GENERAL DISC USSIO N Dr. W. van Megen and Dr. I. Snook (Melbourne) said: In connection with hydra- tion forces a few comments may be relevant: (1) Recent Monte Carlo (IvfC) calculations, of the structure of dense Lennard- Jones liquids adjacent to solid surfaces, reveal a pronounced stratification of the liquid’s density profile (fig. 1). This stratification extends over 4 to 6 molecular dia- 0 2 6 Z/G a 10 FIG. 1 .-Density profile for a Lennard-Jones fluid of reduced bulk density of 0.84 between two solid surfaces. The solid at the right is represented by a hard wall ; the solid at the left wall is represented by a 10-4 potential - and a 9-3 potential - - - - - - . See Steele’s book for details of these potentials. meters, depending on the precise form of the pair potential and the liquid’s bulk den- sity.It should not be unreasonable to assume that the range of the consequential force is of a similar magnitude to the range of the liquid’s stratification. Extrapolating our calculations on simple liquids to water, this would imply a hydration force of range around 2 nm. (2) Steele’ shows that the potential energy of a fluid molecule in the vicinity of a solid surface is sensitive to the structure of the solid and which crystal plane of the solid is exposed. Our MC calculations show that the liquid’s density profile is par- ticularly sensitive to the precise form of the fluid-solid potential. This seems to be in accord with your findings. Since in your experiments there is an interplay between four different kinds of forces which may even be coupled, we wonder whether it would be possible to repeat these experiments with apolar solvents ultimately to pinpoint these hydration or sol- vent structure forces more precisely? To take up Levine’s informal comment on the absence of a crystallized liquid layer in the vicinity of the solid-liquid interface we remark that radial distribution functions in thin rectangular panels parallel to and at various distances from the solid surface indicate that the liquid near the solid surface is not crystal like, but has a structure very similar to that of the corresponding bulk liquid.Fig. 2 shows these panel distri- bution functions (dots) at various distances from the solid surface compared with the radial distribution function for the corresponding liquid.W. A. Steele, The Interaction of Gases with Solid Surfaces (Pergamon Press, Oxford, 1974), chap. 2.44 GENERAL DISCUSSION 3 , FIG. 2.-Panel radial distribution functions for a Lennard-Jones fluid (bulk density of 0.84) at various distances from the solid surface, indicated by heavy dots, compared with the bulk radial distribution functions. (a) z = 0 . 9 8 ~ ~ (6) z = 2.810, (c) z = 5 . 6 8 ~ . Prof. J. Th. G. Overbeek (Utrecht) said: Could the repulsion that you have called " hydration forces " be due to silica chains or silica gel particles possibly formed by hydrolysis of the mica and strongly adhering to the mica surfaces? Prof. A. Silberberg (Rehovot) said: What is known about the mica surface in con- tact with water? Could the structure be different to what is supposed? Could, for example, some of the components of mica have become polymerized and cross-linked leading to a surface layer swollen by water and actually representing a gel.Under such circumstances repulsion at 1-7 nm separation would not be too hard to under- stand. Dr. S. Levine (Manchester) said : Any firm conclusion from experiment on a force between charged mica surfaces other than the usual van der Waals and electric double layer forces depends on an adequate theory of the electric double layer. At concen- trations < of 1-1 electrolyte, the Poisson-Boltzmann (PB) equation, the basis of classical theory of double layer forces, is probably reasonable in aqueous systems, but in the range L10-2 mol dm-3, particularly at higher surface potentials, this equation becomes increasingly unreliable as the electrolyte concentration is increased.In the case of 2-2 electrolytes, the inadequacy of the PB equation occurs at smaller concen- trations and this failure is even more pronounced with unsymmetrical electrolytes. One of the corrections to the PB equation is due to the variation of dielectric constant with distance from the plate walls and the conjecture about hydration in this paper implies that the dielectric constant in the diffuse layer is reduced significantly over a distance of about 6 nm, whereas I would suggest a figure of 1 nm as more reasonable. It must beGENERAL DISCUSSION 45 stressed that there are a number of modifications to the PB equation, of which variation in dielectric constant is only one and not necessarily the most important correction (e.g.self-atmosphere-image and ion-size effects), and because of possible partial com- pensation, it is necessary to consider all the corrections simultaneously. Dr. J. W. White (Grenoble said: I should be glad if you would say more about the experimental characteristics of the hydration forces mentioned in your paper in rela- tion to structural models for water at alumino-silicate interfaces. The essential thrust of the diffusion coefficient measurements made by Olejnik and White [ref. (19) of your paper] was that the water is remarkably unstructured, at least from the point of view of dynamical measurements. Dr. E. J. W. Verwey (Utrecht) said: In the 1930's colloid chemistry developed from a collection of" schools " into an integrated science.The work of Frumkin, connect- ing colloid chemistry and electrochemistry, the experiments of Derjaguin on " thin layers ", and the early calculations of Levine on the interaction of double layers were among the first indications of this integration process. It was officially recognized by the Faraday Society when it organized a General Discussion on the Electrical Double Layer, which was intended to be held in September 1939 in Cambridge but could not take place owing to the outbreak of the war. Although I have not been active in colloid chemistry in the last 33 years (we wrote our book in the winter of 1944-1945) I have always been