the surfactant adsorption. Models have been available for some time to explain the features of adsorption isotherms, and it would seem that the most durable is the four-region model for surfactant adsorption. However, in light of recent evidence the interpretation of the final step in the four-step isotherm must be modified to account for aggregate formation. Based on the data reviewed in Section 3, the two-step model proposed by Gu et al. w15,16x (cf. Fig. 3) would seem to be invalid, as it fails to account for the increase in surface charge and lateral hydrophobic interactions that occur in the second region of the isotherm. It was demonstrated in Section 3.3 that increasing the hydrocarbon chain length of the surfactant, which increases the hydrophobicity of the monomer, displaces the adsorption isotherm to lower bulk concentrations. The rate of increase of surface excess with concentration in regions II and III of the isotherm, in which adsorption is partially or wholly hydrophobically driven, respectively, is more rapid for surfactants with longer tail-groups. Both of these effects become more pronounced as the surfactant chain length is successively increased. The critical intersection point (cip) between isotherms of the same surfactant at different salt concentrations denotes the bulk concentration at which the electrostatic contribution to adsorption changes from attractive to repulsive. At this concentration the orientation of adsorbing surfactant molecules switches from head-groups facing towards the substrate to head-groups facing into solution. Below the cip, the addition

248

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

of electrolyte lowers adsorption by competing with monomers for charged surface sites. Above the cip added electrolyte lowers the electrostatic repulsions experienced between surfactants in aggregates, permitting tighter packing. This increases the surface excess relative to the no added electrolyte case. Acid washing of the silica substrate prior to surfactant adsorption lowers the number of charged surface sites and decreases the number of surfactant molecules adsorbed in region I. Thus, there are fewer nucleation sites for continued adsorption lowering the saturation surface excess for the acid washed substrate. Gemini surfactants are a relatively new class of surfactants that exhibit interesting adsorption behaviour. Most importantly, the saturation surface excess has been shown to be strongly dependent on the length of the spacer group, with shorter spacer length gemini surfactants having the highest saturation surface excess values. The adsorption mechanism for gemini surfactants differs from that of their monomeric analogue only in the initial stages (region I). For spacer lengths greater than 6, it appears that the gemini surfactant is adsorbed to the substrate by one headgroup, with the other head-group electrostatically bound to a bromide co-ion some distance from the interface. For spacer lengths less than 6, the second head-group is held in sufficiently close proximity to the substrate to induce ionisation of a surface hydroxyl group, and both head-groups are bound to the substrate. Calorimetric investigations show that the first step in the adsorption process involves significant displacement of adsorbed water. The energetics of subsequent surfactant adsorption involve a significant cooperative hydrophobic interaction, such that the heat of adsorption is similar to bulk micellisation. Calorimetry also shows that surfactant molecules with chain lengths of 12 or more carbon atoms interact with hydrophobic areas on the silica substrate presumably by adopting a flat conformation on the surface at low coverage levels. While some indirect evidence for the presence of discrete surface aggregates can be obtained from adsorption isotherms and calorimetry, the major disadvantage of these techniques is the lack of information they provide concerning the adsorbed morphology. Without knowledge of the adsorbed layer structure, surfactant adsorption models must always be somewhat speculative. The remainder of this review will essentially deal with techniques that aid in elucidating the precise nature of the adsorbed aggregate. With this knowledge, the most likely adsorption mechanism and the structures present in each region of the isotherm will be presented in Section 7. 4. Atomic force microscopy 4.1. Introduction Perhaps the greatest advance in the study of surfactant adsorption at the solid liquid interface of the last decade is the development of techniques to obtain in situ images of the adsorbed aggregates using the AFM. Imaging has allowed adsorption isotherms to be analysed with knowledge of the aggregate morphology and complemented model dependent techniques such as neutron reflectivity to provide

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

249

greater accuracy. AFM is particularly well suited to detecting the periodicity of discrete surface aggregates i.e. a peak to peak separation. However, AFM is not without its limitations. In order to image aggregates, a surfactant layer must have head-groups facing solution to provide a repulsive force. Thus meaningful AFM data are generally obtained only above the csac, and most AFM experiments are carried out at concentrations greater than the cmc. As only the adsorbed layer facing solution is scanned, for a bilayered aggregate the nature of the underlying layer remains speculative. AFM provides no information concerning the surface excess and the density of adsorption between aggregates is also open to speculation. As a consequence, several possible morphologies can adequately justify a given image, particularly for bilayered aggregates. However, when AFM results are analysed with information from other forms of adsorption data, the most likely structure can be predicted with reasonable confidence. Thus, AFM is most useful when used in conjunction with other measurements. Traditionally, the AFM has been used with the probe in contact with the substrate. While this is suitable for the imaging of a hard surface, it is inappropriate for imaging adsorbed surfactants. Surfactant aggregates are generally fragile and hard contact measurements will destroy the adsorbed morphology. However, the sensitivity of the AFM is such that it is possible to sense the repulsive electrical double layer associated with the adsorbed species. By using a fly height of approximately 1 nm above the adsorbed layer, sufficient contrast is obtained to image the aggregate structure. This type of imaging has been termed soft contact w61x and lowers, but does not eliminate, the risk of aggregate deformation. A considerable volume of surfactant adsorption research predates AFM. Some AFM experimentation has been aimed at better understanding known surface and solution effects. For example, for decades it has been observed that the addition of salt increases the surface excess of surfactant, and the AFM has been used to elucidate the precise response of the adsorbed aggregates to electrolyte. However, other AFM investigations have been motivated solely by a desire to manipulate aggregate morphology, accomplished by studying surfactants with unusual geometries or by specific surface modification. Both types of experimentation have greatly added to our understanding of adsorption phenomena. The size, shape and spacing of adsorbed surfactant aggregates are dependent upon the intermolecular and surfactantsubstrate interactions w62x. These interactions are strongly influenced by solution conditions such as the ionic strength and pH. The solution micellar size and shape is strongly related to the geometry of the monomer, however at a surface this structure must be reconciled with the confinements imposed by the substrate. Nonetheless, the arrangement of aggregates at an interface is often analogous to solution structure observed at higher concentrations w63x. 4.2. The earliest images of surfactant aggregation: CTAB on graphite The first direct imaging of interfacial aggregation concerned the adsorption of CTAB on the cleavage plane of highly ordered pyrolytic graphite w17x. Graphite is a frequently used substrate for AFM investigations because it is available in

250

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 16. Adsorption of CTAB on graphite (a) an AFM image obtained in soft-contact mode using the double layer forces between the tip and sample. The adsorbed layer structure appears as stripes which are spaced 4.2"0.4 nm apart (about twice the length of the adsorbed surfactant). Image size 240=240 nm2, z-range 1.2 nm. (b) The proposed hemicylindrical structure of CTAB on graphite in cross-section. The base molecules (shaded) were probably strongly bound epitaxially by the graphite surface, while the rest of the monomers in the hemimicelle are more labile. 1994 ACS. Reproduced with permission from Ref. w17x.

atomically smooth crystalline form, making it ideal for AFM investigations, and graphon, a form of particulate graphite, is an often-used hydrophobic adsorbate. The graphite lattice consists of three equivalent symmetry axes and the interactions between graphite and surfactants are primarily hydrophobic. The images obtained (one of which is reproduced in Fig. 16a) showed parallel stripes spaced 4.2 nm apart for CTAB concentrations between 0.8 and 5 mM. The orientation was perpendicular to the symmetry axes of the substrate. As the period between the stripes was slightly greater than twice that of the extended monomer, the authors

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

251

suggested that the most likely surface conformation of the surfactant was a hemicylindrical arrangement, as reproduced schematically in Fig. 16b. Adsorption isotherm data acquired more than 30 years prior to the AFM investigation had revealed a two-step adsorption isotherm w64x. Without knowledge of the adsorbed structure it was proposed that the shape of the isotherm resulted from the monomer being adsorbed with its alkyl chain extended in the substrate plane at low concentration, and as the concentration increased the monomers interacted laterally to orient themselves perpendicularly to the substrate. This was thought to result in monolayer formation. However, with knowledge of the ultimate equilibrium structure, the new mechanism proposed by Manne et al. of hemimicelle formation has rightly gained widespread acceptance, this is described below w17x. 4.3. Graphite strongly orientates surfactant aggregates The first step in the isotherm results from surfactant monomer binding strongly to the surface via the hydrophobic interaction, forming a film that has a relatively low rate of exchange with solution surfactants. The adsorbed surfactants align themselves with the graphite symmetry axis, with the tail-groups oriented towards the interior of the hemimicelle. This first layer acts as a template for subsequent surfactant adsorption, as the solution concentration is increased. Aggregation is driven by hydrophobic interactions between the exposed surfactant tail-groups and the tail-groups of surfactants in the bulk. Graphite exhibits the highest degree of control over adsorbed aggregate structure of any substrate investigated to date, due to its large interaction area with the surfactant. Hemicylindrical aggregation has been observed on graphite for ionic w17,62,63,65,66x (conventional and gemini), nonionic w6769x and zwitterionic w70x surfactants on graphite with hydrocarbon tail-groups longer than 12 carbon atoms. Surfactants with tail-groups of 10 or fewer carbon atoms in length form a featureless monolayer on graphite. This is most likely due to the tail length failing to reach a critical length to successfully adsorb epitaxially and act as a template for hemicylindrical aggregation. 4.4. Adsorption studies on mica Considerably different aggregate morphologies have been determined for the crystalline mica substrate. As mica is hydrophilic, it is the head-group of surfactant monomers that interact strongly with the substrate and thus, relative to graphite, the area of interaction per surfactant molecule is greatly reduced. Mica has exchangeable surface cations, the presence of which can bring about higher levels of coverage than the less charged silica substrate. The density of the charges on the mica substrate (0.48 nm2) w61x is such that surfactants are adsorbed to the surface with their head-groups closer together than the equilibrium bulk separation (for comparison, the head-group area for CTAB in a micelle is 0.64 nm2) w26x. This leads to adsorbed aggregates having a lower degree of curvature than the corresponding solution micelles. The negative surface sites on mica are arranged precisely in the

252

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 17. Adsorption of CTAB on mica in the presence of 10 mM KBr. (a) AFM image in 1.8 mM CTAB. (b) Schematic representation of the cylindrical structure of adsorbed CTAB. Part of the cylinder on the right has been cut away to reveal the interior. The cylinders may be flattened on the bottom because of an attractive electrostatic interaction between the surfactant head-groups and the mica. 1999 ACS. Reproduced with permission from Ref. w72x.

surface lattice of the substrate, which is expected to influence aggregate morphologies. However, relative to graphite, the mica surface only weakly orients the axial directions of any adsorbed aggregates w47,61,62x. 4.4.1. Alkyltrimethylammonium halides on mica DTAB, MTAB and CTAB have been shown to form flattened cylindrical aggregates on mica. The period of these aggregates is similar to the diameter of a solution micelle, whilst the length of the aggregate was much larger w62,71,72x. The long axis of neighbouring aggregates was aligned locally in one of three directions, presumably due to the crystallinity of the substrate, giving a striped appearance. Interestingly, it was shown that for CTAB at twice the cmc the cylindrical structures observed 2 h after surfactant was passed into the fluid cell had been transformed into a bilayer after approximately 24 h. Addition of 10 mM KBr had little effect on the type of aggregate formed initially, but with added electrolyte the cylindrical structure was stable up to at least 30 h. An AFM image of adsorbed CTAB and a schematic representation of the structure of the aggregate have been reproduced in Fig. 17. 4.4.2. The influence of electrolyte on aggregate morphology Ducker and Wanless w72x investigated the effect of added electrolyte on aggregate shape. AFM imaging was used to demonstrate that the thermodynamically stable flat bilayer formed by CTAB was transformed to cylinders on the addition of electrolyte (KBr). As the electrolyte concentration was increased further, the long axis of the cylinders was shortened, resulting in the formation of discrete aggregates with no directional orientation. It was postulated that the addition of salt affected

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

253

Fig. 18. AFM images (150=150 nm2) of gemini surfactant aggregates on the cleavage plane of mica. (a) Spherical admicelles of the asymmetric gemini surfactant 12-3-1 (3.0 mM solution). (b) Cylindrical aggregates of the symmetric gemini surfactant 12-4-12 (2.2 mM solution, or 2= cmc); (c) Adsorbed bilayer of the symmetric gemini surfactant 12-2-12 (1.0 mM solution, or 1.3= cmc). 1997 ACS. Reproduced with permission from Ref. w63x.

aggregation in two ways. Firstly, once the co-ion concentration reached sufficiently high levels, it competes effectively with surfactant cations for surface charged sites. Additionally, the free counterion was electrostatically attracted to the head-groups of the surfactant, lowering the tendency of the surfactant to bind to the surface. This released sufficient surfactant from the surface to reduce, then eliminate, the templating effect. This explanation is supported by the fact that more defects are produced in the layer with increasing potassium ion concentration. Additionally, the addition of Hq, which is known to be more strongly surface binding than Kq, produces a greater number of defects at the same concentration. 4.4.3. Gemini surfactants on mica Gemini surfactants have proved particularly useful in the investigation of the relationship between surfactant geometry and substrate interaction on aggregate structure. Gemini surfactants offer the opportunity to systematically vary surfactant geometry by changing chain andyor spacer lengths, which allows the importance of surfactant geometry on aggregate shape to be investigated. Manne et al. w63x studied the adsorption of gemini surfactants of various geometries to graphite and mica with interesting results. On the strongly orientating graphite, where adsorption occurs via hydrophobic interactions, hemicylindrical aggregates formed regardless of the surfactant geometry. As with conventional surfactants, these structures were orientated perpendicular to the underlying symmetry of the substrate. However, on crystalline mica, the surfactant geometry was shown to have a stronger influence on the aggregate formed. This is shown eloquently in Fig. 18. The spherical structures formed by the asymmetric gemini 18-3-1 were also present for 16-3-1 and 12-3-1, although the spacing between aggregates decreased with chain length. Cylinders formed in the presence of 12-4-12 and 12-6-12.

