摘要:

Nuclear Magnetic Resonance Studies of Water in Disperse Systems B Y J. CLIFFORD J. OAKES AND G. J. T. ‘rIDDY Unilever Research Laboratory Port Sunlight Cheshire England Received 5th May 1970 The nuclear magnetic resonance relaxation times of water protons in aqueous suspensions of polystyrene lattices and in lamellar mesomorphic phases have been measured. The results indicate that the surfaces examined have no long range effect on water structure. The main effect is the binding of water inolecules to charged surface groups though an additional effect occurs when pores are formed by flocculation of the polystyrene spheres or when the lamellar phase contains water layers less than 20 A thick. The effect of surfaces on the structure and properties of liquid water has been much discussed 1 v as a possible factor determining the behaviour of disperse systems.For example it has been suggested that the effect of surfaces on water contributes to the stability of colloidal dispersion^,^ and that the state of water in biological systems is very different from that of ordinary bulk water.4 However much of the experimental evidence concerning water in disperse systems is indirect or incomplete i.e. it has been obtained from work on inadequately characterized systems. Con- sequently there remains much doubt as to the nature and range of surface effects in aqueous dispersions. The measurement of the nuclear magnetic resonance relaxation times of water protons is a direct method of determining the mobility of water rnolec~les.~ It has been extensively applied to the investigations of monolayer amounts of water adsorbed on surfaces and to a wide variety of complex biological and other practical systems but less often to well-characterized colloidal dispersions in which solid surfaces are i n contact with bulk water.Most of the work which has been done on such systems (e.g. micelle~,~ silica dispersions,* clays,g inesomorphic phases of soaps lo) has indicated that unless the water is present in small or intermediate l1 pores the perturbing effect of surfaces on water structure is confined to relatively few (1-3) molecular layers of water adjacent to the surfaces and that long-range effects are absent. However investigations of dispersions of polyvinyl acetate spheres in water indicated that (i) fastest flocculation rates were markedly slower than predicted by the Smoluchowski theory (ii) that the particles are surrounded by a 30A-thick layer of water with a viscosity 1,000 times greater than that of bulk water at the same temperature and (iii) that as the concentration of particles is increased a co-operative effect occurs increasing the viscosity of water between particles and that this becomes noticeable at distances of about 0.1 pm.It has been suggested that these effects were due to highly porous particle surfaces.’ Because of this and because the long- range effect seemed anomalous when considered in relation to results obtained in other systems measurements have been made on well-characterized polymer sphere dispersions and are described here. 175 176 NUCLEAR MAGNETIC RESONANCE STUDIES As it was found that water structuring effects were very small for water layers of the dimensions obtainable in these systems the properties of water in lamellar mesomorphic phases were also investigated.In these systems water layers of from 50 to 8 A thick can be obtained. The behaviour of water in such phases is of particular interest in that each water layer together with its two boundary monolayers of soap molecules resembles a thin soap film. Consequently investigations of the state of the water in these systems are relevant to studies of soap film properties. EXPERIMENTAL MATERIALS POLYSTYRENE LATICES Polystyrene latices were prepared by emulsion polymerization of styrene (with sodium decanoate as the emulsifying agent) by a method similar to that described by Ottewill.12 Impurities were removed by filtration followed by dialysis.The dialysis is continued (for about 6 weeks) until the electrophoretic mobility of the particles at pH 9 remained constant. Full details of the preparation and characterization of the latices will be given in a later publication in which other properties of the dispersions will be described. For our work two preparations were used. Their characteristics are summarized in table 1. TABLE 1 mean particle diam. COO- groups per particle area per COO- group A 5 200 A 7 4 0 ~ 103 119 890 A 1 . 8 ~ 103 1370 Histograms of particle size distributions are shown in fig. 1. The particle diameters were determined by electron microscopy and confirmed by light scattering. The concentrations of COO- groups were estimated from potentiometric titrations. 