interested in its interactions with other fields. After the war colloid chemistry has been generally accepted as part of physical chemistry but it is also clear that our understanding of lyophilic colloids is becoming of increasing importance for biochemistry.Israelachvili's paper and especially his observation of " hydration repulsion " (and similar indications in other papers at this Discussion) induces me to suggest a new aspect of this link with biochemistry. I am referring to " hydrophobic inter- action ", which plays an important part in for instance the attraction between the hydrophobic sidechains in the folding process of proteins. The nature of these forces is not quite clear, although we know that the gain in free energy is mostly of an entropic nature. This must be interpreted as an increased amount of order of the water molecules in the neighbourhood of the hydrophobic substance.The " hydration repulsion " between two mica plates at very short distance is obviously connected with a similar phenomenon : an increased " crystallisation " close to the mica surface. In this special case the effect may be reinforced by a paralel- lism in the structure of mica and ice. The more general nature of the phenomenon of the increased amount of order close to a " hydrophobic " wall suggests, however, that there must be another reason for this effect, especially if one considers that at thefree surface of water the situation is the other way round. Here there is more disorder in comparison to the bulk phase (surface entropy is positive). A possible explanation is that the van der Waals-London forces between the hydro- phobic " wall " and the water molecules have the general effect of increasing the order of the latter, but a more detailed investigation of the thin layer in contact with different materials using the methods indicated in Israelachvili's paper may give more informa- tion and create a new link with biochemistry.Dr. J. N. Israelachvili (Canberra) said: Many of the queries raised concerning our experimental results and their interpretation are discussed in some detail in a paper by G. Adams and myse1f.l J. N. Israelachvili and G. E. Adams, J.C.S. Faraday I, 1978,74, 975.46 GENERAL DISCUSSION The possibility that the repulsive “ hydration ” forces are due to the steric repul- sion of solvated polymeric silica chains (or some other polymeric chains) protruding from the mica surfaces cannot be excluded.However, I consider this unlikely for the following reasons : First, the aqueous region between the mica surfaces has the same refractive index as the bulk even when strong “ hydration ” forces are present, and even after the mica surfaces have been in water for more than one day.l Further, the attractive van der Waals forces do not appear to be in any way affected or modified by the presence of these forces. These findings allow us to conclude that the water adjacent to mica sur- faces is not the highly contaminated “ anomalous ” water of bygone days, and that the surface concentration of silica polymers, if they exist, must be very low, and yet sufficient sometimes to produce fairly strong repulsive f0rces.l Second, the hysteresis effects, discussed below, are difficult to interpret in terms of a dilute surface concentration of polymeric chains, unless these are somehow able to control the position of the OHP.Third, the “ hydration” forces are only weakly dependent on ionic strength and pH-an observation that also appears difficult to reconcile with solvated polymer- stabilized interactions. Fourth, a number of experimental studies on the self-diffusion of water, the viscous properties of water, and the heats of immersion of water2 at silicate and other clay mineral surfaces as well as at the ice-water interface have indicated behaviour different from bulk water decaying exponentially away from the surfaces with decay lengths close to 1.0 mn. Since the exponentially decaying “ hydration ” forces were also found to have a decay length of -1.0 nm it is reasonable to conclude that the same phenomenon was being observed in all these different studies.This being so, poly- meric silica chains would have to be invoked to account for some of these observations as well. Also related to this matter is the widely different swelling properties of clays such as the montmorillonites, where the amount of swelling in water has been found to be correlated with the crystallographic b-dimensions of the surface lattices, which closely match that of ice.3 In answer to Levine’s query, we analysed more than 50 (force, distance) curves before concluding that the “ hydration ” forces represented an additional force and not a modification of the double-layer forces.Of course, any theoretical insight as to how the effects Levine mentioned affect the double-layer forces will be welcome. Silberberg asked what is known of the mica-water interface. The mica surface at the cleavage plane is chemically inert in aqueous electrolyte solutions in the normal pH range of the experiments (pH 6-7), and we have verified that our micas do not hydrolyse and that, as expected for micas, they do not swell over the time period of the experiments, about 1 day [see also ref. (l), appendix 11. In answer to Hills’ informal remarks, I think it is important to remember that not only is the mica surface a solid surface, in contrast to the free water surface and the mercury surface, but that the mica surface is also an oxide surface to which water molecules can hydrogen bond.There is also the additional crystallographic simi- larity between the mica and ice lattice which could favour epitaxy [see also ref. (l), section 41. Details will be found in ref. (1). J. N. Israelachivli and G , E. Adams, J.C.S. Furuday I, 1978,74, 975. E. Nyilas, T. H. Chiu, D. M. Lederman and F. J. Micale, Colloid and Interface Science (Aca- demic Press, 1976)) vol. V. 111, p. 471. I. Ravina and P. F. Low, Clays and Clay Minerals, 1972,20, 109.GENERAL DISCUSSION 47 Verwey’s suggestion that “ hydration ” forces may be related to hydrophobic