254

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

However, a reduction in the spacer length to 2 resulted in the formation of a featureless bilayer. These results show the same general trend as solution aggregates (where the curvature of aggregates is lowered as the spacer length is decreased), and demonstrate the dependence of structure morphology on geometry, even when the surface exerts a templating effect. 4.5. Adsorption studies on silica In contrast to the study of adsorption isotherms relatively few AFM studies have employed amorphous silica, perhaps primarily because it is more difficult to obtain clear images. In general, straight chain cationic surfactants have been shown to form essentially spherical admicelles with no long range ordering on the silica substrate w62,7375x. This morphology agrees well with the mechanism predicted by adsorption isotherms: that electrostatically adsorbed monomers act as nucleation sites for further surfactant adsorption. As with other substrates, factors that affect aggregation in solution often have consequences for structures at the silicawater interface. 4.5.1. The influence of electrolyte and counterion type The influence of surfactant and electrolyte concentration and type of counterion on the adsorbed morphology was investigated by Velegol et al. w73x. For CTAB, when the surfactant concentration was increased from 0.9= cmc to 10= cmc the adsorbed morphology changed from short rods to worm-like. This occurred regardless of whether 10 mM KBr was present or absent, although added electrolyte did reduce the aggregate period. When the counterion was changed from bromide to chloride, the aggregate morphology was predominately spheroidal for both concentrations, and once again the presence of electrolyte did not significantly change the aggregate structure. These results are reproduced in Fig. 19. Fig. 19 clearly shows the dependence of adsorbed layer morphology on the type of counterion. These results were rationalised by the greater binding efficiency of bromide over chloride allowing bromide to better stabilise the lower curvature cylindrical structures. Using OR, it was shown that the change in structure from spheroids to worm-like (for CTAC and CTAB, respectively) corresponded to the surface excess increasing by a factor of two thirds. 4.5.2. Adsorption kinetics measured by AFM In a recent study of the coadsorption of surfactants and polymers, Liu et al. w74x made an interesting observation concerning the adsorption and desorption kinetics of CTAC. Although the kinetics of adsorption were not measured directly, force curves were analysed to give some indication of the adsorption rate. Immediately after surfactant solution was passed into the AFM cell a steep repulsive force and adhesion was evident. This indicated the presence of bilayered surfactant aggregates, which were successfully imaged. When the cell was rinsed with water, the repulsive barrier disappeared, and was replaced by an attractive force and larger adhesive force. The length of time that these forces persisted was found to be dependent on

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

255

Fig. 19. AFM images of a range of cationic surfactants on silica: (a) CTAB 10= cmc. (b) CTAB 0.9= cmc. (c) CTAC 10= cmc (d) CTAC 0.9= cmc, with push through the adsorbed layer to the underlying silica substrate shown in the bottom of the image. The peak-to-peak distance between CTAC aggregates with and without 10 mM kBr is 10"2 nm at 10= cmc and 13"2 nm at 0.9= cmc. For CTAB, the peak to peak distances are 10"1 nm without salt and 8"1 nm with 10 mM KBr at 10= cmc. 2000 ACS. Reproduced with permission from Ref. w73x.

how long the CTAC solution was exposed to the substrate. The general features of this effect are shown in Fig. 20. These results suggested that the molecules that serve to reverse the surface charge are easy to desorb; whereas those that are electrostatically bound to the surface are retained by the surface much longer upon rinsing. The time dependence of this process was justified using a model proposed by Chen et al. w76x. This adsorption model proposed that the surfactant ion is initially adsorbed with a counterion bound, and that this counterion is expelled from the film over time, allowing the surfactant to be electrostatically bound to the surface. The electrostatic nature of this adsorption serves to extend the desorption process. Desorption was found to proceed more

256

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 20. Schematic representation of the desorption of CTAC from silica, showing the dependence on the length of exposure of the surfactant to the surface. 2001 ACS. Reproduced with permission from Ref. w74x.

slowly at higher pH values. As increasing the pH should increase the level of electrostatic binding, this result supports the proposed model. 4.5.3. The influence of counterion polarisability Subramanian and Ducker w75x investigated the effect of counterion on the adsorbed structure of CTAq on the basis of the hardness of the ion. It was found that soft, polarisable (e.g. Bry) ions were more effective than hard counterions (e.g. CH3COOy, Cly) at inducing shape changes in admicelles, namely a sphere-tocylinder transition. It was suggested that as hard anions strongly bind water they are relatively unavailable for binding to surfactant ions. Soft counterions, which weakly interact with water, associate more readily with surfactant. Binding of counterions lowers the repulsive force between head-groups, permitting the formation of the less curved cylindrical aggregates. This study reported that surface micelles were clearly observed at 0.5= cmc in the absence of electrolyte. This is because electrostatic adsorption of surfactant to the substrate lowers the activation energy of aggregate formation and permits the formation of admicelles below the solution cmc. 4.5.4. Gemini surfactant aggregates on silica The adsorbed morphology of gemini surfactants at the silicaaqueous interface has been investigated by Atkin et al. w77x. An image of 12-2-12 at approximately 2= cmc is presented in Fig. 21 revealing flattened ellipsoidal structures. Similar structures were imaged for 12-3-12. The adsorption density was found to be approximately 150 aggregates per 10 000 nm2 and the aggregate thickness is

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

257

Fig. 21. An AFM image of the silicasolution interface immersed in an aqueous solution of 2= cmc 12-2-12 revealing adsorbed aggregates with approximately circular profiles in the upper portion of the image. The slow scan direction is down the page. The underlying silica surface is imaged in the lower portion of the page when a higher imaging force is applied. 2003 ACS. Reproduced with permission from Ref. w77x.

estimated to be 3.5 nm, from the push through distance obtained in the force curve. As these surfactants have been shown to form worm-like aggregates in solution, the smaller surface aggregates must be attributed to the influence of the substrate. The physical dimensions and spacing of the admicellar structures presented here are similar to those observed for monomeric quaternary ammonium surfactants on silica (Section 4.5.1). In this study we reported that obtaining clear images of 12-3-12 was considerably more difficult than for 12-2-12. This was attributed to the decreased adsorption density of 12-3-12. The number of adsorbed molecules of 12-3-12 was 25% lower than that of 12-2-12. As the spacer length increased, the adsorption density continued to fall as per Fig. 12. For s)3, AFM images of the adsorbed surfactant layer could not be obtained. This was a consequence of the change in the nature of the interaction force between the AFM tip and the surface, resulting from proximal desorption of surfactant. Proximal adsorption or desorption describes a change in surface excess as a function of surface separation. Ducker et al. w78,79x recently demonstrated that the magnitude of proximal desorption was significant for cationic surfactants adsorbed

258

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

to silica surfaces. At concentrations where the surface charge is reversed due to surfactant adsorption, surfactant monomers desorb from the surfaces upon approach in order to lower the electrostatic repulsion. Subramanian and Ducker has shown that, for CTAB concentrations slightly higher than the cmc, the proximal desorption increases dramatically at surface separations of less than 10 nm w78x. At approximately 9 nm the change in adsorption is ;0.25 molecule nmy2. If the change in adsorption was similar for gemini surfactants then this represents a significant proportion of the total surface excess, up to 50% for the larger spacer groups (as there is a significant decrease in surface excess with increasing spacer length, cf. Fig. 12 and Section 6.3.12). Thus, fewer surfactant monomers are required to desorb from the substrate for the longer spacer sizes in order to expose a hydrophobic surface. Additionally, the electrostatic repulsion is greater for the longer spacer lengths, therefore the driving force for desorption is likely to be greater. Both of these influences strongly suggest that proximal desorption during imaging will have a greater effect on the adsorbed structures as the spacer length increases. This proximal desorption process is depicted schematically in Fig. 22. It is expected that solution-facing monomers will be more likely to desorb from the interface as the tip approaches. As a consequence, a portion of the tail-groups of the monomers adsorbed with head-groups facing the silica substrate will be exposed. This will produce hydrophobic surfaces and consequently the surfaces jump together under the influence of the long-range hydrophobic attraction, which precludes imaging of surface structures. As similar force vs. distance data were obtained for spacer lengths between 4 and 12, a similar process is likely occurring for all of these spacer lengths. As images of the adsorbed structures for spacers of length greater than 3 could not be obtained, the likely structures of the adsorbed aggregates were inferred from the images of the 12-2-12 aggregates and the linear increase in the area per adsorbed molecule w77x with spacer size. It is known that the same surface excess in terms of number of monomers is required to neutralise the surface charge for the gemini surfactants (cf. Section 3.4.6). Further it is generally accepted that surface aggregates form around electrostatically adsorbed monomers (cf. Section 3). Therefore approximately the same number of surface aggregates should be formed for 12-2-12 up to 12-12-12. This was supported by AFM imaging of the 12-2-12 and 12-3-12 gemini surfactants, which reveal similar numbers of aggregates per unit area despite a 25% difference in surface excess. As the same number of aggregates were present per unit area, and the surface area occupied by the aggregates did not change with spacer length, by assuming the mass density of the aggregates does not change, the aggregate volumes and shapes for s)3 were determined. It was shown that the adsorbed aggregates become more flattened as the spacer length is increased. Schematic representations of the variation in adsorbed structure with spacer length are shown in Fig. 23. Clearly as the head-group area increases (larger s), the surfactant chains interdigitate to a greater degree in order to maximise chainchain interactions. This leads to the flattening of the surface aggregates. These flattened aggregates appear to be less than optimal in terms of surfactant packing as they are very easily disrupted.

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

259

Fig. 22. Schematic representation of the release of gemini surfactant molecules from an adsorbed aggregate upon approach of the AFM tip apex (represented by a large filled triangle) for spacer lengths of four or more. The surfactant concentration is sufficient to lead to surface aggregation. No attempt has been made to represent counterions or surfactants that are not adsorbed either at the silica substrate or AFM tip in the first instance, and for simplicity all surfactants are represented in the cis conformation with straight hydrocarbon chains. In (a) the tip is effectively far from the interface, at a separation of ;100 nm. The AFM senses an electrostatic repulsion, but this repulsion is not sufficient to induce release of surfactant from the adsorbed aggregate. (b) Depicts the situation as the AFM tip continues to approach the interface. At these smaller separations the magnitude of the electrostatic interaction is sufficient to force surfactant monomers out of the admicelle. However, the overall electrostatic force is greater than the hydrophobic attraction at these separations. (c) Represents the situation immediately prior to the jump into contact. The outer layer of the surfactant aggregates on both the AFM tip and the substrate have largely diffused into solution, resulting in the residual electrostatic repulsion being considerably reduced relative to the initial situation, and a significant amount of hydrophobic material being exposed. The hydrophobic attraction between the surfactant adsorbed to the AFM tip and the substrate leads to the jump to contact observed in the force curve, the magnitude of which is indicated by the arrow. (d) Shows the expected configuration of surfactant adsorbed to the tip and the substrate in the constant compliance region. Depending on the stiffness of the adsorbed layer, the AFM tip may push through to contact directly with the silica substrate. The situation depicted also leads to the significant adhesion observed for all spacer sizes greater than 3 as the AFM tip is retracted from the substrate. 2003 ACS. Reproduced with permission from Ref. w77x.

4.6. Model hydrophobic substrates Model substrates can be used to study the influence of factors such as surface charge, roughness, crystallinity and hydrophobicity on the adsorption process. Grant et al. w80x systematically investigated the influence of substrate hydrophobicity on the adsorption of a nonionic surfactant, namely octa(oxyethylene) ndodecyl ether (C12E8). In the absence of a charged head-group, the driving force for the adsorption of nonionic surfactants is derived from hydrophobic and van der Waals attractions. In this study, altering the ratio of chemisorbed hexadecane thiol and thiohexadecanol to a gold substrate controlled the substrate hydrophobicity. As

260

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 23. Schematic representation of the variation in the adsorbed gemini surfactant layer as the spacer sizes increased. The diagrams presented are two-dimensional cross sections of three-dimensional aggregates present at the interface. Counterions are not represented. The lateral dimensions of the aggregate are similar at all spacer sizes, and it is the increased size of the surfactant head-group as the spacer length increases that results in reduced aggregation numbers, a lower surface excess and a reduced aggregate thickness. As the adsorbed aggregates cannot be imaged for spacer sizes of four or more, the structures depicted are proposed on the basis of area per molecule and force data. 2003 ACS. Reproduced with permission from Ref. w77x.

the relative amount of thiohexadecane was increased, so was the substrate hydrophobicity. This allowed the preparation of five increasingly hydrophobic surfaces. The contact angle varied almost linearly with the proportion of chemisorbed hexadecane thiol, from 258 at 0%, to 1108 at 100% hexadecane thiol (the three other surfaces prepared were 25, 50 and 75% hexadecane thiol). Force curves and images were collected at a solution surfactant concentration of 2= cmc. Analysis of AFM images and force curves allowed the variation of substrate hydrophobicity on the structure of the adsorbed layer to be commented upon. The general results are summarised in Fig. 24 w80x. On the most hydrophilic surfaces (0 and 25% CH3 terminated) the force curves showed that the surfactant was easily displaced from the surface. This indicated that the interactions between the surfactant and the substrate are relatively weak. On these surfaces, there were many sites at which water could be hydrogen bonded, allowing water to effectively compete for the surface. Imaging of the adsorbed layer indicated diffuse micellar coverage, with an aggregate period of approximately 10 nm. Adsorption was attributed to hydrophobic interactions and a comparatively weak van der Waals contribution, shown in Fig. 24a. On the 50% methyl terminated surface, the adsorbed aggregate period was decreased to 6 nm. A more stable repulsive barrier was present in the force curve, indicating a layer thickness of 6 nm. This value compared well to the solution micellar diameter of 6.2 nm. Increasing the substrate hydrophobicity had increased the strength of the hydrophobic interactions between the surface and the surfactant. The morphology of the surfactants was found to be close packed micellar aggregates

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

261

Fig. 24. Schematic representation of the influence of the variation in the percentage of CH3 terminated alkyl chains on C12E8 adsorption. The remainder of the surface is composed of CH2OH terminated groups. 2000 ACS. Reproduced with permission from Ref. w80x.

(Fig. 24b). When the degree of methyl group termination was increased to 75% force imaging could reveal no discrete aggregation. However, once again the force curve indicated that the adsorbed layer was 6 nm thick. In this case the adsorbed layer morphology was interpreted to be a classical bilayer, represented in Fig. 24c. When the degree of alkyl termination was increased to 100%, a featureless layer was also imaged. However, in this case the layer thickness interpreted from the force curve was approximately 4 nm. This indicates the presence of a surfactant monolayer on the surface, see Fig. 24d. These results also serve to demonstrate the importance of interfacially adsorbed water. The degree of water hydrogen bonded to the surface increases as the level of alkyl termination is reduced from 100 to 0%. Where water is not bound at the interface in great amounts, as on the 100 and 75% alkyl terminated surfaces, monoand bi-layered aggregation results. The 50% alkyl terminated surface provides a greater quantity of sites for water binding, and discrete aggregates are observed. One could interpret this result on the basis of the adsorbed water preventing the bilayered aggregation, leading to the formation of curved aggregates. The 0 and 25% alkyl terminated surfaces were even more hydrophilic, which allowed water to compete more strongly for surface area. This led to the aggregates being spaced further apart and suggested a lower surface excess. Thus, increasing the level of substrate hydrophobicity not only increases the attraction between the surface and the surfactant, it also lowers the level of water interacting with the surface. Both of these factors have implications for structure of the adsorbed surfactant.