30.- 2 0 - 10- I 1000 1200 2 0 0 0 3 0 0 0 4000 diameter 8 diameter 8 C L '000 FIG.1 .-The size distribution histograms for the polystyrene particles. J. CLIFFORD J . OAKES AND G . J . T. TIDDY 177 All n.m.r. measurements were carried out at pH of between 7 and 8 where the COO- groups are completely ionized. A small dependence of water proton spin-spin relaxation rate on pH was noted. This was the same as that observed for pure water the rate being slightly greater at pH 6-8 than at higher and lower pH values (at pH values of less than 5 flocculation occurred). This effect which is attributed to exchange modulated interactions with 01' nuclei does not appear to be affected by the presence of the colloid particles. For the measurements described samples of a high solids content were required. These were obtained by concentrating dilute suspensions by means of centrifugation.Provided that concentration is to no more than about 20 % volume fraction of solid this method is completely reversible. The latices could be redispersed by shaking-or more conveniently by means of ultra-sonic irradiation. The n.m.r. measurements on the water protons were time independent and were unaltered by cycles of concentration and dilution and by cycles of temperature variation. If concentrations of more than 20 % volume fraction of solid were prepared irreversible flocculation occurred and irreversible changes occurred in the n.m.r. properties of the suspensions. Such changes occurred at much lower volume fractions if other methods of concentration-evaporation of water in a stream of gas or under vacuum or freeze drying-were used.Consequently all the results described were obtained on systems concentrated by centrifugation. LAMELLAR MESOMORPHIC PHASES The system decanol+ sodium caprylate+ water was used to provide lamellar meso- morphic phases as Eckwall l 3 has shown that in this system the lamellar D phase is given by a wide range of compositions at room temperature. A triangular diagram based on Eckwall's work showing the D-phase region is given in fig. 2. The D-phase consists of bilayers of caprylate and decanol molecules separated by layers of water. Mixtures of the required compositions were made up melted mixed thoroughly and equilibrated at room temperature for at least one week before n.m.r. measurements were made. X-ray diffraction measurements were also made on the samples.These were in reasonable agreement with the measurements made by Eckwall (the small differences observed do not affect our inter- pretation of our n.m.r. results) and confirm the lamellar structure suggested by him for the D-phase. OECANOL / 1 5 0 SODIUM CAPRYLATE WAT E R FIG. 2.-Lamellar D-phase compositions in the sodium caprylate+ decanol+ water system at 20°C. MEASUREMENTS Nuclear magnetic resonance relaxation times of water protons in these systems were The tempera- Spin-lattice measured with a Bruker Physik pulse spectrometer at a frequency of 60 MHz. ture of the samples was controlled to within 1°C with a gas flow thermostat. 178 NUCLEAR MAGNETIC RESONANCE STUDIES relaxation times were measured with 90"-90" pulse programmes. Spin-spin relaxation times were measured by the Gill Meiboom l 4 technique.Preliminary work shown that for these samples the results of spin-spin relaxation time measurements were dependent on pulse separations decreasing as pulse separations are increased. This effect which has been attributed to the diffusion of molecules in the magnetically inhomogeneous disperse system,' was eliminated by the use of very short pulse separations (80,~s). The precision of the measurements is better than f5 % for TI and for T2 in the polymer latices and f10 % for T2 in the lamellar phase systems where separation of non exponential decay curves into two components is involved. RESULTS POLYMER LATICES The results of the n.m.r. measurements on the polystyrene latices are shown in fig. 3 4 5 and 6. The relaxation rates l/Tl and 1/T2 of the water protons are shown as functions of V/(1- V) where Y is the volume fraction of solid material.For all samples exponential decay curves were obtained which could be characterized by a single relaxation time. (The polystyrene protons gave free induction decays with characteristic times T2 - 1 x s and TI - 1.5 s and did not interfere with the measurements on the water protons). Samples containing less than 20 % v/v of solid gave TI results which were identical with those of pure water within experimental error. Some other observations were (i) Samples with concentration of solid of more than 20 % were