interactions, and of their importance in biological systems is, indeed, the subject of some current theoretical interest.l Finally, I should like to mention that any question like “ How far does the hydra- tion force extend?” is not easy to answer.These forces decay exponentially with distance, and do not stop abruptly at some separation. All that can be said is that in our experiments their effect was negligible (compared to the double-layer and van der Waals forces) at separations above 7.5 nm. Clearly, more work has to be done, e.g., with different solvents, at different temperatures, etc., before we can say more about the origin and nature of these forces. Prof. J. Lyklema (Wageningen) said: The issue whether or not hydration of sur- faces contributes to a repulsive force between colloid particles or between macrocopic bodies is still drawing much attention. The main difficulty is that usually several forces are simultaneously operative.Hydration forces tend to be invoked as contributing to the interaction if the sum of all other forces is inadequate to explain experimental observations. In doing so, imperfections in the theoretical picture of the “non- hydration ” forces are also interpreted as ‘‘ hydration ” forces. Israelachvili has done very elegant measurements, and he concludes that not all his results are explainable in terms of van der Waals attraction and double layer repulsion. As these experiments belong to the best that is available, it is a pity that the author has dubbed the unexplained extra force “ hydration force ” without investigating whether it is really due to hydration and not to some other phenomenon. The only fact that seems to be established is that this force is related to the surface properties of the mica.In fact, several properties of this additional force could be well “ explained ” by assuming that there are polysilicates extending from the mica surface into the solu- tion: (i) the number and extension of such polymers would depend on the process of cleaving and hence be different for different mica sheets (ii) the decay length of any steric repulsion, induced by such polymers could well be of the same order as that observed experimentally and (iii) the hysteresis phenomenon can be attributed to an irreversible reconformation of these chains. The main point I wish to make is the recommendation that the term “ hydration force ” be restricted to those forces where it is unambiguously established that hydra- tion is the sole responsible factor.Dr. J. N. Psraelachvili (Canberra) said: I agree with Lyklema that the use of the For the time being it would be term “ hydration ” forces was possibly premature. better to refer to them as “ additional ” forces. Prof. R. H. Ottewill (Bristol) said: Would you please outline the mechanism by which a mica surface acquires a charge and hence a surface potential. Were the surface potential values quoted in the paper obtained via the force measurements, or by an independent method? If the former, have you compared the results with those obtained by an electrokinetic experiment ? Would you also please comment on the nature of the hysteresis effects which you observed with the mica plates and provide details of the time scale of the experiment.Was this on a very short time scale, i.e., of the order of relaxation time for ions in the double layer, say s or less, or on a long time scale, say hours? I feel certain that the origin of the hysteresis observed in your experiments is quite D. Chan, D. J. Mitchell, B. W. Ninham and B. Pailthorpe, in Water, a Comprehensive Treatise, ed. F. Franks (Plenum Press, London, 1978), vol. 6, in press.48 GENERAL DISCUSSION different from that which we observed with montmori1lonite.l The explanation which we gave for this effect in terms of the rearrangement of a random array of plates to an ordered arrangement has been confirmed recently by neutron scattering experiments. Moreover, the fact that we observed the maximum effect in the most dilute electro- lyte systems would also appear to be in agreement with our interpretation.Dr. J. N. Israelachvili (Canberra) said: The effective surface potentials of the double-layer forces were obtained from the measured forces. No independent mea- surements of the potentials have yet been made, but this is being planned. I do not know by which " mechanism " the surface potentials or charge densities are estab- lished. In section 3 of my paper with Adams,2 a table is given of the chemical com- position of some of the micas used as determined by electron microprobe analysis, and the surface potentials as determined from the forces. It is noteworthy that those micas which exhibited low surface potentials had less Na and more K, though the amounts of Al, Fe and Mg also differed significantly.Hysteresis effects (or irreversible effects) were usually very small or non-existent, but on occasions large enough to be studied in some detail [see ref. (2), fig. 91. The distmce D/nm FIG. 1.-Schematic illustration of the variation of forces P with distance D showing the effects of hysteresis for two curved mica surfaces of radius R . The corresponding pressure P for two planar surfaces, given by the Derjaguin approximation P = a(F/bR)laD, is shown in the inset. - - -, maximum theoretical double layer force; -@-, first approach; -0-, subsequent approaches. general features are illustrated in fig. 1, in which the double-layer forces measured on the first approach in a given aqueous solution, e.g., measured -1 h after raising the concentration from to mol dm-3, were larger than those measured subse- quently, say -1 h to -15 h later (the measurements themselves take about 1 min per datum point). The exponential decay lengths of the repulsive forces were invari- ably the same on the first and subsequent approaches, and close to the theoretical Debye lengths of the electrolyte solutions.However, the magnitude of the forces on the first approach was greater than allowed for by double-layer theory (even assuming an infinite surface potential or charge density). For