262

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Several other AFM studies have used modified hydrophobic substrates and have generally reported adsorption of single layered aggregates with adsorption dominated by substrate tail-group interactions. Briefly, Grant and Ducker w81x have investigated nonionic surfactant aggregation at an amorphous hydrophobic surface, prepared by covalently attaching diethyloctylchlorosilane (DEOS) to a silica substrate. The repulsive forces observed, together with AFM imaging of the adsorbed surfactant, were deemed to be consistent with the formation of a uniform monolayer at the surface, with head-groups facing solution and the surfactant tail-groups and DEOS interacting hydrophobically. Wolgemuth et al. w82x studied the adsorption of various ionic, nonionic and zwitterionic surfactants to an amorphous silica surface, hydrophobised by the covalent attachment of trimethylchlorosilane (TMCS). It was found that the interfacial aggregates formed were roughly hemispherical, in contrast to the half-cylindrical structures formed on crystalline hydrophobic substrates. Similar surfactant morphologies were observed regardless of the type of head-group, leading the authors to suggest that adsorption is driven by hydrophobic interactions between the substrate and the surfactant tail-groups. SDS adsorption was not noted consistently on the TMCS treated substrate, which was attributed to a residual negative charge associated with the silica after reaction with TMCS. 4.7. Summary of AFM investigations AFM has provided direct evidence for the presence of discrete aggregates at the solidaqueous interface and has led to a slow revolution in our interpretation of surfactant adsorption data from all sources. AFM imaging is only useful above the csac, where surfactant head-groups are facing towards solution, imparting a repulsion (electrostatic for ionic surfactants) that allows soft contact imaging to be accomplished. The clarity of the AFM images obtained varies with the surfactant and the substrate under investigation. As the AFM is most useful at detecting periodicity, surfactantsubstrate combinations that produce highly regular morphologies produce the clearest AFM images. The hydrophobic cleavage plane of graphite orients the adsorbed surfactant structures more strongly than any other substrate. This is primarily due to a high level of (hydrophobic) interaction with initially adsorbing monomers, which templates the subsequently formed hemicylindrical structures for surfactant chains with greater than 10 carbon atoms. Chains of less than 10 carbon atoms are not templated by the substrate and result in a laterally featureless adsorbed layer. The hydrophilic, crystalline mica substrate also orients the adsorbed surfactant structures, but not as strongly as graphite. The alkyltrimethylammonium surfactants form cylindrical aggregates, which were in some cases transformed into a featureless layer. However, the addition of electrolyte stabilises the cylindrical admicelles. The adsorbed morphology of gemini surfactants on mica can be manipulated by altering the geometry of the surfactant monomer. Surfactants are less strongly oriented on silica surfaces, due to a lower surface charge and the amorphous state of the substrate and perhaps also due to an increase in surface roughness. The adsorbed structure of CTAB and CTAC on silica has

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

263

been investigated. The adsorbed configuration of CTAB was concentration dependent, with short rods at 0.9= cmc and worms at 10= cmc. These structures were insensitive to the presence or absence of electrolyte. For CTAC, the adsorbed morphology was insensitive to both the surfactant concentration and the presence or absence of electrolyte, with spheroidal structures present at both 0.9= cmc and 10= cmc in the presence and absence of electrolyte. In all cases there was little or no long range ordering of the adsorbed aggregates. Similar structures have been observed for MTAB and DTAB on silica. By altering the substrate hydrophobicity the adsorbed morphology of nonionic surfactants can be controlled. By sequentially increasing the substrate hydrophobicity the structure of adsorbed nonionic surfactants can be changed in a controlled manner from diffuse adsorbed micelles on the least hydrophobic surface, to densely packed micelles, to a bilayer, to a monolayer on the most hydrophobic interface. It is clear that adsorbed aggregate structures are often formed upon surfactant adsorption at the solidaqueous interface, and that the properties of the substrate may exert a high degree of influence on the type of aggregate formed. These structures can be reconciled with the adsorption isotherm data presented in Section 3, and this will be discussed in Section 7. The intervening discussion is aimed at more completely characterising the structure of the adsorbed layer. 5. Fluorescence quenching experiments 5.1. Introduction Fluorescence quenching studies statistically analyse the fluorescence emission of a micelle bound probe molecule. A second molecule, known as a quencher, suppresses this emission. Fluorescence quenching is a well accepted method for the determination of micelle aggregation numbers in bulk solution w83,84x. The fluorophore probe, typically pyrene, is excited at a particular wavelength and the fluorescence intensity monitored at a second wavelength. All fluorescence quenching studies assume that both the probe and quencher molecules are entirely contained within micelles, that there is no exchange between aggregates, and that the micelles are unperturbed by the presence of the probe and quencher. A random distribution of the fluorophore and the quencher throughout the system under investigation is assumed. Therefore, some aggregates will contain both the probe and the quencher, some contain either the probe or the quencher, and some aggregates will contain neither. Statistical analysis will allow the average number of quencher molecules per micelle in the system to be calculated. As only micelles containing the fluorophore are probed, the aggregation number of the micelles can be determined from the intensities of the fluorescence emission in the absence and presence of a known concentration of quencher. This relatively simple analysis is known as the static emission method and for this method to be successful it is essential that quenching is rapid. It is assumed that there is no fluorescence from micelles containing quencher, thus any emission from these micelles alters the aggregation number determined. This effect is expected to be more pronounced for

264

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 25. Representative fluorescence decay curves for pyrene fluorescence from admicelles on silica. Two curves are shown for each surfactant: one with no quencher present and one in the presence of quencher. The concentration of surfactant is as follows TTAB (Cs1.01 mM) and CTAB (Cs0.801 mM). The amount of pyrene in each sample is approximately 1 pyrene per 50 surface-bound micelles. 2000 ACS. Reproduced with permission from Ref. w85x.

large aggregates, where the rate of quenching will be dependent on the diffusion rate of the probe and quencher throughout the micelle core. 5.2. Time resolved fluorescence quenching The time resolved fluorescence quenching experiment is not as dependent on the assumption of rapid quenching. This experiment uses a delta pulse excitation of the probe molecules and the decay of the fluorescence emission with time is recorded. As the rate of quenching is dependent on molecular collisions between the probe and the quencher molecules, this gives rise to a fast initial decay, followed by a slower unquenched decay. Examples of the form of these curves are presented in Fig. 25, reproduced from Strom et al. w85x. These curves were obtained from admicelles on a silica surface. For time resolved fluorescence quenching experiments, the aggregation number can be determined from the curved portion of these data, or by extrapolation of the long time data back to ts0. This gives the intensity at time 0 in the presence and absence of quencher molecules, and allows the aggregation number to be determined in the same manner as for the static emission method.

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

265

5.3. Determination of aggregation numbers In order to determine aggregation numbers, the concentration of aggregate bound surfactant must be known. In solution, this can be determined from the cmc. When studying the properties of aggregates adsorbed at an interface, the surface concentration must be determined by other methods, and it is assumed that all surfactant is aggregate bound. If aggregates are present in solution and at the interface, the probe molecule will be distributed through both solution and surface aggregates, leading to results that are inaccurate. However, if the surfactant concentration is chosen such that it is below the solution cmc, but above the concentration at which aggregates form on the surface (csac), then the probe molecule will be fully surface bound. This allows the adsorbed aggregates to be probed directly. 5.4. The earliest fluorescence probe studies The first fluorescence probe studies of adsorbed aggregates were performed by Levitz et al. w86,87x, using the static emission method. The adsorption of nonionic surfactants such as Triton X-100, were investigated. It was shown that the aggregation numbers for adsorbed surfactant determined by fluorescence emission followed similar trends to bulk micelles. It was thus concluded that the most likely form of surface aggregate was that of a small surface admicelle. Fan et al. w19x performed similar studies for ionic surfactant adsorption. The aggregation of alkyltrimethylammonium bromides on an alumina substrate was investigated and the presence of small surface bound aggregates was also determined. The size of these aggregates appeared to grow as the surface excess of surfactant increased. This result is shown in Fig. 26. These results agree well with the adsorption model for ionic surfactants to an oppositely charged substrate suggested by the adsorption isotherm data described in Section 3. That is, electrostatically adsorbed monomers act as nucleation sites for further surfactant adsorption, leading to the formation of surface aggregates. This accounts for the observed increase in aggregation number. These data also show that fluorescence probe quenching gives insight into the structure of the adsorbed layer at surface excess values below that where aggregates are fully formed, and are thus not readily investigated using AFM. 5.5. Time resolved fluorescence quenching of adsorbed aggregates The first time resolved fluorescence quenching study of adsorbed aggregates was performed by Strom et al. w85x. The adsorption of DTAB, TTAB, CTAB and dodecyl-1,3-propylene-pentamethyl-bis(ammonium chloride) or DoPPDAC, a divalent cationic surfactant, was studied. The quencher molecules used in this study were alkyl pyridinium surfactants. This choice was deemed suitable in light of the structural similarity between the quaternary ammonium surfactants under study and the alkyl pyridinium quencher, which minimised the degree to which the system was perturbed. Emission data typical of that obtained by this study are shown in

266

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 26. Aggregation numbers for adsorbed surfactant aggregates on alumina formed using alkyltrimethylammonium bromide surfactants of three different chain lengths. The aggregation numbers corresponding to a given surface excess are given on the figure for the steeply increasing region of the isotherms. 1997 ACS. Reproduced with permission from Ref. w19x.

Fig. 25. In particular, note the initial curved decay, which results from micelles containing both quencher and probe molecules, and note that the exponential portion of the curve, due to micelles containing probe only, has the same slope as the result recorded without quencher present. These features provide strong evidence for the formation of discrete surface aggregates, which was the principal finding of this study. This result agrees well with AFM imaging results, and cannot be rationalised simply in terms of mono- and bi-layer aggregation. The evolution of aggregation number with surfactant concentration and with increasing electrolyte concentration is reproduced in Fig. 27. It was judged that the uncertainty in these values was at most 20%. The increase in aggregation number with surface excess observed in Fig. 27a supports the trend observed by Fan et al. for the adsorption of alkyltrimethylammonium bromides on alumina w19x. Increasing the electrolyte concentration was also shown to increase the aggregation number. This was rationalised on electrostatic grounds, with added electrolyte decreasing the electrostatic repulsions between monomers within adsorbed aggregates. This facilitates tighter packing within admicelles, leading to the observed increase in aggregation number. 5.6. Summary of fluorescence quenching experiments For the purposes of this review, fluorescence quenching experiments provide two important results. Firstly, fluorescence quenching provides strong evidence for the

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

267

Fig. 27. Aggregation numbers of micelles on silica for DTAB (circles), TTAB (squares), DoPPAC (triangles), and CTAB (cross). Aggregation numbers are presented as a function of (a) bulk concentration and (b) as a function of added electrolyte. 2000 ACS. Reproduced with permission from Ref. w85x.

presence of discrete aggregates. This supports the results obtained using AFM, that discrete admicellar structures are present at the substrate as summarised in Section 4. Perhaps more importantly, several fluorescence quenching studies have shown that the aggregation number of the adsorbed structures increases with surface excess.

268

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

This result provides the strongest evidence for the nucleation mechanism of surfactant adsorption, described previously in Section 3. 6. Reflectance techniques 6.1. Introduction and underlying principles The principle of reflectivity is essentially the same with respect to scattering, regardless of whether light, X-rays or neutrons are used as the radiation source. Whereas X-rays are sensitive to the electron density, and hence the atomic number of the species under investigation, the scattering of neutrons is dependant on the nuclear properties of the molecules being probed. Optical techniques such as ellipsometry and optical reflectometry (OR) derive their sensitivity from the dependence of reflection and refraction on refractive index and interference effects that give phase information. These effects allow calculation of the adsorbed layer thickness or surface excess. The significantly shorter wavelengths associated with X-rays and neutrons allow the structure of the adsorbed layer to be examined more intimately. 6.2. Neutron reflectivity and adsorbed layer structure NR provides sufficient resolution to study the orientation of surfactant molecules in the adsorbed layer, as different nuclei scatter neutrons with different amplitudes. In the case of protons and deuterons, neutrons are scattered with opposite phases, which allows contrast variation to be achieved by use of a combination of protonated and deuterated adsorbates. This type of substitution alters the reflectivity of the adsorbed layer, but leaves the surfactant morphology unaltered. Additionally, by careful adjustment of the proton to deuterium ratio of the solvent, the contrast between the solvent and the surface can be set to 0, resulting in the recorded reflectivity profile being dependent only on the adsorbed layer. The resolution, or the smallest length scale able to be probed, is a function of the wavelength. The short wavelength of neutrons gives increased resolution over that of optical techniques. This, combined with the contrast control available using isotopic substitution, allows NR to elucidate concentration profiles close to the surface, and even the orientation of the adsorbed species. 6.2.1. Limitation of NR The major limitation of NR is that interpretation of the results is model dependent. Conventionally, the adsorbed layer is treated as a laterally unstructured film w88x. This allows the interference fringe that results from the reflections at the solid layer and layersolvent interfaces to be characterised by a thickness and a scattering length density profile w89x. In practice, the adsorbed film is often more complicated than a homogenous layer, and the only satisfactory method for determination of the perpendicular composition profile is to divide the adsorbed film into a series of layers with separate density profiles using the optical matrix formulation of

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

269

reflectivities. Each layer is characterised by its scattering length density, thickness, and if necessary an interfacial roughness factor. The division of layers generally reflects the different scattering length densities of the surfactant head-groups and alkyl chains. 6.2.2. Contrast control The additional contrast achieved by isotopic substitution provides information that aids in selecting the most appropriate model. In spite of this detail, a determined profile is not necessarily a unique solution i.e. several models of the adsorbed layer structure may fit the experimental result. For the most part, neutron reflection results for surfactant layers have been interpreted as a monolayer on hydrophobic solids and bilayered or a patchy bilayer morphology on hydrophilic substrates. Whilst these results are somewhat surprising in light of AFM imaging, if one takes into account the limitations of the technique, useful information can be derived from these studies. Neutron reflection gives the most direct measurement of the thickness of adsorbed aggregates, and provides the orientation of molecules perpendicular to the surface. However, it provides no data concerning the lateral ordering of molecules, and hence one cannot make definitive statements concerning the inplane dimensions of aggregates w90x. 6.2.3. NR studies of cationic surfactants on silica Several NR studies have examined the adsorption of CTAB to silica w8991x. The consensus of these studies is that the surface is incompletely covered. This is rationalised by either a defective bilayer, or by the adsorption of discrete micellelike aggregates. The AFM and fluorescence data presented above indicate that the admicelle morphology is the correct interpretation. McDermott et al. w92x showed that whilst the fraction of the surface covered increased from 51% at 1y3= cmc to 65% at 1= cmc, the layer thickness of 3.4 nm was consistent at both concentrations. From this result the authors suggested that the increase in the level of surface coverage was unlikely to be derived from a change in the aggregate structure. Due to the incomplete coverage of the surface it was postulated that the water and surfactant occupy separate domains on the surface, with islands of surfactants suggested. It was noted that these islands would most likely have head-groups protruding from the sides in order to shield the alkyl chains from the water intermittent to the islands. It was shown that when the dimensions of these islands were sufficiently small the head-groups at the sides of the aggregates would become important to the fitting of a consistent optical model, in that both head-groups and chains would be present in the midplane of the aggregates. Secondly, inclusion of the intermittent water in the optical model suggested a lateral size of 9 nm at the higher surfactant concentration. This result is consistent with the dimensions observed using AFM in Section 4.5.1. Further, the fluorescence quenching studies show that the aggregation number increases with surface excess. This result is inconsistent with the above interpretation. These results were clarified by Fragneto et al. w90x, who also concluded that if the adsorbed aggregates were micelle-like, then they must be strongly flattened.

