Physical Adsorption in Condensed Phases : Introductory Paper BY D. H. EVERETT Received 5th May, 1975 In selecting topics for Faraday General Discussions one objective has been to identify an area of physical science in which new developments appear to be imminent, and so to set the stage for, and give impetus to, fresh advances. Looking back over the past seventy years one can pick out many examples of Faraday Discussions which have indeed proved to have been milestones in the evolution of a subject. It is also true that since many areas of science seem to evolve in a cyclic fashion, one may find several Discussions ten or twenty years apart which have returned to, and given new life, to a particular field of study. One may cite, for example, the electrolyte discussions of 1927,1957 and that planned for 1977; and those on liquids of 1936,1953 and 1967.The present Discussion, however, does not form part of such a sequence. In part at least this is because adsorption in liquid systems is relevant to a wide range of phenomena so that various aspects of the subject have formed part of many earlier Discussions; in particular the problems of ionic adsorption have been dealt with under the umbrella of the Electrical Double Layer.* Developments over the past few years, and foreseen in the near future, led the Colloid and Surface Chemistry Group of the Faraday Division to the view that a broadly based Discussion on the fundamental aspects of adsorption would be timely. Because of the wide range of topics which could have been included, it was decided to exclude work specifically concerned with the physical adsorption of gases by solids and to limit discussion to problems relevant to gas/liquid, liquid/liquid and liquid/ solid interfaces.In the event, papers in the first and third areas only were submitted, although many of the basic principles to be discussed here will be equally applicable to liquid/liquid interfaces. The preambles to several of the papers in this Discussion emphasise that work on adsorption in condensed systems is not only of considerable fundamental importance and interest in its own right, but is also very relevant in a wide range of problems in colloid science. It is clear indeed that the motivation for much of the work to be reported at this meeting stems from this consideration.The particular areas of interest relate first to vapour/liquid interfaces, which play a fundamental role in the formation and stability of foams, in controlling the drainage of liquid from or through porous media, and an understanding of which will contribute to knowledge of the properties of liquid/liquid interfaces of biological importance. Secondly, we shall be considering adsorption at liquid/solid interfaces, which determines among other things the properties of thin films, and the phenomena of lubrication and colloidal stability. The subject matter of this Discussion can alternatively be grouped according to the nature of the fluid phase: pure components, solutions of organic substances, surfactant and ionic aqueous solutions and polymer solutions.All of these, in their different ways, have considerable relevance to practical problems of industrial and environmental importance. * Trans. Faraday SOC., 1951, 47, 409-414 (abstracts only). 78 PHYSICAL ADSORPTION I N CONDENSED PHASES In this introductory paper I should like to outline how, in my view, the themes to be discussed are inter-related, and how work on them may be stimulated by this meet- ing. Let us start at the beginning and reflect on the fact that our understanding of the basic concept of surface tension is still lamentably incomplete 170 years after its importance was first realised by Young and Laplace. That surface tension was a manifestation of the asymmetry of intermolecular forces acting upon surface molecules was appreciated by Laplace.But in the absence of satisfactory theories of these forces-especially in dense systems-detailed understanding of the relationship between molecular structure and surface tension, both in pure liquids and solutions has proceeded only slowly. Following the elucidation of the nature of the forces between individual molecules by Debye, Keesom and London (which confirmed the empirical formulation of Mie later exploited by Lennard-Jones) the intermolecular energy of condensed systems was evaluated by pair-wise summation, and this procedure has formed the basis of theories of the bulk liquid state, and of Hamaker’s treatment of interparticle forces. This emphasis on energies at the molecular scale has earned for this theoretical approach the appellation ‘‘ the microscopic or Hamaker theory ”.In contrast, the alternative approach via quantum electrodynamics, pioneered by Lifshitz, has enabled the energies of molecular assemblies (regarded as continuous media) to be calculated from the electromagnetic properties of the individual bulk phases. This theory again has earlier roots in Maxwell’s work, and indeed current applications are often expressed in semi-classical rather than fully quantum mechanical terms : this approach is designated the “ macroscopic, or Lifshitz, theory”. The idea that electromagnetic properties, intermolecular forces and surface tension were closely linked has also a long history-as one example of this we may quote the correlations sought for, and found, between the parachor and molar refractivities.In recent years a certain rivalry between these two aspects of the problem has developed and it has been questioned whether they are