this reason I concluded that the hysteresis arises from an irreversible inward shift of the OHPs, in addition to any I. C . Callaghan and R. H. Ottewill, Furuday Disc. Chem. Soc., 1974,§7, 110. J. N. Israelachvili and G. E. Adams, J.C.S. Faruday I, 1978,74,975.GENERAL DISCUSSION 49 possible reduction of the potential or charge, during the first (forced) approach of the surfaces.By applying the Derjaguin approximation to the results for the force F against distance D for the curved mica surfaces one may obtain the corresponding pressure P against distance D for two planar surfaces (see inset, fig. 1). The " transition" region, wherein the OHP are presumably being displaced, shows up as a " plateau " region for the planar configuration at pressures usually close to P z 0.1 atm. This suggests the possible existence of a two-dimensional phase transition mechanism. The forces themselves, however, are still reversible in this " transition " region; the surfaces have to be brought much closer (closer than 5 nrn) before the forces become irreversible (hysteretic).Finally, hysteresis effects were generally accompanied by large " hydration " or '' additional " forces at smaller separations. Dr. J. W. White (Grenoble) said: My second question concerns the forces between alumino-silicate surfaces at low separations. Our original work using neutron diffrac- tion, and more recent work at very high energy resolution with instruments at the Institut Laue-Langevin in Grenoble, show that the diffusion of water in the crystalline swelling region of clays is probably different in kind from that in the osmotic region of swelling. This implies a quite different structure for the water in this region. Dr. J. Visser ( Vlaardingen) said : (1) How general are your observations; in other words, how far can your results be applied to other systems, such as polystyrene latices? (2) You do not mention the temperature at which you did your experiments. I suppose they were done at room temperature.What kind of effect may one expect at other temperatures, let's say at 40 "C and at 50 "C, particularly as regards the hydra- tion effects observed? (3) In all your measurements, you stayed away from the i.e.p.; would you please tell us why? Would you not get more information by measuring at that point in the sense that you reduce electrostatic contributions to the interactions ? (4) Suppose you achieved contact; is all the water then squeezed out from be- tween the adherents or not? (5) To what extent does surface deformation either account for your resuIts, or interfere with the interpretations ? Dr.J. N. Israelachvili (Canberra) said: We have yet to do more work on the ad- hesion forces and on the interactions at separations below 1.5 nm. What has so far emerged is that the forces that ultimately determine the adhesion are already evident at separations of -1 nm. Thus when the adhesion forces are strong the surfaces come into contact (in a primary minimum) at a separation of 0.0 & 0.4 nm relative to contact in uncleaved mica. This would allow for no more than a monolayer of water between the surfaces when in strong adhesive contact. In cases where the adhesive forces were weaker we noted that contact occurred a few Bngstroms further out. I am not sure what the i.e.p. of mica at the cleavage surface is, probably close to pH E 3. At pH NN 3 the adhesive forces are always much stronger (by 2 orders of magnitude or more) than at pH M 6-7, and the surfaces tend to tear up on separating them from contact.For this reason we stayed away from the i.e.p. All the experiments were carried out in a temperature controlled room at 20-22 "C. I do not know what effect changing the temperature would have. We shall try it. Surface deformation (flattening) occurs when the curved mica surfaces are brought50 GENERAL DISCUSSION close together against very strong repulsive forces. Such deformations were negligible at separations above 1.5 nm (see my paper with Adams for more details). As regards the applicability of our results to other systems, I would say that since the measured double layer and van der Waals forces are in good agreement with theory we may expect these results to apply more generally to other systems.At small separations, however, the forces often deviated drastically from theoretical ex- pectations and here the results, and the conclusions drawn from them, may not be generally applicable to other systems. Dr. B. Vincent (Bristol) said: I am wondering if the apparently very large values Lyklema reports for the " equilibrium " separation in the thin films stabilised by PVA might be due to multilayer adsorption at the aqueous/air interface. We have evidence' for such multilayers in the case of PVA adsorbed at the aqueouslpolystyrene interface, at the equilibrium PVA concentrations you have used (400 and 1400 p.p.m.), from both adsorption isotherms and adsorbed layer thickness (hydrodynamic tech- nique) measurements.Maybe such multilayer adsorption would not be so apparent from ellipsometric measurements since this technique relies on a significant refractive index increment between the adsorbed layer and the bulk solution. Although the first layer may have a reasonable refractive index increment, it could be that in the second (and subsequent) layers the coils are hardly distorted from their solution con- formations, and that therefore the refractive index increment relative to the bulk solu- tion is much less here. I would also query the shape of the molecular weight distribution assumed for the sample of PVA used by the authors. They have assumed a distribution which " tails " towards high molecular weights.We have carried out a g.p.c. analysis2 of a similar sample of PVA (admittedly not from the same suppliers, but also 88% hydrolysed poly(viny1 acetate) and of similar average molecular weight, 45 000). The distribu- tion tails in fact to low molecular weights, with a sharp cut off at high molecular weights (-67 000). Dr. Th. F. Tadros (Jealott's Hill) said: The " equilibrium " film thickness of PVA obtained