270

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

This result has been corroborated by several authors using various techniques, and has now become well accepted. This study also investigated the effect of surface roughness on adsorption. As does surface preparation, surface roughness could contribute to the variation in results for surfactant adsorption on silica substrates. The authors reached several important conclusions. Firstly, the level of surfactant adsorption decreased as the surface roughness increased. This result may seem counter-intuitive, as increasing roughness should increase the available surface area. It was suggested that this effect could be derived from an inability of the rougher surface to match the curvature of the adsorbed aggregate. Additionally, the thickness of the surfactant aggregates was increased on the rougher surface. This could be attributed to reduced electrostatic interactions between the substrate and the aggregate leading to a greater aggregate curvature. Alternatively the surface roughness could add to the thickness of the layer by varying the position of the centre of the aggregate relative to the surface. It has been consistently noted that adsorption studies using two or more experimental techniques have a far greater probability of producing unequivocal results. This is also the case with NR. On the basis of the numerous discrete adsorbed aggregates determined by AFM imaging, Schulz et al. w93,94x have recently produced reflectivity models using unit cells consisting of spheres and cylinders. The authors point out that adsorbed films consisting of discrete aggregates will naturally produce a level of surface coverage that is less than 100%. This is an important observation, as the patchy bilayer or island interpretations suggested by many NR studies are simply corrections made to account for fractional surface coverage, with little supporting evidence and can be seen as originating from outdated interpretations of adsorption isotherms where surface excess values considerably below that required for bilayer adsorption were interpreted as patchy bilayers. The most recent investigation of Schulz et al. combined AFM and NR studies of surfactant adsorption onto crystalline quartz and showed that NR can correctly distinguish between adsorbed morphologies by using bulk solution contrast variation w94x. Three surfactant-quartz combinations were studied, and the structure of the adsorbed layer was determined using AFM imaging. At concentrations above the solution cmc, TTAB was shown to form spherical admicelles, TTAB with 200 mM NaBr formed cylindrical structures and the double chained surfactant DDAB produced a laterally unstructured film. These morphologies are consistent with the solution structures, where spheres, cylinders and bilayers have been determined for TTAB, TTAB and 200 mM NaBr, and DDAB, respectively. The authors produced theoretical scattering models for spherical, cylindrical and bilayer morphologies for each surfactant. The best fit parameters used for each of the models is reproduced in Table 2. While each model had a unique scattering length density profile, it was not possible to determine the structure of the adsorbed layer solely on the basis of goodness of fit, as all structures seemed equally likely for each surfactant. However, the structure of the adsorbed surfactant layer can be ascertained by comparing surface excess values obtained in pure D2O, and in a system where the

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Table 2 Fits to adsorbed layer structure for cationic surfactant systems on quartz from neutron reflectometry NR in D2O G (mmol my2) TTAB Spheres Cylinders Bilayer Spheres Cylinders Bilayer Spheres Cylinders Bilayer 6.6"0.6 6.1"0.6 5.4"0.5 7.3"0.3 7.0"0.4 6.3"0.4 4.7"0.3 4.5"0.3 4.2"0.2 d (nm) 5.3"0.3 5.1"0.5 5.2"0.2 5.5"0.3 3.0"0.2 2.5"0.2 t (nm) 4.6"0.4 3.6"0.4 2.7"0.3 4.7"0.1 4.0"0.2 3.0"0.2 3.2"0.2 2.6"0.2 2.1"0.2 NR in contrast matched D2OyH2O G (mmol my2) 6.5"0.4 6.4"0.3 6.4"0.3 6.9"0.3 6.9"0.3 6.8"0.3 4.9"0.2 4.8"0.2 4.8"0.2 d (nm) 5.2"0.3 6.0"0.3 4.7"0.2 5.2"0.2 3.5"0.2 3.3"0.1 t (nm) 4.6"0.3 4.0"0.2 3.3"0.1 4.4"0.2 3.9"0.2 3.2"0.1 3.6"0.2 3.1"0.1 2.6"0.1 AFM d (nm) 6.9"0.5

TTABq200 mM NaBr

4.9"0.5

DDAB

Consistent fits to NR and AFM images for each system are set in bold, showing agreement between adsorbed amounts in the D2O and quartz contrastmatched D2OyH2O measurements. Also listed are the film thicknesses t and nearest-neighbour spacings d, from NR fitting and AFM images, showing agreement between the two techniques for the best fit NR case. Redrawn from Ref. w94x.

271

272

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

solution is contrast matched to the quartz substrate. This is achieved by varying the ratio of H2O to D2O in the bulk. For the correct model, the surface excess (and the fitted parameters) must be equivalent at both contrasts. As can be seen in Table 2, the surface excess values (G) are indeed independent of the model used when the quartz and the bulk solution are contrast matched. In the unmatched system, the surface excess values show considerable variation. The interfacial model which best fits these data, between these limits in contrast, is shown in bold in Table 2. The results obtained agree well with those obtained with AFM imaging. This led the authors to suggest that this method of adsorbed layer structure determination may be of particular use when the necessary conditions for soft contact AFM images cannot be obtained. This elegant work indicates that AFM and NR data are compatible provided an appropriate model is employed to interpret the NR data. 6.3. Ellipsometry, optical reflectometry and adsorption kinetics The evolution of the adsorbed layer structure with concentration is often commented upon when studying surfactant adsorption using equilibrium techniques. These studies imply a series of processes that occur as the interfacial concentration increases. As we have seen previously for ionic surfactants, the general consensus is that surfactant adsorption is by electrostatic means at low concentrations, and as the surface excess increases lateral hydrophobic interactions lead to induction of surface charge and hemimicelle formation. This is followed by the formation of aggregates at higher surface excess values. This interpretation is quite well established and accepted. However, whether these processes occur kinetically is open to question. That is, for a concentration where aggregates result, is the surface charge first neutralised, followed by hemimicelle formation, then by bilayered aggregate formation, or do these processes occur simultaneously? Additionally, the contribution or otherwise of micelles to the adsorption process is often questioned. Investigations of surfactant adsorption kinetics should elucidate these issues. 6.3.1. Dynamic aspects of surfactant adsorption In contrast to equilibrium adsorption characteristics, the dynamic aspects of surfactant adsorption have been far less studied and are consequently not as well understood. This is in spite of the importance of adsorption kinetics to many realworld applications such as wetting, lubrication and spreading. The difficulty in monitoring the expected fast kinetics of the adsorption process is the primary reason for the lack of investigation of adsorption rates. However, the recent development of ellipsometric techniques with high temporal resolution has made it possible to study the kinetics of adsorption and desorption with adequate precision. 6.3.2. Principles of optical techniques OR and ellipsometry both monitor the variation in the reflectivity of an interface upon adsorption. These variations are induced by the change in the refractive index profile of the substrate upon the adsorption of a surfactant layer. Both techniques use a linearly polarised light source that is reflected off the substrate to which

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

273

adsorption takes place. The polarisation characteristics of the reflected light are monitored. These polarisation changes are highly sensitive to the presence of a surfactant layer. Ellipsometry allows the average thickness and refractive index of the layer throughout the adsorption process to be calculated by monitoring the ellipsometric angles c and D with time. For an extensive explanation of ellipsometry, see Ref. w95x. OR records the amplitudes of the perpendicularly polarised components of the reflected beam. Upon adsorption, the ratio of these two intensities is altered and this allows for calculation of the surface excess. OR is described in detail in Ref. w96x. Unlike NR, reflectance techniques using light as the radiation source cannot give information concerning the molecular structure of the adsorbed layer. 6.3.3. Hydrodynamic considerations In order to make quantitative statements concerning the adsorption kinetics it is critical that the hydrodynamics of the measurement be well defined. Whether the adsorption process is transport-limited or limited by the rate of layer organization cannot be discerned unless the rate of diffusion of surfactant to the substrate is known. Several studies have suffered from a lack of hydrodynamic control w97 101x. This has complicated the accompanying attempts at modeling the observed kinetics. However, the qualitative results obtained are still of value. When hydrodynamic conditions are well defined, such as that provided by use of a stagnant point flow w102x, the character of the adsorption process can be commented upon with much greater certainty. 6.3.4. Ellipsometric measurements of nonionic surfactant adsorption Several ellipsometry studies have investigated the adsorption of nonionic surfactants w98101x. Whilst the hydrodynamics of the interface were not well controlled, interesting results were nonetheless obtained. For a nonionic surfactant adsorbing to silica, only weak interactions are expected between the surfactant and the substrate. In this case, a transport limited adsorption process is expected, and the results obtained were in good agreement with this. Tiberg et al. w98x used in situ ellipsometry to study the adsorption and desorption kinetics of polyethylene glycol alkyl ethers at the silicaaqueous interface. A typical adsorptionydesorption cycle was determined that comprised of five separate kinetic regions. It was found that the rate of adsorption was strongly dependent on the bulk surfactant concentration. The rate of adsorption continued to increase above the cmc, clearly demonstrating the involvement of micelles in the adsorption process. Representative results are presented in Fig. 28. The increasing section of the adsorption profile was divided into two separate regions: first, a linear region where adsorption increases monotonically with time, followed by a transition region where the adsorption rate decreases. The linear region of the result was attributed to the rate of diffusion of monomers and micelles from the bulk through the stagnant layer in the immediate vicinity of the surface being lower than the rate of adsorption to the substrate. That is, in this portion of the data the adsorption rate was thought to be transport limited. As the surface

274 R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304 Fig. 28. Left: Time evolution of the surface excess (G, open diamonds) and mean optical thickness (df , filled diamonds) for C14 E6 at concentrations, from bottom, 0.007, 0.009 and 0.01 mM. All concentrations are below the cmc. Right: Time evolution of the surface excess (G, open diamonds) and mean optical thickness (df, filled diamonds) for C14E6 at concentrations, from bottom, 0.025, 0.1 and 0.25 mM. All concentrations are above the cmc. 1994 ACS. Reproduced from Ref. w98x.

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

275

Fig. 29. Desorption kinetics from silica for three different hexaethylene glycol monoalkyl ethers: C12E6, (filled diamonds); C14E6 (open diamonds); C16E6 (filled triangles). The inset shows the corresponding process for different polyethylene glycol monododecyl ethers: C12E5 (open triangles); C12E6 (filled triangles); C12E8 (open diamonds). 1994 ACS. Reproduced with permission from Ref. w98x.

excess increased, the concentration gradient between the surface and the bulk decreased. This lowered the rate of adsorption to the surface and led to the curved region of the adsorption data. Throughout this transition region, the rate of adsorption steadily decreased, presumably as the surface became more saturated. This eventually led into the plateau region of adsorption. For surfactant concentrations above the csac, it was shown that the equilibrium surface excess increased as the size of the hydrocarbon moiety of the surfactant increased. Similar results have been reported for ionic surfactants by solution depletion methods cf. Section 3.3. Interestingly, for all surfactant concentrations studied, the measured mean optical thickness was found to rapidly increase to a steady value of approximately 4.7 nm. This was interpreted as indicating that the adsorbed layer was built up of micellar like structures, which had a well-defined thickness even in the initial stages of the adsorption process. Higher surface excess values were achieved by the formation of more surface aggregates that eventually saturated the surface, or by increasing the packing density of monomers in the existing aggregates. When the adsorption plateau was reached using a bulk concentration above the cmc, no desorption was apparent until the solution surfactant concentration fell below the csac. This demonstrated the cooperative nature of nonionic surfactant adsorption. The general features of the desorption kinetics are given in Fig. 29, showing the effect of alkyl chain length and head-group size. Two regions were also identified in the desorption results. The surface excess initially decreases in a linear fashion with time. It was suggested that the rate of diffusion away from the surface was the rate-limiting step in this region. As shown in Fig. 29, the rate of desorption increases by an order of magnitude when the length of the alkyl chain is increased by two carbon atoms. However, variation in

276

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 30. Reversibility of the adsorptiondesorption cycle of C12 E5 adsorbed on polystyrene coated silica. C1s0.17 mM. First desorption upon dilution with C2s0.0006 mM. Further desorption with water. Second cycle: adsorption of a 0.17 mM surfactant solution. Desorption by water. The soluteysolvent switches are indicated by arrows. 2000 ACS. Reproduced with permission from Ref. w103x.

the head-group size had only a slight effect on the desorption rate. These effects were thought to be directly correlated with the corresponding changes in the csac. That is, as surfactant diffuses away from the surface, the adsorbed equilibrium adjusts such that in the region of the solution immediately adjacent to the surface the surfactant concentration is kept constant at the csac. The lower csac of the surfactants with longer tail-groups was therefore able to bring about the shorter desorption times observed. The increased strength of hydrophobic interactions between adsorbed surfactants was not considered. The decrease in surface excess eventually becomes non-linear. This slowing of desorption suggested that the monomer concentration could no longer be maintained at the csac. This shows that as the surface excess decreases, the rate of dissociation of surface micelles becomes the rate-determining step. 6.3.5. OR studies of nonionic surfactant adsorption The adsorption of nonionic surfactants to synthetic hydrophobic substrates has also been investigated. Geffroy et al. w103x used OR with stagnant point flow hydrodynamics to study the adsorption of ethylene oxide surfactants to a polystyrene coated silica surface. Interestingly, it was shown that a portion of the surfactant layer was irreversibly bound to the substrate. The hydrodynamics associated with the stagnant point flow make it particularly suitable for cycling experiments, where solvent, then surfactant solution is successively flowed toward the substrate. An example of such an experiment is shown in Fig. 30, which shows that the initial adsorption with a surfactant concentration of 0.17 mM rapidly leads to a surface excess of ;1 molecule nmy2. This was first

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

277

desorbed with a much lower concentration of surfactant, 0.0006 mM, which led to the surface excess being reduced to ;0.4 molecule nmy2. When pure water was passed into the cell, no change in the surface excess was detectable. Upon reintroduction of the higher surfactant concentration the original surface excess was obtained, and rinsing with water returned the surface excess to the lower adsorption density. This led the authors to suggest that an adsorbed amount of ;0.4 molecule nmy2 would be obtained at very low surfactant concentrations. The authors suggested that this irreversibly bound surfactant was adsorbed flat at the surface based on the form of the desorption kinetics and molecular area considerations. In this study the kinetics of adsorption were relatively fast, and the equilibrium surface excess values were reached within a few minutes. Below the cmc, the rate of adsorption increased linearly with concentration. Above the cmc the rate of adsorption continued to increase albeit more slowly. In this system, it was not expected that the micelles would adsorb to the surface directly, as the head-groups are quite hydrophilic. Nonetheless, the observed increase in the rate of adsorption above the cmc suggested that the micelles were acting as a source for monomers adsorbing to the surface, which increased the adsorption rate. 6.3.6. Adsorption kinetics of CTAB on silica Several studies have investigated the adsorption kinetics of CTAB at the silica aqueous interface. The presence of the charged head-group should provide a greater initial driving force for adsorption. However, the native charge associated with the silica surface will be neutralised at low surface excess values. Goloub et al. w2x have shown that the surface charge will compensate for the charge associated with the adsorbed surfactant to a significant extent, but the maximum potential of the surface has been reached well below the saturation surface excess. Thus, for a surfactant concentration above the csac, adsorption is against an electrostatic repulsive barrier throughout most of the adsorption process. 6.3.6.1. Ellipsometry. Eskilsson and Yaminsky w104x used in situ ellipsometry to study the adsorption of CTAB to silica. The adsorption isotherms were determined for both CTAB and CTAq, reproduced in Fig. 31. These isotherms showed that the level of surfactant adsorption began to increase significantly at approximately 0.1 mM, with the most rapid increase in the level of adsorption occurring between 0.4 and 0.8 mM. In this concentration range the surface excess rose from 1 to 4 mmol my2, which was the saturation surface excess. At the beginning of this region of sharp increase in surface excess, the level of counterion adsorption became appreciable. This indicates that surfactant is starting to be adsorbed with the headgroup facing away from the surface. These results are in reasonable agreement with those obtained by Goloub et al. w2x for a similar cationic surfactant. For solution surfactant concentrations above 0.4 mM the thickness of the adsorbed layer was determined to be approximately 2.5 nm. This value did not increase with surfactant concentration, even though at this concentration the surface excess was only 25% of the saturation value. It was suggested that the Coulombic repulsions between adsorbed surfactant serve to orientate the surfactants normal to the surface.

278

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 31. Adsorption isotherms for CTAB (crosses) and CTAq (circles) at the silicaaqueous interface. 1998 ACS. Reproduced with permission from Ref. w104x.