flocculated and could not be entirely redispersed by shaking or by ultrasonic irradiation. (ii) All samples freeze at temperatures between 0 and N 2"C i.e. no observable n.m.r.signal remains below this temperature. (iii) At solid concentrations of below 20 % v/v the chemical shift of the water protons in the lattices is identical with that of bulk water within experimental error At concentra- tions greater than this a concentration dependent chemical shift is observed-about 1 p.p.m. to high field relative to water at the same temperature for a 75 % solid content system. (iv) Preliminary results indicate that these suspensions have a maxi- mum flocculation rate equal to that predicted by the Smoluchowski theory.16 LAMELLAR PHASE The compositions of samples of the D-phase of the sodium caprylate decanol water system are given in table 2. These samples containing both water and lipid TABLE 2 sample 1 2 3 4 5 6 7 8 9 10 11 12 mol ratio caprylate decanol water water 0.060 0.148 0.045 0.110 0.032 0.079 0.029 0.060 0.018 0.099 0.014 0.034 0.045 0.135 0.049 0.147 0.074 0.143 0.102 0.142 0.129 0.141 0.160 0.142 thickness of water layer A 11.8 16.7 21.7 30 42 48 43 15 16.5 9.7 8.3 8.0 1 IT2 218 79 173 39 33 30 172 135 206 426 891 1,211 1ITlW 1.48 1.04 0.57 0.52 0.49 0.45 0.58 1.7 1.95 2.63 3.20 2.0 J.CLIFFORD J . OAKES AND G . J . T. TIDDY 179 protons and their relaxation behaviour was more complex than was observed for the polystyrene suspensions. The spin-spin relaxation curves were non-exponential and consisted of two components. One had an intensity proportional to the amount of non-aqueous component present and spin-spin relaxation times of from 2.5 x to 5 x s. The other proportional to the fraction of water present in the sample had relaxation rates l/T which are given in table 2 and shown in fig.7 as a function of mol caprylate/mol water. For some samples the slower decay curve was itself non-exponential. This is considered to be due to the presence of an additional lamellar phase probably because of incomplete equilibration in these highly viscous systems. For these samples the spin-spin decay curve is resolved into its two com- ponents and an average relaxation rate calculated from their proportions and relaxa- tion rates is included in fig. 7 and table 2. Thus the measured spin-lattice relaxation rate l/Tl shown in fig. 8 as a function of mol caprylate/mol water is an average of the relaxation rates of the lipid protons and the water protons. This averaging is due to the combined effects of spin diffusion in the non-aqueous part of the sample and molecular diffusion in the water.As we are interested only in the water protons in the present study the relaxation rate of the lipid protons was measured by using caprylate + decanol + deuterium oxide systems (the results are shown in fig. 8) and its effect on the average relaxation rate calculated and subtracted from the measured rate for each sample and the spin lattice relaxation rate of the water protons l/Tlw estimated and recorded in table 2 and given as a function of mol caprylate/mol water in fig. 9. The spin-lattice relaxation curve was always exponential for all samples. DISCUSSION POLYSTYRENE LATICES It has been shown l7 that if protons exist in two states exchanging so that on average n.m.r. relaxation rate is obsei.ved,l then provided that TA % TB S A 4 S, PA+& 1 where T is the observed relaxation time TA and TB are the relaxation times in the two states PA and PB are the populations of the two states and SB is the average residence time of a proton in the state B.If it is assumed that water affected by the particle surface exchanges with bulk water then we may denote state A as bulk water and state B as water modified by the particles. Consequently there should be a linear relationship between the observed relaxation rate 1/T and V/(l - V) where V is the volume fraction of solid provided that there is no interaction between the particles in their affect on water structure. Fig. 3-6 can be considered as describing two distinct concentration regions 0-20 % solids content where a non-flocculated dispersion is being examined and more than 20 % solid content where flocculation has occurred.Below 20 % solid content a linear relation between 1/T2 and V/(l - V ) is observed for both the 890 and 5,200 8 diam particles. There is no measurable effect of the particle surfaces on Tl of the water protons when solids content is less than 20 %. The lack of any measurable decrease in Tl and the small effect on T2 for water protons in the non-flocculated systems indicates that there is little long-range influence of these surfaces on water structure. 