by the authors for a sample of HviSc = 42 500 extrapolated to Ph = 0 is of the order of 70-80 nm. Such a value is a little higher than twice the hydrodynamic layer thickness (6) of a PVA sample of similar m.w. (Rw = 45 000) and similar acetate content, on polystyrene latex particles (330 nm diameter); 6 = 32.7 -i- 3.0 nm.3 Moreover, we have also shown that the adsorbed layer thickness increases as the particle radius increases and it depends on the surface properties of the latex partic1es.l Thus, these thick PVA films are to be expected and I do not feel there is a need to in- voke the presence of tails to explain the results, particularly if one realises that multi- layer absorption of PVA is a possibility.2 Most likely, at the plateau of the adsorption isotherm, the polymer coil rearranges and becomes more elongated normal to the interface (the volume occupied per molecule remains the same as in bulk solution, but the thickness of the adsorbed layer is greater than the diameter of the hydrodynamic sphere in solution).Prof. J. Lyklema and Dr. T. van Vliet ( Wageningen) said : For a number of reasons we rejected the idea of multilayer formation. In the first place it must be noted that Th.van den Boomgaard, T. A. King, Th. F. Tadros, H. Tang and B. Vincent, J. Colloid Inter- face Sci., 1978, 66, 68. M. J. Garvey, Th. F. Tadros and B. Vincent, J. Colloid Interface Sci., 1974,49, 57. M. J. Garvey, Th. F. Tadros and B. Vincent, J. Colloid Interface Sci., 1976,55, 440.GENERAL DISCUSSION 51 hydrodynamic thicknesses, measured at the PS/solution interface are at best circum- stantial evidence that in other systems multilayer adsorption can occur. More perti- nent are adsorption studies at GL boundaries, from which no indication for multi- layer formation could be 0btained.l More direct evidence stems from a number of experiments in which, after attain- ment of equilibrium, the pressure was reduced.Within the, admittedly not minor, range of error the same thickness was then found as when this film was made directly at the final pressure. The time scale to attain this new equilibrium state was of the order of an hour, which does not seem enough for polymer molecules to diffuse from the bulk through the film to the surface to establish a multilayer. Hence, we conclude that only convective liquid flow takes place. A third, more indirect indication is that a tenfold increase in concentration gives rise to not more than 15% increase of h at given Fs (our fig. 2). If multilayer adsorp- tion occurred, a much stronger dependence would be expected. Regarding the MW distribution, GPC analysis of PVA samples very similar or identical to ours and from the same supplier showed the distribution to approach the “ most probable ” one2 and no indication for a sharp shut-off, as found by Garvey et al.was obtained. Apparently this cut-off is typical for the specific sample used. Prof. A. Silberberg (Rehouot) said: It is my impression that the curves of fig. 1 in the paper are essentially osmotic dilution curves of the polymer solution trapped between the adsorbed layers of the film. Referring to the experiments, it is clear that due to the surface curvature at points a ” the pressure pa will be less than pat,,, and (pa - patmos) will approximately equal the applied suction. Suction not only causes partial draining of the fluid layer but also a reduction in the polymer concentration in the core region of the film.Polymer con- centration will drop until the osmotic pressure difference between the core of the film and the bulk polymer solution (which is in tube T) balances (patmos - pa). Convec- tion through the film will continue until the bubble b has disappeared. During this time there is a pressure difference pb - pa applied where p b > patmos > pa. The final stages of equilibration probably involve diffusion of polymer from within the core layer of the film to the bulk supply. The (overall) thickness corresponding to two adsorbed layers with an effectively zero thickness core layer, i.e., zero polymer con- centration between the adsorbed layers corresponds to p b = pa - patmos = n, where 7r is that part of the osmotic pressure of the solution which is due to the polymer solute.On the other hand, when Ph = 0 ( p a = patmos), the film thickness will tend to corre- spond to that of two adsorbed layers plus that of a core layer, at least one undistorted polymer coil diameter thick. We see that for PVA 205 (400 p.p.m.) for example h N 80 nm whenp, = 0 and h N 40 nm whenp, = 50 N m-2. The film thickness 80 nm atp, = 0 agrees reasonably with two layers of 20 nm each and a coil diameter of 40 nm, whereas h = 40 nm at Ph 21 50 N m-2 corresponds to two touching adsorbed films. A pressureph 51: 50 N mV2 is a reasonable osmotic pressure for a 400 p.p.m. PVA 205 polymer solution. The data for the other polymers show similar behaviour. The agreement is approximate but is close enough to what may be expected from experimental accuracy.I do not believe that the explanation offered by the authors can be right. The exponential distribution (whether of tails or of other segments) has been calculated against an (essentially) zero concentration bulk solution. A different case arises 6 6 J. M. G. Lankveld and J. Lyklema, J. Colloid Interface Sci., 1972, 41,466. B. J. R. Scholtens, Meded. Landbouwhogeschool, Wageningen, Neth., 1977, 77, 7.52 GENERAL DISCUSSION when we deal with systems at finite concentration. The adsorbed layers if brought close to each other would tend to equalize concentrations and the few long tails would collapse or penetrate the other side. In calculations of the stable separation distance between two polymer molecules, Flory and Krigbaum considered the free energy minimum which is a compromise between strong overlap (a higher free energy of mixing and a lower free energy of distortion) and small overlap (a lower free energy of mixing and a higher free energy of distortion).In the case of the approach of two colloid surfaces, irreversibly coated by polymers, similar problems arise, but different