The kinetics of adsorption were found to be strongly dependent on the bulk concentration. For surfactant concentrations less than 0.1= cmc up to 2 h was required for equilibrium to be obtained at the surface. However, as the surfactant concentration approached and exceeded the cmc, adsorption was complete within minutes. For all surfactant concentrations studied, the rate of desorption was fast, complete within a few minutes and no concentration dependence was found. Examples of adsorption and desorption experiments are shown in Fig. 32. 6.3.6.2. Optical reflectometry. Pagac et al. w105x investigated the adsorption of CTAB to silica using OR with laminar flow hydrodynamics. Unfortunately, this study was influenced by a trace impurity that caused large overshoots in the surface excess values for surfactant concentrations below the cmc. This impurity was later shown to be due to leaching from PVC tubing contained within a peristaltic pump w73x. Results obtained above the cmc appear to be unaffected, and display kinetics similar to those obtained by Eskilsson and Yaminsky w104x. The equilibrium was reached within approximately 1 min of adsorption commencing. Once again, the involvement of micelles in the adsorption process was indicated by an increase in the rate of adsorption above the cmc. 6.3.7. The role of micelles in adsorption The role of micelles in the adsorption process for CTAB w106x and CPBr w107x was elucidated by our group using an optical reflectometer with stagnant point flow hydrodynamics. The well defined hydrodynamic flow fields associated with stagnant point flow permits the measured initial rate of adsorption to be compared to the

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

279

Fig. 32. Surface excess vs. time during adsorption from 1 mM CTAB solution. The surfactant is injected at time 0. Rinsing of the sample cell with pure water is initiated at ts1420 s. The inset shows the adsorption from 0.1 mM CTAB. 1998 ACS. Reproduced with permission from Ref. w104x.

theoretically derived diffusion limited flux to the surface w102x. The quotient of these two values is recorded as the sticking ratio w106x, and is simply the number of surfactant molecules that are adsorbed to the surface normalised by the theoretically derived diffusion limited flux to the surface. Note that this is an oversimplification of the actual process, as adsorbed surfactant molecules are freely exchanging with surfactant in the bulk. However, trends in the sticking ratio can give valuable insight into the nature of the adsorption process. An increase in sticking ratio with concentration indicates cooperativity, a decreasing sticking ratio suggests competitive adsorption, while a constant sticking ratio infers that surfactant molecules are adsorbing independently. The sticking ratios for CTAB and CTAB with 10 mM KBr are shown in Fig. 33. In this figure the concentration of surfactant has been normalised by the cmc for each system (1.25=10y4 M with 10 mM KBr, w108x, 9.0=10y4 M without salt w109x). For concentrations above the cmc the theoretical flux consists of two components, the flux due to monomers and the flux due to micelles. Any increase in the concentration above the cmc is reflected only in the flux due to micelles as the monomer concentration is assumed to be constant. The important result from Fig. 33 is the marked increase in the sticking ratio for both systems at the cmc, indicating an increase in the efficiency of the adsorption process above the cmc. If surfactant molecules were competing for adsorption sites, this would be reflected in a reduction in the sticking ratio. In the absence of salt, the sticking ratio is essentially constant up to the cmc indicating that the monomers are adsorbing independently. With 10 mM KBr present the sticking ratio is higher

280

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 33. The sticking ratio (determined from the initial rate of adsorption) vs. concentration for solutions of CTAB, normalised by the corresponding cmc, with (squares) and without (diamonds) added 10 mM KBr. The sticking ratio increases at the cmc for both systems. 2000 ACS. Reproduced with permission from Ref. w106x.

than in the absence of salt and is seen to gradually increase even below the cmc, indicating that adsorption of monomers is cooperative in this case. Cooperative adsorption is clearly aided by the screening of the head-group charge by electrolyte. Perhaps such screening allows hydrophobic interactions to play a greater role in adsorption? The sticking ratio increases sharply at the cmc, both in the presence and absence of electrolyte, which clearly indicates that above the cmc adsorption has become cooperative. Similar trends were determined for CPBr w107x. The most obvious explanation for cooperativity is that micelles are directly adsorbing to the surface, either partially or wholly. For this to lead to the observed trend in sticking ratio the increased success with which micelle bound monomers adsorb to the substrate must be due to more effective penetration of the electrical double layer of the interfacial layer by micelles compared to monomers. Opposing adsorption is the electrostatic repulsion of this double layer, and as a micelle has a significant number of counterions associated with it, we propose that surfactant contained within a micelle will more effectively penetrate the double layer than a monomer. This is due to the reduced charge per monomer contained in a micelle. 6.3.8. The influence of electrolyte on adsorption The effect of changing the counterion from bromide to chloride on the surface excess, kinetics of adsorption and adsorbed layer structure was investigated using OR and AFM in the study of Velegol et al. w73x. The implications for the adsorbed layer structure have already been discussed in Section 4.5.1. For now we shall concern ourselves primarily with the adsorption kinetics. Below the cmc, the surface excess reached 80% of its maximum value within 2 min, while above the cmc, equilibrium was attained within this time interval. When the surface was rinsed with water, the surfactant was completely desorbed within seconds.

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

281

Fig. 34. Response of the adsorbed layer to a change in the type of counterion in solution. All solutions contain 2 mM C16TAq Xy in 10 mM KX where X is either Bry or Cly . The composition is varied as follows: (1) CTAB in KBr; (2) CTAC in KCl; (3) CTAC in KBr; (4) CTAB in KCl; (5) CTAB in KBr. The surface excess for the upper curve is calculated from the optical data by using the refractive index increment of CTAC and for the lower curve by using the refractive index increment of CTAB. 2000 ACS. Reproduced with permission from Ref. w73x.

The suitability of OR for performing cycling experiments is demonstrated by Fig. 34. This experiment utilised a 2 mM surfactant solution with 10 mM electrolyte, which was continuously flowed over the substrate. The composition of the solution was varied between combinations of CTAB, CTAC, KBr and KCl. In any combination, the added electrolyte provided 83% of the total counterions in solution. For CTAC with 10 mM KBr, the surface excess obtained is the same as that of CTAB with 10 mM KBr. However, when the chloride concentration was greater than that of bromide, as with CTAB with KCl, the surface excess is intermediate to that reached for CTAC with KCl and CTAB with KBr. These results show that when both the chloride and bromide counterions are present, there is competition for binding to the aggregate surface, both in solution and adsorbed at the interface. This influenced the surface excess obtained. The results obtained in this experiment were consistent with the bromide ion having a much higher affinity for the aggregate than the more hydrated chloride ion. This was evident in the higher surface excess values obtained for the bromide counterion.

282

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 35. Adsorption Isotherms for CTAC; without added electrolyte (closed diamonds), with 10 mM KCl (open diamonds), with 10 mM NaCl (open triangles) and with 10 mM LiCl (open circles). For more information see Ref. w112x.

6.3.9. The influence of co-ion type on CTAC adsorption We have also investigated the effect of variation in the co-ion type on adsorption behaviour. The CTAC system was used to investigate this effect, as it has been demonstrated w73,75,110,111x that the concentration of chloride ions has little influence on the surface aggregate structure formed upon adsorption of CTAC. Thus, if variation in adsorption is observed, it is likely to be a consequence of altering the electrolyte co-ion. The adsorption isotherms for CTAC and CTAC in 10 mM NaCl, KCl and LiCl (henceforth referred to as CTAC and CTACqNaCl, CTACqKCl and CTACqLiCl, or collectively as CTACqXCl) are presented in Fig. 35. The adsorption isotherms are shifted to lower concentrations in the presence of electrolyte. Importantly, the size of this shift is independent of the co-ion identity. In fact, the adsorption isotherms for the different salts are remarkably similar, indicating that the co-ion has little effect on the surface excess above or below the cmc. It is possible that the different co-ions cause different adsorption behaviour at very low surfactant concentrations where electrostatics provide the driving force for adsorption. However, the surface excess values at these adsorption levels are below the sensitivity of the reflectometer. The rates of adsorption for the three electrolyte types were also alike at all surfactant concentrations, indicating that the co-ion has little effect on the adsorption rate. These results agree well with the AFM study of Subramanian and Ducker w78x (Section 4.5.3) that suggested that the addition of LiCl up to a concentration of 500 mM had little effect on the adsorbed aggregate morphology of CTAC. The results obtained here extend this finding, and show that KCl, NaCl and LiCl have no

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

283

Fig. 36. (a) Adsorption isotherms for CTAB (open squares), MTAB (closed diamonds) and DTAB (open triangles) in the presence of 10 mM KBr (b) demonstrates the similarities in the form of the adsorption isotherms when the same data are plotted as a function of the solution cmc. For more information see Ref. w112x.

measurable effect on the saturation surface excess up to 10 mM XCl. Thus, it could be expected that similar surface aggregate structures would be present for CTAC in the presence or absence of 10 mM XCl. 6.3.10. Effect of chain length on adsorption We have also studied the influence of the length of the hydrocarbon chain of quaternary ammonium surfactants using OR w112x. The isotherms in Fig. 36 demonstrate that the data are qualitatively similar for DTAB, MTAB and CTAB in the presence of 10 mM KBr. At concentrations below 0.1= cmc, where surface

284

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 37. Initial adsorption rate for CTAB (open squares), MTAB (closed diamonds) and DTAB (closed triangles) in the presence of 10 mM KBr. The dashed vertical lines represent the solution cmc for each surfactant system. Reproduced from Ref. w112x.

excess values are low, the isotherms are essentially the same within instrumental limitations. For concentrations greater than 0.1= cmc the isotherms separate, with the surface excess increasing in the order DTAB-MTAB-CTAB for a given fraction of the cmc. The increase in the surface excess is too large to be simply due to the increased mass of individual surfactants i.e. the number of surfactant molecules adsorbed at the interface increases in the order DTAB-MTAB-CTAB. The kinetics of adsorption for DTAB, MTAB and CTAB in 10 mM KBr are presented in Fig. 37. The general form of the data are the same for the three surfactants. Below the cmc the rate of adsorption increases steadily with concentration. In the vicinity of the solution cmc there is an abrupt increase in the rate of adsorption, as noted previously in Section 6.3.7. This is due to the direct adsorption of micelles, which facilitates rapid and effective surface aggregate formation. Comparison of the adsorption kinetics for the three surfactants is not as simple as it may at first appear. If one chooses a surfactant concentration and compares the adsorption rates, unless all three surfactants are below or above their cmc values, different surface and bulk structures will be present. This will affect the rate at which the adsorbed layer is created and thus the initial adsorption rate. It may seem that the solution to this problem is to normalise the data by the solution cmc so that the surface structures and hydrophobic driving force at a given fraction of the cmc are similar. However, in doing this the absolute concentration of the surfactants being compared will be vastly different, which will have a similarly large effect on the flux of surfactant to the substrate. This obviously hampers any comparison. However, it is possible to make some direct comparisons in certain regions of the isotherms, and several more general observations. Below 0.02 mM, where the surfactants are far below their solution cmc values, the initial adsorption rates are low for all three surfactants. At these concentrations

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

285

it is expected that electrostatic interactions between the surfactant and the substrate provide the driving force for adsorption and that hydrophobic interactions between the tail-groups will have little, if any, effect on the adsorption rate. Thus, the initial adsorption rates are all quite similar. At 0.1 mM the three surfactants are still below their respective cmc values, but there is a clear trend with the adsorption rate increasing in the order DTABMTAB-CTAB. If one examines the adsorption isotherms (Fig. 36) at this concentration, we note that the surface excess values increase in the same order. That is, not only does an increased hydrophobic driving force lead to the assembly of a more complete adsorbed layer, it also leads to significantly faster adsorption rates. Above 1 mM CTAB and MTAB are both above their solution cmc values. Once again CTAB has the faster adsorption rate. The increased rate of CTAB adsorption in this case is most likely due to the fact that the CTAB solution will have a much higher proportion of its surfactant contained within micelles. As micelles are considerably more effective at covering the substrate than monomer units w106,107x, this brings about a faster adsorption rate. Above the cmc for DTAB no data were obtained for CTAB. However, it is still possible to compare data for DTAB and MTAB. The same effect occurs in this case, with MTAB having a significantly faster adsorption rate for the concentration corresponding to the DTAB cmc. In general, once the surface excess is appreciable (i.e. above 0.2 mM) the rate of adsorption is always greater for the surfactant with the longer hydrocarbon chain. When the surfactants are all below their solution cmc values it is the increased hydrophobic interactions between monomers that leads to the faster adsorption rate. A longer tail-group surfactant will also have a greater proportion of surfactant contained within micelles (when present). As micelles are more effective in creating an adsorbed layer on the substrate, this means that the longer chain surfactant will also always have a faster adsorption rate above the cmc. Thus, at any concentration, the adsorption rate always increases in the same order as the tail-group size. 6.3.11. The slow adsorption region Our OR investigations have revealed a concentration range in the vicinity of the csac where long-term increases in adsorption are observed w106,107,111,112x. This region is known as the slow adsorption region (SAR), and was first noted for 0.6 mM CTAB in the absence of salt, where the adsorption of surfactant occurred over a period of several hours. This is shown in Fig. 38. The surface excess rapidly increased up to ;0.6 mg my2, but then proceeded very slowly until it reached the final equilibrium adsorption value of ;1.6 mg my2 which was equivalent to that reached at all higher concentrations of CTAB. Similar long-term increases in surface excess have also been elucidated for CTAC w111x for surfactant concentrations between 0.6 and 0.9 mM. These unusual results were investigated in more detail using CPBr. As CPBr is UV active, solution surfactant concentrations could be determined with high precision spectrophotometrically. This allowed the effect of small increases in surfactant concentration on slow adsorption phenomena to be ascertained. As shown

286

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 38. Surface excess of CTAB at the pyrogenic silicasolution interface vs. time for 0.6 mM CTAB. Adsorption occurs over a much greater time period than for other concentrations. The surface excess initially increases rapidly to 0.6 mg my2 (which equals the equilibrium excess achieved at slightly lower concentrations) then over a number of hours adsorption increases to the plateau level obtained for slightly higher concentrations. At these higher concentrations equilibrium adsorption levels are obtained within minutes Upon introduction of pure water at 12 000 s the adsorption returns to baseline levels. Reproduced with permission from Ref. w106x.

in Fig. 39, solution CPBr concentrations of 0.2 and 0.55 mM were sufficient to lead to a plateau coverage of 0.4 and 1.6 mg my2, respectively. It was between these two boundaries, and only between these boundaries, that long-term increases in CPBr adsorption were noted. For CPBr concentrations between 0.274 and 0.306 mM, the surface excess rapidly increased to approximately 0.4 mg my2, and then continued to increase at a greatly reduced rate. Changes in solution CPBr concentrations of only 0.01 mM were sufficient to bring about notable increases in the rate of the secondary adsorption. The rate of this secondary increase in adsorption was proportional to the surfactant concentration. Increasing the solution surfactant concentration to 0.336 mM CPBr led to a fast increase in the level of surface coverage to 0.9 mg my2, followed by a similar slow increase in the level of adsorption. The rate of the secondary adsorption in this case was decreased over that observed for concentrations between 0.274 and 0.306 mM. For each of these experiments the surface excess had not reached plateau levels after 30 h. Limitations associated with data collection prevented longer observation times. These results suggest that in the secondary phase of the adsorption process adsorbing surfactants are filling surface adsorption sites stochastically. As the concentration is increased, the probability of a monomer adsorbing is also increased, leading to a faster rate of secondary adsorption. The initial level of surface excess is higher for 0.336 mM (0.9 mg my2), which suggests that the adsorbed surfactant layer is more complete in the initial stages of adsorption than that obtained with slightly lower bulk concentrations. Consequently, there was a reduced number of