compatible and whether it is justifiable to apply continuum theory at very close separations where the detailed molecular structure of one phase is “ seen ” by molecules in the other. It turns out, as Ninham and his co-workers have shown in their prolific work of the last few years, that the term “ continuum theory ” is not really justified, for the underlying ideas can also be applied to molecules which can be represented by a region of space having an inhomogeneous polarisability conveniently described by a Gaussian distribution whose width is a measure of the size of the molecule : this leads to the concept of the self-energy of a molecule.Ninham’s paper in this Discussion reviews some of the progress which has been made in this area. Although the specific applications outlined in this paper are concerned mainly with gas/solid interactions, we may expect further developments dealing with liquid/solid systems in more detail, especially if, as is foreshadowed, the theory will be essentially parameter-free. However, the continuum theory pays minimum attention to the statistical mechanical aspects of the distribution of molecules in interfacial regions. Yet the properties of interfaces are largely dependent on their structure in molecular terms. Unfortunately, there is little prospect of experimental data on surface properties providing more than broad indications of this structure. Thus the measured quantities such as surface tension or adsorption correspond to integrals of excess quantities taken through the interface: surface tension results from the distribution of excess tangential stress through the surface while the adsorption is the integral of the excess amount of matter in the interface.A given distribution of stress or matter leads unambiguously to a definite value for the surface tension or adsorption; but a singleD. H. EVERETT 9 measurement of either can tell us little about the distribution: in principle a given surface property can arise from different detailed distributions in the surface. Indeed in non-equilibrium surface states such a situation may arise in practice. It is only in the equilibrium state-where nature decides the distribution-that a one-to-one relation exists.Much effort has been devoted to the theoretical study of the distribution of matter and stress in the gas/liquid interface, using thermodynamic and statistical mechanical arguments. The simplest theories consider an abrupt step-function in density at the surface; more detailed theories deduce that the density falls monotonically from that of the liquid to that of the vapour over a few molecular diameters. More recently, following their successful application to the elucidation of the structure of bulk liquids, Monte Carlo and molecular dynamic techniques have been used to check other approaches. A major problem which emerges is that of deciding whether or not molecular layering (leading to an oscillatory density distribution function through the surface) occurs at the vapour/liquid interface: the papers by Rowlinson and his co-workers, and by Perram and White, address themselves, inter alia, to this problem.Findenegg and Fischer consider the similar problem for a fluid near a solid wall. It is clear that, whatever their detailed structure, “ surface layers” even in a simple system must extend in a direction normal to the surface for distances corres- ponding to several molecular diameters. As indicated above, this is to be expected on theoretical grounds, but is also in many instances a necessary qualitative deduction from experiment : analysis of adsorption measurements, using realistic estimates of molecular size, shows that the data cannot be accounted for without assuming multi- layer adsorption.Although much experimental work on adsorption from solution has been analysed in terms of models which assume that all composition changes occur within one molecular layer (i.e., monolayer theories) it has been known for some time that, except possibly for ideal solutions, such models are unrealistic and in most cases are thermodynamically inconsistent. Nevertheless they provide a convenient basis for the correlation of data, and as such have proved useful. But it is important to assess the validity of attempts to establish more realistic multi-layer models : several theories are available but comparison with experiment is not always easy because of the difficulty of knowing with adequate certainty the appropriate values of the various parameters involved.Furthermore, the body of accurate experimental data with which to test the theories is woefully inadequate. It is important, therefore, to have presented here Lane’s careful measurements and analysis of the properties of the vapour/liquid interface for the cyclohexane + carbon tetrachloride system. That we are still unable satisfactorily to account in detail for the surface properties of this system represents a major challenge. The papers from Edmonds and McLure and from Couper, Gladden and Ingram deal with rather more complex systems. The former authors provide interesting thermodynamic data which in due course will provide tests of more elaborate theories : in particular they show that, as with bulk properties, fluorocarbon + hydrocarbon mixtures exhibit some unusual surface properties.The systems studied by Couper et al. form micelles in the bulk solution, and measurements of surface tension provide a convenient and precise method of studying such phenomena. Similar types of system have been studied