considerations apply. As Hesselink, Vrij and Overbeek have shown when the polymer bulk concentra- tion is zero it is the osmotic (mixing term) repulsion which dominates. At finite concentration, the situation is more complex. If the polymer film is in equilibrium with polymer in solution the case discussed in my paper at this Discussion is approached, If adsorption is irreversible, there will be a tendency for coated colloid particles and polymer in solution to mix as though we had two molecular weight species of the same polymer in solution.In this case, in addition, the coated colloid particle is structurally different from the polymer, and long range van der Waals forces arise and may influence interaction between the colloid particles. Lyklema and van Vliet indeed assume that polymer is irreversibly adsorbed, but that in other respects their films reach equilibrium. More than a possibility exists, however, that in the draining of the film more polymer molecules than required for equilibrium were trapped and that their presence enhances the film thickness even more. " Trapping " would be most effective since the diffusion coefficient of such polymer molecules must be expected to be extremely low.h t "T" pa I Prof. J. Lyklema and Dr. T. van Vliet ( Wageningen) said: Silberberg raises a num- ber of points regarding the interpretation of equilibrium film thicknesses in polymer stabilized films. The main problem appears to be whether or not there are polymer molecules in the film between the two adsorbed sheaths. A similar question has been asked by Tadros and by Vincent. The presence of such molecules would require considerable modification of the theory by Hesselink et al. In our (semi quantitative) examples of computations we neglected any polymer remaining within the film after establishment of the stationary state. The main arguments for this decision are: 1. The reversibility of Fs as a function of pressure, i.e., as a function of h: a given Fs(h) situation can be obtained both from the side of higher Fs reducing the pressure or from the lower side by increasing it.The time to attain the new equilibrium thick- ness is of the order of an hour and apparently too fast to be attributable to the diffu- sion of polymer molecules. Our conclusion is that the rate of adjustment of the newGENERAL DISCUSSION 53 situation is determined by convective flow, so that, apart from the adsorbed polymer, the interior of the film must have roughly the same composition as the bulk. 2. Accepting that there is some polymer between adsorbed layers, we inferred from the fact that a tenfold concentration increase (from 400 to 4000 p.p.m.) increases h at fixed pressure by only z lo%, that the contribution of any molecules between the adsorbed layers is small and we decided to neglect it in this stage of the work. Some additional comments that can be made in this connection are: (i) the chemi- cal potential of the water is mainly determined by the glycerol (1 molar) hence there is no reason to invoke polymer concentration changes as the most likely way to counteract hydrostatic (or, for that matter, capillary) pressures and (ii) drainage of the film pro- ceeds by marginal regeneration and/or by the sucking into the border of thicker regions.In other words, it is, at least in part, of a convective rather than of a diffusive nature. This may be one of the major differences with the measurements by Cain et al. Dr. P. Walstra ( Wageningen) (communicated) : The system PVA 205 at 400 p.p.m.of Lyklema and the PVA 88/10 at 0.2 g/lOO cm3 of Cain appear to be quite similar. But in the comparable range of distances (h = 50-60 nm) the film pressures measured by both groups differ almost exactly by a factor 1000, Cain’s pressures being higher. How can this be explained? Prof. J. Lyklema and Dr. T. van Vliet (Wageningen) said: One of the factors that may be responsible for the different pressures at given range of interaction between measurements by Cain et al. and ours is the amount r of PVA adsorbed. In our case T - 3.1 mg m-2 while from the data by Cain et al. it can be inferred that at a bulk concentration of 0.5 g cm-3 Tis of the order of 20 mg m-2. This difference, in turn, can be due to the quite different mechanisms by which the films were formed in the two cases.As outlined in the discussion remark by Lyklema to Cain et al., the history of the film plays an important role. In particular, the different surface rigidities in the two cases deserve attention. For rigid films more experimental information is clearly needed to settle this point. Apparent stationary state thicknesses may be found that do not represent true equilibria. They are due to the very slow drainage of liquid from the film. Dr. J. B. Smitham (Bristol) said: No entirely satisfactory explanation can be offered at this time. We have examined our calibration procedure closely with special atten- tion being given to the measurement of the applied force, the area over which the force acts and the equating of the applied force to the force of steric repulsion.Direct application of forces comparable with those obtained by our earlier extra- polation procedure confirms that the forces are approximately equivalent and accurate to within a factor of two or three. The error is greatest at very low applied forces. At separations of order 100 nm, the force is x4 x N and this increases to 2.5 x For the PVA films the first detectable interactions occur at separations around 100 nm (fig. 5 of our paper). Only a slight change of shape of the FECO fringes is observed and it is not until the separation has become closer to 75 nm that the fringes show a marked flattening, corresponding to a flat thin film area. Thus in the region 75-100 nm the area over which the interaction occurs is less well defined.An additional criticism that has been made of the PMMA surfaces is that their surface roughness could increase the area over which the force is applied. However, electron micrographs of N at separations near 60 nm. The area over which the force is applied