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

287

Fig. 39. Measured surface excess of CPBr at the hydroxylated silicasolution interface vs. time for CPBr concentrations of 0.202 (filled circles), 0.274 (filled diamonds), 0.284 (filled squares), 0.306 (filled triangles), 0.315 (open squares), 0.336 (open diamonds) and 0.554 mM (open triangles) with no added electrolyte. Adsorption occurs over a much greater time period than for concentrations outside this range. The surface excess initially increases rapidly to 0.4 mg my2 (or 0.9 mg my2 for 0.336 mM CPBr) then continues to increase over many hours. 2000 ACS. Reproduced with permission from Ref. w107x.

adsorption sites available on the surface, which led to the observed reduced secondary rate of adsorption at this concentration. In this work AFM imaging was used to investigate the structural changes that must accompany these slow increases in adsorption. AFM images of the silica surface were obtained 20 min and 22.5 h after the injection of 0.3 mM CPBr. After 20 min had elapsed there was no evidence of any structure present in the adsorbed layer on the silica surface, despite a surface coverage of ;0.4 mg my2 obtained by reflectometry. However, after 22.5 h the presence of an adsorbed layer structure consisting of elongated admicelles is clear w107x. Changes in the force vs. distance data collected over this time period were also consistent with the slow formation of surface aggregates. It appears that the SAR is a consequence of kinetic barriers to the formation of the thermodynamically stable arrangement of adsorbed surfactant. As the SAR occurs at concentrations below the solution cmc, adsorption is due only to monomers. These monomers must participate in an aggregation process at the interface that results in the formation of surface structures analogous to bulk micelles. During this process, a monomer may have to sample many sites before it is successfully incorporated into a surface aggregate. Additionally, an incoming monomer must also overcome an electrostatic barrier to reach the surface, as the surface excess of 0.4 mg my2 is sufficient to cause charge reversal of the silica surface. Thus, we concluded that the SAR is a result of kinetic barriers to the adsorption process, in the form of structural and electrostatic components. The boundaries of the SAR were determined by both the surface structure and coverage, and by the aggregate structure of surfactant in solution. At concentrations below the SAR the concentration in bulk is not sufficient to raise the chemical potential of the monomer to a level where surface aggregation is favourable. The

288

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

surface excess therefore indefinitely remains below that required to give rise to surface aggregates. At concentrations above the SAR surface aggregates are clearly forming and are doing so rapidly. One explanation is that at concentrations above the SAR but below the cmc a number of transient semi-formed aggregates are present in solution and these premature aggregates adsorb to the surface, negating any requirement for monomer units to slowly pack into a preformed aggregate structure. Alternatively, as the bulk substrate concentration is increased the electrostatic repulsion between the monomer and the surface will be screened more effectively. This will decrease the energy barrier to adsorption for monomers and facilitate more rapid accumulation at the surface. The electrostatic screening interpretation is supported by the presence of a SAR for CTAC in the presence of 10 mM LiCl w112x. This important result shows that long-term increases in adsorption can occur in the presence of electrolyte, provided the electrolyte does not influence the surface structure. Recall that the addition of 10 mM XCl to CTAC solutions has little effect on the equilibrium surface excess w112x (cf. Fig. 35) or the adsorbed structure (cf. Section 4.5.3). Thus, while the presence of 10 mM XCl diminishes the electrostatic repulsion an adsorbing monomer experiences in the secondary adsorption phase, the same structural barriers to adsorption remain in the presence of electrolyte. This permits long-term increases in adsorption for CTACq10 mM LiCl and indicates that it is the structural barrier to adsorption that is critical for the evolution of slow adsorption effects. Using OR in conjunction with AFM, similar secondary increases in surface excess have been observed for low surfactant concentrations at high pH w111x. The adsorption kinetics were monitored using OR, and it was shown that for 0.11 mM CTAB at pH 9.4, a rapid increase in adsorption occurred in the first 10 s, followed by a slow rise (23 h) after which the equilibrium surface excess was reached. This result was complemented by variation in force vs. apparent separation data as a function of time and AFM imaging. Immediately after surfactant was passed into the AFM cell, the substrate-tip interaction was dominated by an attractive jump into contact starting from an apparent separation of approximately 10 nm. This originates from a hydrophobic attraction between the tail-groups of the surfactant adsorbed to the tip and surface. With additional time, a steeply repulsive force was observed, typically from a separation of approximately 5 nm. This permitted images of the adsorbed aggregates to be obtained, suggesting an adsorbed morphology similar to that obtained at much higher surfactant concentrations, cf. Section 4.5.1. These results show that a critical density of surfactant molecules adsorbed at the interface is required to initiate spontaneous aggregation on the surface, rather than a certain bulk surfactant concentration. Increasing the pH of aqueous solutions in contact with silica leads to a substantial increase in surface charge. This promotes the adsorption of cationic surfactants from dilute solution. Once a certain surface excess is reached (0.6 mg my2 for CTAB on silica), the density of hydrophobe on the surface is such that bilayered aggregation is initiated i.e. the adsorption of more surfactant via hydrophobic interactions. The successful imaging of these aggregates indicates that the csac for silica at pH 9.6"0.4 is in the vicinity of 0.055 mM

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

289

Fig. 40. Surface excess of CTAB on hydroxylated silica vs. time for (a) sequentially increasing and (b) sequentially decreasing CTAB concentrations in the absence of electrolyte. A stable baseline was obtained in water; the surfactant solutions were passed into the cell in series. Surfactant concentrations used were 0.05 mM (filled circles), 0.15 mM (open diamonds), 0.3 mM (open triangles), 0.5 mM (open circles), 0.6 mM (closed squares), 0.7 mM (closed diamonds), and 0.8 mM (open squares). Note the long-term increase observed for 0.6 mM. Surfactant concentrations of 1, 3 and 5 mM were also used. These concentrations all produced an equilibrium surface excess of 1.4 mg my2 within experimental limitations and are not shown here for clarity. Water (crosses) is passed into the cell at 16 500 s and the non-zero surface excess obtained is due to baseline drift. Only representative data points are provided. However, the line to guide the eye provides a reasonable indication of the form of the experiment. Reproduced from Ref. w113x.

CTAB, about an order of magnitude lower than at neutral pH w111x. As such it is no longer logical to normalise the csac as a function of the bulk concentration but to acknowledge that a certain surface concentration of adsorbed surfactant is required to initiate surface aggregation. This approach not only considers the surfactantsurfactant interactions which are correlated with the bulk behaviour but includes the influence of the substrate.

290

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

OR enables different surfactant concentrations to be consecutively analysed, this allowed the SAR to be further probed by cycling experiments w113x. Fig. 40a shows the increasing concentration cycle and Fig. 40b the decreasing concentration cycle, where surfactant concentrations were sequentially equilibrated at the interface. As expected, a stepwise increase in surface excess is noted for all concentrations in the increasing concentration cycle, except for 0.6 mM, where long-term increases in surface excess are noted. In the decreasing concentration cycle of this experiment the surfactant solutions were passed into the cell in reverse order and a stepwise decrease in surface excess results. When water was passed into the cell at 16 500 s, a surface excess of approximately 0.15 mg my2 results. This reflects drift in the baseline over the course of the experiment, rather than residual surfactant adsorbed at the surface. We note that the drift in these experiments is greater than for others as the temperature control is adversely influenced by the large number of solution changes. The long-term increase for 0.6 mM observed in this concentration cycling experiment (Fig. 40a) is somewhat different to that described above in Fig. 38, as in this case the surface excess does not reach saturation levels of coverage. Whilst this may seem problematic, there are important differences between the experiments presented in Figs. 38 and 40a. In the first case, 0.6 mM CTAB is introduced to a substrate which is initially bare and a fast increase in surface excess to 0.6 mg my2 results. It is assumed that, in the period of fast adsorption, the surfactant adsorbs in a manner which is favourable for continued adsorption, albeit slow. This allows the adsorption to continue, and saturation surface excess is eventually reached. In the concentration cycling experiment, when the 0.6 mM surfactant is passed into the cell, it encounters a pre-adsorbed surfactant concentration of 0.6 mg my2 that has been sequentially built up over a period of approximately 1 h. Thus, in the cycling experiment when 0.6 mM is passed into the cell, the structure initially present at the interface is the equilibrium structure for 0.5 mM CTAB. This pre-adsorbed structure leads to a different structural adsorption path that becomes kinetically trapped at a surface excess below the equilibrium surface excess. Because of this, it is not until the surfactant concentration reaches 0.8 mM that saturation levels of coverage are reached. Surprisingly the desorption cycle reveals an increase in surface excess for concentrations below 0.5 mM. Some of this discrepancy can be attributed to the drift in the baseline of ;0.1 mg my2, but not all of it. This suggests that in the increasing cycle, equilibrium is not achieved at the lower concentrations despite the observation that the surface excess is stable. Perhaps this is an extreme case of kinetically trapped conformations and suggests that the SAR may be indicative of a more general effect at low concentrations and that accurate equilibrium adsorption isotherms may only be obtained by desorption from higher concentrations. These results highlight the importance of the surfactant structure in slow adsorption, and also serve to illustrate the folly of implying kinetic information from equilibrium data.

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

291

Fig. 41. The adsorption of 12-s-12 gemini surfactants at the silicasolution interface at 1= cmc: 12-212 (filled diamonds), 12-3-12 (closed squares), 12-4-12 (closed triangles), 12-6-12 (closed circles), 128-12 (open squares), 12-10-12 (crosses) and 12-12-12 (open triangles). A typical adsorption result for 1 mM CTAB (open circles) is included for comparison. The surfactant solution is first passed into the cell at ;15.5 s leading to surfactant adsorption. The plateau level of adsorption is maintained whilst the surfactant solution is flowing into the cell. 2003 ACS. Reproduced with permission from Ref. w77x.

6.3.12. Adsorption of gemini surfactants to silica by OR The adsorption kinetics of 12-s-12 gemini surfactants to silica has been investigated w77x, and is presented in Fig. 41. The surfactant concentrations shown are ;1= cmc. For a given spacer length there was no variation in the form of the data over the range of surfactant concentrations studied. For each surfactant a baseline was obtained for ;15 s prior to surfactant being introduced into the cell, at which time the surface excess rapidly increases. For all spacer lengths, equilibrium adsorption levels are attained within 4 s of adsorption commencing. The surface excess rises rapidly to approximately 70% of the equilibrium surface excess, followed by a slower increase to the final value. The equilibrium surface excess remained constant whilst the surfactant concentration was maintained. For comparison, the adsorption curve of CTAB, also at 1= cmc, is included. CTAB has a longer alkyl chain (C16) than the gemini surfactants under investigation here, but a similar cmc (0.9 mM), and is therefore an appropriate comparison. The form of the adsorption for CTAB is similar to that of all the gemini surfactants and identical to that obtained for 12-3-12. Upon water being passed into the cell, desorption was complete and rapid for all surfactants (not shown). This study reported the most complete set of adsorption isotherms obtained for gemini surfactants at the silicaaqueous interface, with adsorption isotherms for seven different spacer sizes investigated, presented in Fig. 42. Below 0.3= cmc, the adsorption isotherms are coincident. This indicates that electrostatic interactions between the substrate and the adsorbed surfactant play a dominant role in the adsorption process up to this concentration. The saturation surface excess decreases

292

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 42. Adsorption isotherms for 12-s-12 gemini surfactants at the silicasolution interface with the concentration axis as a function of the cmc: 12-2-12 (filled diamonds), 12-3-12 (closed squares), 12-412 (closed triangles), 12-6-12 (closed circles), 12-8-12 (open squares), 12-10-12 (crosses) and 12-1212 (open triangles). Below 0.3= cmc there is no apparent variation in surface excess with spacer size. Above 0.3= cmc the isotherms begin to separate, and the saturation surface excess decreases with increasing spacer size. 2003 ACS. Reproduced with permission from Ref. w77x.

with increasing spacer size. The implications of this decrease for the adsorbed morphology are detailed in Section 4.5.4. There are two important differences between the data obtained in this study and the depletion studies examined in Section 3.4.6. Firstly, solution depletion studies (cf. Fig. 12) show a substantial increase in surface excess above the solution cmc, an effect that was more pronounced for short spacers. No evidence of this effect was detected in the current study for any spacer length. The isotherms for all spacer lengths reached saturation at or slightly below the solution cmc. The same behaviour is found for conventional cationic surfactants adsorbing to silica w106,107x. The cause of the discrepancy between this study and those previous is unclear. The second discrepancy relates to the concentration required to lead to the second adsorption stage. Previous adsorption isotherms reported that the second adsorption stage for the longest spacer investigated (12-10-12) commenced at lower concentrations than for those with shorter spacer groups. This was attributed to the increased hydrophobicity of the longer chain surfactant. In this study the concentration leading to the second adsorption step was independent of the spacer length. The difference between the two studies may be related to a change in the ionic strength of the solution on surfactant adsorption that can occur in depletion studies. Surfactant adsorption to charged surface sites on silica can induce nearby hydroxyl groups to become more acidic w2x. For longer spacer groups this may result in both headgroups interacting with the substrate and a consequent increase in solution ionic strength w41x. An increased ionic strength will result in the solution cmc being decreased, shifting the features of the adsorption isotherm to lower concentrations.

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

293

Fig. 43. Adsorption isotherms for CTAB (closed squares), DTAB in no added electrolyte (closed triangles) and DTAB in 10 mM NaBr (open triangles) on the AAPP substrate. The filled lines are drawn to guide the eye. The dashed lines separate the regions of the adsorption isotherms. Within instrumental limitations, the adsorption isotherms for CTAB and DTAB with no added electrolyte are indistinguishable below 0.1 mM, as are the DTAB no added electrolyte and DTAB with 10 mM NaBr isotherms above approximately 3 mM. 2003 ACS. Reproduced with permission from Ref. w114x.