by Lucassen: here the measurement of surface properties has been applied to a study of the kinetics of micellar aggregation. These two papers illustrate the way in which studies of adsorption, through surface tension measurements, can be applied to other problems. Evidence that under certain conditions quite major structural changes can occur10 PHYSICAL ADSORPTION IN CONDENSED PHASES at the solid/liquid interface has been accumulating for several years and the paper from Brown et al. collects together and analyses information on alkane/carbon, alkanoll carbon and various solution/carbon interfaces.For pure liquids the adsorption data of Findenegg and the calorimetric data of Thorne lead to clear and qualitatively similar conclusions about the thickness of the structured layers of n-alkanes and n-alkanols at the graphite surface: these layers exhibit properties closely similar to the solid material. Similar conclusions result from measurements of adsorption by graphite from n-alkane and n-alkanol mixtures by Bown and by Brown. Perhaps the most significant feature of this work, which is closely related to later papers on polymer adsorption, is the dependence of these effects on the length of the hydrocarbon carbon chain. It is also interesting and important to discover that chain-branching, in at least one instance, eliminates the structuring, as does the substitution of a less well ordered (" non-graphitised ") surface for one consisting essentially of basal planes.Groszek's paper on the adsorption of condensed ring hydrocarbons by carbon surfaces exhibits similar features which are still not well understood. One might hope to be able to learn more about the interface from spectroscopic and similar techniques. Some of these (e.g., ellipsometry) while giving general information about the thickness of surface layers are of limited use, since to interpret the data it is first necessary to introduce a simple model of the interfacial structure, and what is deduced is a parameter dependent upon some average properties.N.M.R. methods using isotopically labelled compounds may prove more valuable since it may be possible to identify those groups in the molecule which are perturbed in the adsorbed state. But this technique is still being developed and, because of the small fraction of the molecules in the system which are in the interface, makes considerable demands on high sensitivity. Nevertheless, one may look forward to progress in the measurement and in the interpretation of n.m.r. data relating to surface phases. The use of infra-red spectroscopy, already widely used for gas/solid adsorption has only recently been extended to the solid/liquid interface. Marshall and Rochester's paper illustrates some of the possible applications of this technique: it is, of course, particularly useful (as with gas adsorption) for those systems involving silica where strong hydrogen bond interactions between the adsorbate and the adsorbent lead to substantial modifications of the infra-red spectrum of both the solid and adsorbate.The second half of the Discussion moves on to generally more complex problems, in which there is a strong interplay between forces of different kinds-dipolar and electrostatic forces being added to dispersion forces. Three groups of systems are represented here; first, those in which two dipolar species compete for adsorption on an electrically charged interface, and in doing so modify the potential drop across the surface layer ; secondly, those in which the solid surface itself can take part in ionic exchange or dissociation equilibria ; and, thirdly, those in which the adsorption of an ionic surfactant is influenced by pW.These papers are clearly important and relevant to practical problems involving oxide/aqueous solution interfaces. The concept of the formation of" hemimicelles " at a surface may be relevant to the structural changes at the graphite/alkane interface mentioned above. These pagers lead on naturally to the major problem of polymer adsorption, to consideration of the conformation of adsorbed polymers, and of the phase changes which they can undergo. Silberberg's paper introduces the general problems of polymer adsorption and directs attention to the distorting effect of a surface upon the segment distribution function and as a consequence on the centre-of-gravity con- centration function: it would be interesting in connection with studies of alkane adsorption to know to what extent the results for polymers may be applied to shorter chains.For example, the thickness of adsorbed layers of alkanes seems to be aD. H . EVERETT 11 linear function of chain length for n > 6 , while for polymers the thickness is pro- portional to the radius of gyration which varies with n*. Chan, Mitchell and White make use of the methods which have been so successful in the hands of Edwards and de Gennes for polymer solutions, in conjunction with the " continuum '' approach to intermolecular forces to show how a " phase change " in the adsorbed polymer can occur depending upon temperature or solvent composi- tion; Clark, Lal, Turpin and Richardson apply the Monte Carlo technique to the particular problem of the configuration of a polymer molecule, one end of which is terminally anchored, to throw further light on the effect of solvent/polymer interaction on conformation.One of the major experimental problems in the study of polymer adsorption arises from the slow attainment of an equilibrium conformation; and the even longer time needed