is subject to some uncertainty.54 GENERAL DISCUSSION shadowed carbon replicas of the surface did not reveal any gross defects of the surface. This was strongly confirmed by the nature of the FECO fringes which faithfully repro- duced any irregularities in the contact area. Moreoever, an approximate estimate of the roughness can be made by calculating the distance corresponding to half the width of a fringe. This is an overestimate because the fringe width is determined by the condition of the silver layer and not only by the smoothness of the transparent over- layers, whether they be mica or PMMA.Nevertheless, the maximum asperity height was 10 nm. Roughness of this order is not likely to cause the thousandfold dis- crepancy commented on by Walstra. The equating of applied pressure with the steric repulsion in a good solvent ap- pears straightforward. Our experiments indicate that the effect of added electrolyte For this reason electrostatic repulsion has been neglected since it was not significant. The repulsive pressures obtained in the compression experiment seem quite high. However, several theories exist for predicting the free energy change on the approach of two sterically stabilised flat plates. The steric pressure can be obtained by taking the derivative of the analytical expressions for the free energy change. Using appro- priate parameters for PVA and calculating theoretical curves of steric repulsive pres- sure against distance, we find that the calculated pressures are of the same order of magnitude as those observed experimentally.However, the theory predicts that the interaction should occur at smaller distances than those observed. Finally, it is interesting to note that Sonntagl found values of the repulsive pres- sures intermediate to those of Lyklema et al. and ourselves at comparable surface separations for PVA films. mol dmV3 NaC1) on the repulsive pressure is small. Prof. J. Lyklema (Wageningen) said: In trying to explain the differences between our free liquid film approach and the procedure adopted by Cain et al.I wonder to what extent the results may be influenced by the drainage pattern of the film, prior to the attainment of the stationary state. One typical feature of the free film method is that after formation a drop or dimple forms that after some time is sucked into the border. After this has happened the film is visually plane parallel, as can be confirmed within 5 nm by scanning. This step appears not possible in the drainage from between two rigid walls. Dr. J. B. Smitham and Prof. R. H. Ottewill (Bristol) said : In both types of apparatus we have used, continuous monitoring of the film area is possible. This can be achieved in the reflectance apparatus by direct viewing of the illuminated area and in the mul- tiple beam apparatus by direct examination of the FECO fringes.The latter provide an extremely sensitive method of monitoring the film profile. Thus any dimple formed by compression of the rubber is immediately visible and drainage of liquid and recovery of the dimpled surface to a plane interface can be monitored. Dimples only occur if the surfaces are pushed together rapidly. It has been our experience, however, that by using slow application of pressure, with precise fine control of the micrometer drive, the formation of dimples can be avoided. Dr. J. Klein (Rehouot) said: There are two related questions I would like to ask of (a) What is the resolution of the multiple beam interferometric technique as used (b) Can they give an estimate of the size of surface asperities both on the silvered Cain, Ottewill and Smitham.in their experiments ? H. Sonntag, Croatica Chem. Acta, 1976,48,439.GENERAL DISCUSSION 55 silicone rubber surface and on the poly(methylmethacry1ate) surface on which poly- mer is adsorbed? Dr. J. B. Smitham and Prof. R. H. Qttewill (Bristol) said: The resolution of the technique is limited by the measurable wavelength shifts of the fringes of equal chro- matic order (FECO). The question of the surface asperities was dealt with in some detail in the reply to Walstra. As pointed out there, the maximum size of the asperities can be related to the half-width of the fringes and this was estimated to be about 10 nm. In effect, this is also the resolution of the technique. The reproduci- bility of the method is well within these limits.The surface separations involved in the measurements are in the 60-120 nm and hence are large compared with the resolu- tion. The experimental accuracy is thus adequate to ensure meaningful measurements. Prof. A. Silberberg (Rehouot) said: In my comment on the work of Lyklema and van Vliet I have already pointed out the need to consider the fate of the bulk solution between two approaching surfaces. If equilibrium is to be discussed then clearly there must at all times be equilibrium between the film, the trapped unattached polymer and the bulk solution per se. Whether in equilibrium, or out of it, the presence of a polymer solution of finite concentration tends to give thicker films, certainly in cases where the adsorbed layers are (effectively) irreversibly attached.Direct comparison between data is, however, made very diAicult by the fact that different polymers, different solvents and different surfaces were used in these and in the measurements of others. The paper does not detail the time to equilibrium but I would judge that a wait period of the order of 25 min, though no apparent change in film thickness was noted, is too short. A more than small possibility exists therefore that present results do not reflect equilibrium values. The method is one of great beauty, however, and flexible enough to permit most of the questions raised to be answered properly in the future. Dr. J. B. Smitham and Prof. R. H. Qttewill (Bristol) said: Silberberg raises a num- ber of interesting points about the attainment of equilibrium in these studies.Quali- tatively, it takes several hours after the addition of polymer solution to the bare sur- faces before a stable film is formed when the rubbers are