In the experiments using the reflectometer, the ionic strength of the solution is unaffected by surface ionisation. 6.3.13. Adsorption of ionic surfactants to a charged hydrophobic substrate For comparison, surfactant adsorption at a negatively charged, hydrophobic, plasma polymer substrate was used to study the influence of the substrate on adsorption behaviour w114x. The acetaldehyde plasma polymer (henceforth denoted AAPP) surface used is similar to silica in charge density but differs in that the regions between the charge sites are hydrophobic. The adsorption isotherms for CTAB, DTAB and DTAB with 10 mM NaBr are presented in Fig. 43 and for SDS and SDS with 10 mM NaBr in Fig. 44. The isotherms can easily be divided into four-regions. Region I is a low surface excess region where the level of adsorption was independent of the bulk concentration. Regions II and III occurred where the surface excess was increasing linearly with concentration, the latter increasing more steeply, and a final plateau region where no further adsorption is observed as the bulk concentration of surfactant is increased (region IV). It was noted that the first region was absent for the anionic surfactant systems. This difference was attributed to the negative surface charge on the AAPP substrate, which drives electrostatic adsorption of cationic surfactant and hinders adsorption of anionic surfactant. In Fig. 43 it can be seen that up to a surfactant concentration of ;0.02 mM the adsorption isotherms of CTAB and DTAB are indistinguishable. This indicates that the adsorption in the first region is independent of the hydrophobicity of the surfactant monomer, and is thus driven purely by Coulombic attractions between

294

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

Fig. 44. Adsorption isotherms for SDS in no added electrolyte (closed circles) and in 10 mM NaBr (open circles) on the AAPP substrate. The lines are drawn to guide the eye with regard to the trend in the data series. The dashed lines separate the regions of the adsorption isotherms. Note that region I is absent. Within experimental limitations, the isotherms are indistinguishable above 3 mM. 2003 ACS. Reproduced with permission from Ref. w114x.

the surfactants and the charged sites on the surface. In the DTABq10 mM NaBr system the surface excess is much lower in this region of the isotherm. We attribute this difference to screening of the surfactantsubstrate electrostatic interaction andy or reduction in the number of charged sites suitable for adsorption on the surface due to adsorption of Naq ions. Similar effects have been observed on silica for the adsorption of CPBr w107x. Note no measurable adsorption occurs for the SDS systems. That is, the first region is absent in Fig. 44. For all isotherms, further adsorption beyond region I takes place against a repulsive electrostatic interaction and therefore must be driven by hydrophobic interactions. The concentration which separates the first and second regions of the isotherm is analogous to the cac w115x for surfactantpolymer interactions in solution and is denoted the surface cac. In region II, all systems exhibit a linear increase of surface excess with the log of concentration above the surface cac. The slope of the isotherm in this region is indicative of the magnitude of the favourable hydrophobic surfacemonomer interactions. The slope of the CTAB isotherm exceeds that of the DTAB isotherm. This is a direct consequence of the greater hydrophobicity of the C16 vs. C12 hydrocarbon chain. Importantly, all the systems remain linear until the same surface excess of ;0.3 mg my2 is reached. Saturation of this adsorption process occurs at this concentration and strongly suggests that the arrangement of surfactants on the surface is the same in all systems. It is postulated that coverage of the substrate in this region is by random sequential adsorption. Adsorbed surfactant chains are confined to the plane of the surface by hydrophobic interactions whilst the head-groups may protrude from the surface in order to maximise hydration. For the cationic surfactants, the surfactants adsorb opposite a

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

295

surface charge, orientated with head-groups electrostatically bound to the charged surface group, and hydrocarbon chains adsorbed to the substrate. Between the charged substrate groups, surfactants are adsorbed with hydrocarbon chains against the hydrophobic surface, while the head-groups may be orientated towards solution. For SDS, it is anticipated that the surfactant will be adsorbed in the same manner as for DTAB and CTAB but lie between charged surface sites. The next sub-region occurs for surface excess values from 0.3 mg my2 up to the saturation level of coverage for each system (;0.5 mg my2 for C12 systems and 0.7 mg my2 for CTAB). The surface excess is now too large for all the surfactant monomers to be confined to a prostrate layer against the substrate; a more complex structure must exist. This region of the isotherm exhibits the greatest rate of increase in surface excess with increasing bulk concentration. This rapid increase is a consequence of cooperative hydrophobic interactions between adsorbed and adsorbing monomers. The end result of this process is the plateau region of the adsorption isotherm, and further increases in concentration have no effect on surface excess. The level of adsorption beyond the prostrate monolayer coverage achieved at the end of region II is not large (;40% of the total). The proposed structure is one of monomers adsorbed to the most hydrophobic patches of the prostrate monolayer. This will leave the more hydrophilic charged sites exposed to water. Thus the overall structure is one of a prostrate monolayer with a partially (;65%) complete second prostrate monolayer attached to it. The sparseness of the second layer ensures that head-group interactions are not important. Note the second layer should not be considered a distinct layer as it will be anchored in the prostrate layer and very incomplete. The disorder in the structure ensures that AFM images cannot be obtained. The times to obtain equilibrium for surfactant adsorption on the AAPP substrate were also reported. In regions II, III and IV of the isotherms when electrolyte is absent, the adsorption of both SDS and DTAB to the AAPP substrate typically required approximately 2500 s to achieve equilibrium. On the addition of electrolyte, these equilibration times were considerably reduced, to approximately 100 s. It is suggested that the presence of added salt permits more rapid coverage of the substrate by reducing the range of electrostatic repulsions between head-groups. Lengthening the hydrocarbon chain from 12 to 16 carbon atoms (i.e. DTAB to CTAB) resulted in a similar reduction in equilibration time. In this case, the reduction could only be due to the increased hydrophobicity of the monomer imparted by a longer tail-group, as electrostatic influences are not expected to be significantly different between DTAB and CTAB in the absence of electrolyte. The initial adsorption rates were investigated, and it was shown that CTAB had the fastest initial adsorption rate of the surfactants studied. This was also attributed to the greater hydrophobic driving force imparted by a longer tail-group. The adsorption rates for the DTAB, DTABq10 mM NaBr and SDSq10 mM NaBr systems were quite similar at all concentrations. The SDS no added electrolyte system had the lowest initial adsorption rate, due to the electrostatic repulsions between the substrate and the monomer.

296

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

6.4. Summary of reflectance observations Results from NR provide the most accurate measure of the thickness of the adsorbed layer of CTAB on silica, of approximately 3.5 nm, as well as a lateral size of discrete aggregates of approximately 9 nm. It was suggested that if these aggregates were micelle-like, then they must be strongly flattened. This result agrees well with push through distances measured using AFM, and the flattening of the aggregate is most likely due to the surface charge of the substrate. SAR on silica have been identified for CTAB and CPBr at normal (unadjusted) pH in the absence of electrolyte, for CTAC both in the presence and absence of added electrolyte, and for CTAB at pH 10 using OR. We propose that the nature of the surfactant arrangement in solution and on the surface at equilibrium determines if an SAR region is likely. An SAR can only occur if the packing constraints of a monomer in a surface aggregate is very much greater than the packing constraints in solution. In the absence of electrolyte, the equilibrium surface structure present is small micelle-like surface aggregates adsorbed on the surface, whereas in solution only monomers are present. It is this process of building order at the surface that gives rise to the slow kinetics. The presence of electrolyte increases the aggregation in solution and at the same time results in surface aggregates with less curvature, thus removing the mismatch in packing constraints between the bulk and the surface. The evolution of these structures were followed by AFM. As the SAR corresponds to the first concentration where aggregates are formed at the interface, the lower concentration boundary of the SAR is equivalent to the csac. Slow adsorption effects have not been elucidated in the presence of the electrolyte of a strongly binding coion i.e. the addition of 10 mM KBr to CTAB or CPBr solutions. Slow adsorption effects have been observed in the presence of the weakly binding chloride counterion. This supports a structural interpretation for the origin of slow adsorption. The gemini surfactants did not exhibit slow adsorption for any spacer size. For surfactants in the presence of the electrolyte of a strongly binding co-ion, equilibrium is rapidly reached at all surface excess values and increases in the surface excess above the cmc are noted. This is a result of a higher packing density of the surface aggregates and a change in aggregate morphology. Thus, for these systems, the csac is actually greater than the cmc. In these systems aggregates occur in bulk before they are present on the surface indicating that the energy cost of the geometrical constraints of the surface on the aggregate structure exceed the adsorption energy. In both the presence and absence of electrolyte, chloride surfactants exhibit similar features to bromide surfactants in the absence of electrolyte, due to the chloride ion having a much lower affinity for surfactant aggregates than the bromide ion. The decreased binding efficiency of the chloride ion results in the saturation surface excess of the chloride system being approximately 50% less than that of the corresponding bromide surfactant. For concentrations up to 10 mM, the identity of the electrolyte co-ion does not influence adsorption behaviour within the experimental limitations of OR.

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

297

For constant solution conditions, lengthening the hydrocarbon tail-group of the surfactant monomer leads to two important effects. Firstly, the entire adsorption isotherm is shifted to lower concentrations, in line with the reduction in solution cmc, and the width of the increasing region of the isotherm is decreased. Secondly, the saturation surface excess is increased for the longer chain surfactants. Calculation of the number of surfactant monomers per unit area reveals that this increase in surface excess is due to the presence of a greater number of molecules adsorbed at the interface, and not merely a result of the increased mass of the monomer units. For 12-s-12 gemini surfactants adsorbing at the silicaaqueous interface it has been shown that below 0.3= cmc, the surface excess is independent of the spacer size. At these concentrations attractive electrostatic interactions between the adsorbed surfactants and the substrate determine the level of adsorption. At increased surfactant concentrations further adsorption is due to hydrophobic interactions between adsorbed monomers. The saturation surface excess is a function of the spacer size, with the shortest spacer size producing the largest saturation surface excess. Whereas the equilibrium surface structures determined for cationic surfactants on silica have been shown to be elongated spherical structures, the adsorbed morphology of the surfactant layer on the AAPP substrate is a prostrate monolayer with a partially (;65%) complete second prostrate monolayer attached to it. The second layer is not a distinct layer as it will be anchored in the prostrate layer and very incomplete. CTAB has the highest saturation surface excess of the surfactants investigated on this substrate. However, the area per molecule data shows that approximately the same numbers of molecules are adsorbed for all surfactant substrate combinations at saturation and so the increase in surface excess is due to the increased mass of the CTAB monomer. On this substrate, DTAB and SDS have the same saturation surface excess both in the presence and absence of added 10 mM NaBr, despite the surface having a substantial negative charge. This demonstrates the dominance of hydrophobic interactions for surfactant adsorption onto hydrophobic substrates. Studies of surfactant adsorption kinetics give insight into the mechanism of surfactant adsorption. Comparison of the measured kinetics of adsorption (for cationic surfactants adsorbing to silica) to the theoretical diffusion limited flux of surfactant to the substrate has revealed a cooperative character in the adsorption process above the solution cmc. This cooperativity is due to the direct adsorption of micelles to the substrate. The kinetic aspects of gemini surfactant adsorption have been shown to be similar to that of conventional surfactants. The adsorption process is complete within seconds for all spacer sizes, at all concentrations. Thus the slightly slower micellar monomer exchange rates for gemini surfactants has little influence on the adsorption kinetics. For a constant head-group type, a longer tail-group will have a higher initial adsorption rate at any concentration, on both silica and the AAPP substrate. This is due to the increased hydrophobicity of the monomer providing a greater driving

298

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

force for adsorption at pre-aggregate concentrations. This increased hydrophobicity is evidenced by the fact that longer tail-group surfactants form micelles at lower concentrations. As micelles facilitate more effective substrate coverage, this leads to faster adsorption rates. The addition of electrolyte to surfactant solutions increases the rate of adsorption at a given concentration for both the silica and AAPP substrate. In addition to screening electrostatic interactions the addition of electrolyte lowers the solubility of the hydrocarbon chain, rendering the surfactant molecule more hydrophobic. This results in the adsorption rate being increased as described above. As the increase in the rate of adsorption is not dependent on the binding efficiency of the counterion to the surfactant, in the absence of specific interactions, the species of the electrolyte does not influence the rate of adsorption. In general, the kinetics of adsorption is slower on the AAPP substrate than for silica. This is most likely due to the difference in the adsorbed conformation on the two substrates. The mixed nature of the surface may also play a role, in that favourable electrostatic and hydrophobic interactions are possible on the AAPP surface but the strength of these interactions may be reduced. The adsorption of DTAB and SDS in the absence of electrolyte to the AAPP substrate is comparatively slow to reach equilibrium, but does not correspond to a SAR. For these systems, adsorption at all concentrations is slow, not just in a discrete portion of the isotherm. Secondly, the increase in surface excess with time does not correspond to a structural transition in the adsorbed aggregates. The role of the substrate in the pace of adsorption should not be underestimated. 7. Summary Adsorption isotherms are the traditional method for investigating surfactant adsorption phenomena, and they continue to yield much valuable information. By comparing isotherm data with the more recent structural information provided by techniques such as AFM, calorimetry, fluorescence probe studies, NR, and ellipsometry, changes in the interfacial conformation of the surfactant as a function of the solution concentration can now be determined. In a further advance, analysis of adsorbed aggregate structures in the light of new and detailed kinetic data indicates that it is the type of aggregates that form on the surface that determines both the surface excess and the rate of adsorption. 7.1. Mechanism of adsorption and the adsorption isotherm The new knowledge now available allows us to comment on the adsorption process with greater precision than ever before. This is particularly true for cationic surfactants on oxide surfaces. Adsorption is controlled by electrostatic and hydrophobic interactions. The relative influence of these interactions is determined by both the properties of the substrate and the nature and concentration of the surfactant. We divide the adsorption isotherm into three concentration spans. In each span a new adsorption process is possible. The spans are named on the basis of the

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

299

Fig. 45. Proposed Mechanism of surfactant adsorption. Each span is described in detail in Section 7.

adsorption mechanism that is newly available in that concentration range. They are, in increasing concentration, the electrostatic concentration span, the electrostatic hydrophobic concentration span and the hydrophobic concentration span. The mechanism of adsorption in each span differs and is depicted schematically in Fig. 45. Note that the hydrophobic concentration span may be further divided into above-

300

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

cmc and below-cmc spans to reflect the direct adsorption of micelles. We provide our description of the adsorption mechanisms in order of increasing concentration but this should not be taken to imply that the kinetics of adsorption operate in this manner. That is, adsorption at a concentration in, for example, the hydrophobic span does not imply that the adsorption process occurs sequentially through the electrostatic span, the electrostatichydrophobic span then the hydrophobic span. Rather in the hydrophobic span the mechanisms applicable to all regions may be operating simultaneously and at different rates. For example, as we have evidence that micelles adsorb directly to the surface, in this case the hydrophobic interactions are essentially in operation before the micelle approaches the surface, at which time the electrostatic mechanisms begin to operate. The mechanism we have assigned to each span is available to surfactant adsorption at greater concentrations but not at lesser concentrations. 7.1.1. The electrostatic concentration span In the first span surfactant molecules are electrostatically adsorbed to the charged surface sites. The presence of a positively charged head-group at the interface renders nearby hydroxyl groups more acidic, which induces more charged sites in the vicinity of the initial charge site. 7.1.2. The electrostatic and hydrophobic concentration span Surfactant tail-groups will interact with any hydrophobic regions that are present on the substrate. Regardless, the hydrophobic tails of the surfactant, along with the newly induced sites of surface charge, act as nucleation points for further surfactant adsorption. The adsorption is thus driven both by hydrophobic interactions and electrostatic attraction. Throughout the second span the charge on the underlying substrate continues to increase. At the end of this concentration span the adsorbed morphology is described as a teepee structure, the substrate ionisation is at a maximum and the overall surface charge is neutralised. 7.1.3. The hydrophobic concentration span Any further adsorption is purely hydrophobically driven, and will be against a repulsive electrostatic barrier that arises as a result of overcompensation of the surface charge by the adsorbed surfactant. Thus span three is characterised by the hydrophobic adsorption of surfactant molecules to the teepees present at the interface, with head-groups oriented away from the surface. In the third concentration span, the level of counterion adsorption becomes appreciable. The kinetic evidence evaluated in this review suggests that the explanation of the hydrophobic concentration span is relevant for all surfactant concentrations above charge neutralisation. The only variation in the adsorption mechanism is the direct adsorption of micelles above the solution cmc. Not all systems will exhibit different adsorption regions corresponding to the three concentration spans described. We have seen that for an anionic surfactant adsorbing to a partially hydrophobic anionic surface w114x, there is no electrostatic