to reach equilibrium on desorption. That adsorption proceeds in two steps- initial adsorption of one or more segments by collision with the surface, followed by rearrangement of the chain conformation-is investigated by Grant, Smith and Stromberg and reported here. In the final three papers we enter, I think, even more difficult areas. Protein adsorption even at the waterlair interface is a complex phenomenon, depending not only on the nature of the protein but also on the electrical state of the molecule.Until the simpler systems are understood, we shall be a long way from a realistic theory of the more general case of proteins at liquid/liquid interfaces and in membranes. The paper from two Unilever groups on casein adsorption provides some interesting data, some of which are not at first sight easy to understand. In particular, I antici- pate that some will wish to enquire more deeply into the apparent failure of the data to conform to the Gibbs adsorption equation. Studies of electrical double layers have played a major role in the elucidation of the colloidal stability of hydrophobic dispersions : and even though the problem has not been solved fully in simple systems, bolder experimentalists, recognising many practically important systems, have begun to examine the interplay between the structure of double layers and the adsorption of non-ionic polymers.Koopal and Lyklema have worked with one classical dispersion-silver iodide-while Kavanagh, Posner and Quirk have studied another-gibbsite. Both conclude that the polymers form relatively thick layers and that these layers can account for the influence of such polymers on colloidal properties. We may hope that from this Discussion will emerge a clarification of a range of problems, and an appreciation of the way in which further progress in others can usefully be pursued. But one must stress that the papers in this Discussion only take us part-way to the elucidation of the nature of interparticle forces and their dependence on adsorption phenomena.For all the papers in this Discussion are concerned with adsorption on a single surface-what is important in determining colloidal stability is the adsorption between two neighbouring surfaces. The basic problems involved require answers to the following questions. (i) Does the adsorption depend upon the distance between the surfaces? (ii) If so, how rapidly is adsorption equilibrium achieved when the separation changes? (iii) If the adsorption does not change (either because the equilibrium adsorption is insensitive to separation, or because the time constant for equilibration is too long in relation to the time scale of the process concerned), how do the forces between the adsorbed layers themselves influence the overall interparticle energy ? The first question is important because of the relationship between the effect of separation on the adsorption isotherm and interparticle energy as shown in separate thermodynamic treatments of Hall, and of Ash, Radke and myself.The relationship12 PHYSICAL ADSORPTION I N CONDENSED PHASES between the interparticle energy (Vp(c)) at a given separation (h) in a c-component solution, and that at the same separation in pure solvent (1) is linked to the change in the relative adsorption of the various species with separation and with bulk chemical potentials pf through the equation i=c Pul The primary information needed is that on the relative adsorption isotherm--I'i,l as a function of pi-at infinite separation and at separation h for each of the components.We have shown incidentally that this equation applies both to non-ionic systems and electrolyte solutions provided proper attention is paid to the electrical state of the surfaces: in the latter case application of a simple point charge model of ionic adsorption leads to equations essentially identical to the DLVO results. However, apart from this case, little theoretical or experimental attention has been paid to the evaluation of adsorption isotherms as a function of the separation between adsorbing surfaces; this I believe to be one of the more important tasks in the future. The problem of the kinetics of adsorption, too, needs further study, since in many instances the duration of an encounter between two particles undergoing Brownian motion may be too short for adsorption equilibrium to be achieved. In this case the interparticle forces are those corresponding to particles coated with a layer of adsorbed material-and this is a problem which has been tackled both in terms of the Hamaker approach by Vold, and of the Lifshitz approach by Ninham. However, even though the magnitude of the adsorption may not change with separation, the conformation of the adsorbed layers (especially if they are polymeric) may: the problem here is that of the so-called steric stabilisation upon which much work has already been done- but where a much deeper analysis still has to be attempted. In conclusion, it is clear that our Discussion will cover much ground-but that there are links between the various topics which evolve from apparently simple problems to which we may soon expect satisfactory answers, to very much more complex problems which represent a substantial challenge. One may hope that the stimulus given by this Discussion will lead to substantial progress in the future. Note that the paper by Edmonds and McLure, referred to by Prof. Everett on p. 9 and by Dr. Dickensoii on p. 89, was read and discussed at the meeting but was:not submitted for printing. Ed.