compressed. It was our ex- perience in a study by ellipsometry of the adsorption of gelatin into silver bromide surfaces that after about six hours both the surface excess concentration of gelatin and the optical film thickness reached a constant va1ue.l For this reason, compression studies were not commenced for a period of between 14 and 24 h after the addition of polymer solution. Thus we assumed that the surfaces were in adsorption “ equili- brium ” with the bulk polymer solution phase. We have not investigated in detail the effect of bulk polymer concentration on the equilibration time.The second equilibrium to consider is the “ approach equilibrium ” in the thin film which is formed on the close approach of the surfaces. The experimental evidence obtained, so far, indicates that this ‘‘ approach equilibrium ” was not a sensitive func- tion of electrolyte concentration, or bulk polymer concentration. Once the film had formed it remained stable for periods of several hours, i.e., although the film may not have been in true thermodynamic equiIibrium, it appeared to be so within the time scale of the experiment and it was not practicable to wait any longer before changing the applied pressure. We are in agreement with Lyklema and van Vliet that changing the bulk polymer concentration does not cause a marked change in repulsive pressure.T. J. Maternaghan and R. H. Ottewill, J. Phot. Sci., 1974, 22,279.56 GENERAL DISCUSSION This does not hold, however, if the polymer desorbs from the rubber surfaces; when this occurs the repulsion pressure decreases dramatically. Dr. Th. F. Tadros (Jealott’s Hill) said: The concentration of PVA in the thin liquid film between the rubber surfaces found by the authors, namely 0.55 g cm-3 seems to me extremely high. At such separation of -65 nm, this should correspond to an adsorption value of the order of 18 mg m-2 (if there is no polymer entrapped between the adsorbed layers), an extremely high value never obtained before on other surfaces, e.g. polystyrene, AgI, silica etc. The surface used by the authors is PMMA and ad- sorption of PVA on this could be different, but it is highly unlikely that such high adsorption value will be obtained.Dr. J. B. Smitham (Bristol) said: The reflectance method measures the refractive index of the thin film between the two silicone rubbers. Since the unadsorbed PVA is not removed, it is possible that, in addition to a PVA layer that is strongly adsorbed to the rubber surface, there are other layers weakly associated with the adsorbed layer. Indeed a study of the drainage curves as a function of increasing time of ad- sorption supports this view. Such multilayers would contribute to the refractive index of the thin film while not necessarily contributing to the surface coverage by the PVA. Thus what is determined by our experimental technique is the total polymer concentration between the surfaces.Adsorption isotherms obtained by centrifuging a dispersion stabilised by PVA may well remove the multilayers and measure only that PVA that is strongly adsorbed to the surfaces involved. Dr. A. Lips and Mr. E. J. Staples (Port Sunlight) said: In connection with studies of the adsorbed layer thickness of PVA on polystyrene latex particles,l we have performed osmotic pressure measurements on concentrated PVA solutions. We used the same polymer, Alcotex 88/10, as that employed by Cain et al. The results are summarised in fig. 1. At low polymer concentrations, C, the osmotic pressure, II, is consistent with the a value of the Flory Huggins parameter x 21 0.48. However, the observed behaviour indicates deviation from the Flory Huggins model in the direction of x increasing with C which is suggesting substantial attraction between the polymer molecules. This view is supported by the highly viscous, approaching gel-like nature of the solutions. Direct measurements of osmotic pressure for concentrations >0.2 g cmm3 have so far proved difficult. Nevertheless it can be inferred from the optical behaviour of more concentrated solutions that dII/dC is continuously and rapidly increasing with C, as suggested in fig. 1. The osmotic behaviour thus indicates the absence of a phase transition, and this has been confirmed for concentrations up to 0.45 g ~ m - ~ . Our measurements lend support to the view proposed by Cain et al. that the steric forces in their system are largely controlled by the compressibility of highly concen- trated polymer solutions. A quantitiative evaluation of this suggestion can be attempted if it is assumed that the approach of the two surfaces is accompanied by negligible change of adsorption and that the intervening polymer solution is sufficiently concentrated so that it can be treated as effectively homogeneous across the gap. The applied force, i.e., the steric force, can then be taken as equal to the differ- ence between the osmotic pressure of a polymer solution of concentration correspond- ing to that in the gap and the pressure of a solution representing conditions far away from the surface. The former osmotic pressure is usually dominant. Comparing our osmotic pressure measurements with the measurements of steric D. S. Duclcworth, A. Lips and E. J. Staples, this Discussion.GENERAL DISCUSSION 57 C/g ~ r n - ~ FIG. 1 .-Main diagram represents osmotic pressure measurements on PVA, Alcotex 88/10. The full lines are theoretical expectations based on Flory-Huggins theory for the values of x as indicated. Insert represents fig. 5 of paper by Cain et al. [applied pressure plotted against (separation)-'] force (fig. 5 of Cain et al), we can obtain a reasonable representation in terms of the homogeneous model of the dependence of the steric force on separation if the con- centration at a separation of 60 nm is taken as -0.2 g cm-3 (fig. 1). The quality of the fit may be fortuitous. Nevertheless, the value of osmotic measurements on concentrated solutions both in assessing a polymer system for con- formity to theoretical models and in providing a base-line for predicting steric forces is clearly demonstrated.