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

301

adsorption, as such the first span is not present in this isotherm. Highly hydrophobic surfaces give a similar result. 7.2. Adsorption kinetics and the adsorption isotherm We have demonstrated that the adsorption process can continue over long periods due to kinetic trapping. The concentration range over which this occurs is called the SAR. The SAR spans concentrations that are sufficient to lead to aggregate levels of coverage on the substrate, but where no aggregates are present in solution and is a consequence of kinetic barriers to the formation of the thermodynamically stable arrangement of adsorbed surfactant. This phenomenon may yet prove to be widespread. We have seen that adsorption levels obtained on dilution exceed those that are reached when the surfactant concentration is increased in a step-wise manner. Further, the adsorption levels obtained by increasing the concentration in a stepwise fashion were less than those obtained when the concentration is increased directly, in one step. These differences exist despite the fact that in all cases the surface excess had stabilised and apparent equilibrium had been reached. The fact that stepwise introduction of surfactant led to values of surface excess less than those seen at the same concentration for stepwise reductions in surfactant concentrations can be interpreted as evidence that equilibrium has not been reached in the former case, and that this is due to kinetic trapping akin to that which gives rise to the SAR. However in this case the kinetic trapping is so severe that continued adsorption has ceased. Thus, equilibrium adsorption isotherms that are determined by stepwise changes in surfactant concentration should always be conducted under a dilution regime to prevent false equilibrium values being obtained. An assessment of the available literature on the adsorption of cationic surfactants to silica surfaces has led to a new understanding of the mechanisms operating during adsorption. The mechanisms available are determined by which concentration span of the adsorption isotherm the solution surfactant concentration lies in. The adsorbed surfactant morphology determines both the kinetics of adsorption and the surface excess. The influence of the kinetics of adsorption can be dramatic, leading to very slow adsorption processes and possibly complete kinetic trapping at surface excess values below the true equilibrium surface excess. References
w1 x w2 x w3 x w4 x w5 x w6 x w7 x w8 x A.M. Gaudin, D.W. Fuerstenau, Trans. AIME 202 (1955) 958. T.P. Goloub, L.K. Koopal, B.H. Bijsterbosch, M.P. Sidorova, Langmuir 12 (1996) 3188. J.L. Trompette, J. Zajac, E. Keh, S. Partyka, Langmuir 10 (1994) 812. M.R. Bohmer, L.K. Koopal, Langmuir 8 (1992) 2649. (a) R.K. Iler, The Chemistry of Silica, Wiley, 1979 (b) J.H. Bolt, J. Phys. Chem. 61 (1957) 1166. L.K. Koopal, T.P. Goloub, Surfactant adsorption and surface solubilization, in: R. Sharma (Ed.), ACS Symposium Series 615, vol. 78, American Chemical Society, Washington, DC, 1995. P. Wangnerud, G. Olofsson, J. Colloid Interf. Sci. 153 (1992) 392. T.P. Goloub, M.P. Sidorova, Kolloid Zh. 34 (1992) 17.

302

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304

w9x D.B. Hough, H.M. Rendall, Adsorption of ionic surfactants, in: G.D. Parfitt, C.H. Rochester (Eds.), Adsorption from Solutions at the SolidLiquid Interface, Academic Press, London, 1983. w10x B.H. Bijsterbosch, J. Colloid Interf. Sci. 47 (1974) 186. w11x H. Rupprecht, E. Ullmann, K. Thoma, Fortsch. Kolloid Polym. 55 (1971) 45. w12x H. Rupprecht, J. Pharm. Sci. 61 (1972) 700. w13x H. Rupprecht, T. Gu, Colloid Polym. Sci. 269 (1991) 506. w14x A. Thibaut, A.M. Misselyn-Bauduin, J. Grandjean, G. Broze, R. Jerome, Langmuir 16 (2002) 203. w15x Y. Gao, J. Du, T. Gu, J. Chem. Soc. Faraday Trans. 1 (83) (1987) 2671. w16x T. Gu, Z. Huang, Colloids Surf. 40 (1989) 71. w17x S. Manne, J.P. Cleveland, H.E. Gaub, G.D. Stucky, P.K. Hansma, Langmuir 10 (1994) 4409. w18x P. Somasundaran, D.W. Fuerstenau, J. Phys. Chem. 70 (1966) 90. w19x A. Fan, P. Somasundaran, N.J. Turro, Langmuir 13 (1997) 506. w20x P. Chandar, P. Somasundaran, N.J. Turro, J. Colloid Interf. Sci. 117 (1987) 148. w21x P. Chandar, P. Somasundaran, K.C. Waterman, N.J. Turro, J. Phys. Chem. 91 (1987) 148. w22x P. Somasundaran, C.V. Kunjappu, C.V. Kumar, N.J. Turro, J.K. Barton, Langmuir 5 (1989) 215. w23x T.P. Goloub, L.K. Koopal, Langmuir 13 (1997) 673. w24x A. De Keizer, M.R. Bohmer, T. Mehrian, L.K. Koopal, Colloids Surf. 51 (1990) 33. w25x M. Chorro, C. Chorro, O. Dolladille, S. Partyka, R. Zana, J. Colloid Interf. Sci. 210 (1999) 134. w26x J. Israelachvili, Intermolecular and Surface Forces, Part III, second ed., Academic Press, London, 1992. w27x Y.F. Deinega, Z. UlBerg, L. Marochko, V. Rudi, V. Denisenko, Kolloid Zh. 36 (1974) 649. w28x (a) F.M. Menger, C.A. Littau, J. Am. Chem. Soc. 115 (1993) 10083 (b) M. Dreja, S. Gramberg, B. Tieke, Chem. Commun. 13 (1998) 1371 (c) S. De, V.K. Aswel, P.S. Goyal, S. Bhattacharya, J. Phys. Chem. B 102 (1998) 6152. w29x R. Zana, in: I. Robb (Ed.), Specialist Surfactants, Chapman Hall Ltd, New York, 1996. w30x R. Zana, M. Benrraou, R. Rueff, Langmuir 7 (1991) 1072. w31x D. Danino, Y. Talmon, R. Zana, Langmuir 11 (1995) 1448. w32x R.J. Hunter, Foundations of Colloid Science, Oxford University Press, 1985. w33x M.J. Rosen, Chem. Tech. 23 (1993) 30. w34x R. Zana, Curr. Opin. Colloid Interf. Sci. 1 (1996) 566. w35x R. Zana, Y. Talmon, Nature 362 (1993) 228. w36x E. Alami, G. Beinert, P. Marie, R. Zana, Langmuir 9 (1993) 1465. w37x K. Esumi, M. Goino, Y. Koide, J. Colloid Interf. Sci. 183 (1996) 539. w38x K. Esumi, Y. Takeda, M. Goino, Y. Ishiduki, Y. Koide, Langmuir 13 (1997) 2585. w39x K. Esumi, S. Uda, M. Goino, et al., Langmuir 13 (1997) 2803. w40x M. Chorro, C. Chorro, O. Dolladille, S. Partyka, R. Zana, J. Colloid Interf. Sci. 199 (1998) 169. w41x L. Grosmaire, M. Chorro, C. Chorro, S. Partyka, R. Zana, J. Colloid Interf. Sci. 243 (2001) 525. w42x R. Zana, J. Colloid. Interf. 248 (2002) 203. w43x J. Leimbach, J. Sigg, H. Rupprecht, Colloids Surf. A 94 (1995). w44x L.S.C. Wan, J. Pharm. Sci. 55 (1966) 1395. w45x S. Grasvholt, J. Colloid Interf. Sci. 57 (1976) 575. w46x A.L. Underwood, E.W. Anacker, J. Colloid Interf. Sci. 106 (1985) 86. w47x H. Hoffmann, G. Platz, H. Rehage, W. Schorr, W. Ulbricht, Ber. Bunsen-Ges. 85 (1981) 255. w48x J. Leimbach, H. Rupprecht, Colloid Polym. Sci. 271 (1993) 307. w49x J. Seidel, Thermochim. Acta 229 (1993) 257. w50x J.L. Trompette, J. Zajac, E. Keh, S. Partyka, Langmuir 10 (1994) 812. w51x E. Bury, E. Souhalia, C. Treiner, J. Phys. Chem. B 95 (1991) 3824. w52x G.H. Findenegg, B. Pasucha, H. Strunk, Colloids Surf. 37 (1989) 223. w53x J.L. Kavanau, Water and Solute Water Interactions, Holden-Day, San Francisco, 1964. w54x H. Kihira, E. Matijevic, Langmuir 8 (1992) 2855. w55x L. Lajtar, J. Narkiewicz-Michalek, W. Rudzinski, S. Partyka, Langmuir 10 (1994) 3754.

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304 w56x w57x w58x w59x w60x w61x w62x w63x w64x w65x w66x w67x w68x w69x w70x w71x w72x w73x w74x w75x w76x w77x w78x w79x w80x w81x w82x w83x w84x w85x w86x w87x w88x w89x w90x w91x w92x w93x w94x w95x w96x w97x w98x w99x w100x w101x w102x w103x

303

M. Bouzerda, Ph.D. Thesis, CNRS Lab 330 Montpellier, 1991. M.J. Rosen, Surfactants and Interfacial Phenomena, Wiley, New York, 1978. Z. Kiraly, R.H.K. Borner, G.H. Findenegg, Langmuir 13 (1997) 3308. Z. Kiraly, G.H. Findenegg, J. Phys. Chem. 102 (1998) 1203. J. Zajac, J.L. Trompette, S. Partyka, Langmuir 12 (1996) 1357. H.N. Patrick, G.G. Warr, S. Manne, I.A. Aksay, Langmuir 13 (1997) 4349. S. Manne, H.E. Gaub, Science 270 (1995) 1480. S. Manne, T.E. Schaffer, Q. Huo, et al., Langmuir 13 (1997) 6382. A.M. Koganovkii, Colloid J. USSR 24 (1962) 702. E.J. Wanless, W.A. Ducker, J. Phys. Chem. B 100 (1996) 3207. E.J. Wanless, W.A. Ducker, Langmuir 13 (1997) 1463. H.N. Patrick, G.G. Warr, S. Manne, I.A. Aksay, Langmuir 13 (1997) 4349. L.M. Grant, F. Tiberg, W.A. Ducker, J. Phys. Chem. B 102 (1998) 4288. N.B. Holland, M. Ruegsegger, R.E. Marchant, Langmuir 14 (1998) 2790. W.A. Ducker, L.M. Grant, J. Phys. Chem. B 100 (1996) 11507. W.A. Ducker, E.J. Wanless, Langmuir 12 (1996) 5915. W.A. Ducker, E.J. Wanless, Langmuir 15 (1999) 160. S.B. Velegol, B.D. Fleming, S. Biggs, E.J. Wanless, R.D. Tilton, Langmuir 16 (2000) 2548. J.-F. Liu, G. Min, W.A. Ducker, Langmuir 17 (2001) 4895. V. Subramanian, W.A. Ducker, Langmuir 16 (2000) 4447. Y.L. Chen, S. Chen, C. Frank, J. Israelachvilli, J. Colloid Interf. Sci. 153 (1992) 244. R. Atkin, V.S.J. Craig, E.J. Wanless, S. Biggs, J. Phys. Chem. B 107 (2003) 2978. V. Subramanian, W.A. Ducker, Phys. Chem. B 105 (2001) 1389. W.J. Lokar, W.A. Ducker, Langmuir 18 (2002) 3167. L.M. Grant, T. Ederth, F. Tiberg, Langmuir 16 (2000) 2291. L.M. Grant, W.A. Ducker, J. Phys. Chem. B 101 (1997) 5337. J.L. Wolgemuth, R.K. Workman, S. Manne, Langmuir 16 (2000) 3077. R. Zana, in: R. Zana (Ed.), Surfactant Solutions: New Methods of Investigation, vol. 22, Marcel Dekker, New York, 1987, p. 241. M. Almgren, in: M. Gratzel, K. Kalyanasundaram (Eds.), Kinetics and Catalysis in Microheterogeneous Systems, Marcel Dekker, New York, 1991. C. Strom, P. Hansson, B. Jonsson, O. Soderman, Langmuir 16 (2000) 2469. P. Levitz, H.V. Damme, D. Keravis, J. Phys. Chem. 88 (1984) 2228. P. Levitz, H.V. Damme, J. Phys. Chem. 90 (1986) 1302. X.-L. Zhou, S.-H. Chen, Phys. Rep. 257 (1995) 223. A.R. Rennie, E.M. Lee, E.A. Simister, R.K. Thomas, Langmuir 6 (1990) 1031. G. Fragneto, R.K. Thomas, A.R. Rennie, J. Penfold, Langmuir 12 (1996) 6036. D.C. McDermott, J.R. Lu, E.M. Lee, R.K. Thomas, A.R. Rennie, Langmuir 8 (1992) 1204. D.C. McDermott, J. McCarney, R.K. Thomas, A.R. Rennie, J. Colloid Interf. Sci. 162 (1994) 304. J.C. Schulz, G.G. Warr, W.A. Hamilton, P.D. Butler, J. Phys. Chem. B 103 (1999) 11057. J.C. Schulz, G.G. Warr, P.D. Butler, W.A. Hamilton, Phys. Rev. E 63 (2001) 604. R.M.A. Azzam, N.M. Bashara, Ellipsometry and Polarised Light, North-Holland, Amsterdam, 1989. J.C. Dijt, M.A. Cohen Stuart, G.J. Fleer, Adv. Colloid Interf. Sci. 50 (1994) 79. P. Frantz, S. Granick, Phys. Rev. Lett. 66 (1992) 899. F. Tiberg, B. Jonsson, B. Lindman, Langmuir 10 (1994) 3714. J. Brinck, F. Tiberg, Langmuir 12 (1996) 5042. F. Tiberg, M. Landgreen, Langmuir 9 (1993) 927. K. Eskilsson, F. Tiberg, Macromolecules 30 (1997) 6323. T. Dabros, T.G.M. van de Ven, Colloid Polym. Sci. 261 (1983) 694. C. Geffroy, M.A. Cohen Stuart, K. Wong, B. Cabane, V. Bergeron, Langmuir 16 (2000) 6422.

304 w104x w105x w106x w107x w108x w109x w110x w111x w112x w113x w114x w115x

R. Atkin et al. / Advances in Colloid and Interface Science 103 (2003) 219304 K. Eskilsson, V.V. Yaminsky, Langmuir 14 (1998) 2444. E.S. Pagac, D.C. Prieve, R.D. Tilton, Langmuir 14 (1998) 2333. R. Atkin, V.S.J. Craig, S. Biggs, Langmuir 16 (2000) 9374. R. Atkin, V.S.J. Craig, S. Biggs, Langmuir 17 (2001) 6155. The cmc was determined from the breakpoint in the conductivity vs. concentration data. P. Lianos, R. Zana, J. Colloid Interf. Sci. 84 (1981) 100. L.J. Magid, Z. Han, G.G. Warr, M.A. Cassidy, P.D. Butler, W.A. Hamilton, J. Phys. Chem. B 101 (1997) 7919. B.D. Fleming, S. Biggs, E.J. Wanless, J. Phys. Chem. B 105 (2001) 9537. R. Atkin, V.S.J. Craig, E.J. Wanless, S. Biggs, J Colloid Interf. Sci, submitted for publication. R. Atkin, Ph.D. Thesis, University of Newcastle, 2002. R. Atkin, V.S.J. Craig, E.J. Wanless, P.G. Hartley, S. Biggs, Langmuir (2003) in press. K. Thalberg, B. Lindman, G. Karistrom, J. Phys. Chem. 94 (1990) 4289.

You might also like