180 NUCLEAR MAGNETIC RESONANCE STUDIES r( I in c - U 0 *2 I n I o'---l---f-- 0.5 1.5 ' 2 V/U - V ) FIG. 3.-The dependence of the spin-spin relaxation rate of water protons 1 /T2 on V/(l - V ) where V is the volume fraction of solid for suspensions of 5,200 A diameter polystyrene particles 0 58°C ; x 47°C; 0 37°C; 0 30°C; A 20°C; 'J 10°C.On the basis of the Bloembergen Purcell and Pound theory for dipolar n.m.r. relaxation? in liquids the observations could be explained either by a negligibly small effect of the surfaces on the motion of water molecules or to the presence of tightly- bound water molecules with correlation times for molecular rotation of five or six orders of magnitude greater than those in normal water. The difference between TI and T2 in these systems indicates that the second alternative is correct but the magnitude of the T2 effect shows that the amount of tightly-bound water must be small-much less than a monolayer of water molecules on the surface. o-2'I FLOCCULATED 0 I FLOCCULATED 1 I I I I 0 0.5 I 1.5 2 .0 V/U - v> FIG. 4.-The dependence of the spin-lattice relaxation rate of water protons 1/T on ?'/(I - V ) where V is the volume fraction of solid for suspensions of 5,200 A diameter polystyrene particles 0 58°C; X 47°C; 0 37°C; 0 30°C; A 20°C; V 10°C. J . CLIFFORD J . OAKES AND G . J . T. TIDDY 181 FIG. 5 . T h e dependence of the spin-spin relaxation rate of water protons l/Tz on V/(l- V) where Vis the volume fraction of solid for suspensions of 890 A diameter poly- styrene particles 0 58°C; X 47°C ; 0,37"C ; 0,30"C ; A 20°C ; v 10°C. 1.0 0.75 - 0 . 5 L !s r3 0.2 5 NOT -1 FLOC C U L ATED -ATE D (3.5 t i FLOCCULATED FLOCCULATED FIG. 6.-The dependence of the spin-lattice relaxation rate of water protons l/Tl on V/(1- V) where V is the volume fraction of solid for suspensions of 890"A diam.polystyrene particles 0.58"C ; x 47°C; .* -37°C; 0 30°C;. A 20°C ; v 10°C. I I I J 0 0.5 1.0 1.5 182 NUCLEAR MAGNETIC RESONANCE STUDIES The main effect of the particles on the water may be due to the binding of water molecules to a few active sites on the particle surfaces. If it is assumed that one water molecule is bound to each COO- surface group relaxation rates 1/(T2,+SB) of 1,200 s-1 for the 5,200 L! particle system and 4,000 s-l for the 890 L! particle system can be calculated for water protons bound to COO- groups. There would be no measurable effect on the average Tl for these systems. The difference between the two kinds of particle indicates that active groups other than COO- are present. In such polymer systems-0-H groups can also be present on the surface.lg If the surface concentration of these groups is the same for both particles the effect for a given particle concentration would be greater for the smaller particles as is observed.(Although the values of 1 /(T2B + S,) are different for the different size particles the slopes of the (log[l/(T,,+S,)] 1/TK) curves are the same indicating that the activation energies are similar for the water bound on both types of particle. The rates decrease with increase of temperature indicating that T2B is substantially larger than SB). Thus for these non-flocculated polystyrene latices the main effect on the aqueous solvent appears to be the binding of a small number of water molecules at active sites COO- and perhaps OH groups. If there is any other effect it is too small to be measured by the techniques we have used.It can be calculated that any such extra effect must be no greater than e.g. the doubling of the correlation time for molecular motion of three layers of water round the particles. These conclusions are in accord with the preliminary finding that the flocculation rate of these particles obeys the Smoluchowski equation. Systems containing more than 20 % solid material are at least partly flocculated i.e. they cannot be entirely redispersed by shaking or ultrasonic irradiation. In flocculated systems “ pores ” will be formed by contact between particles and in regions near where the contact takes place the distance between the solid surfaces will be of molecular dimension and water properties in these regions may be modified.Previous work has shown that in porous systems the mobility of water molecules is decreased and the chemical shift of water protons moved to higher fields.’ An up-field chemical shift is indeed observed in the flocculated polymer latex systems and fig. 9 and 6 show that l/Tl is increased on flocculation for all temperatures indicating that some slower moving water molecules are now taking part in the averaging process that determines the observed Tl. These are presumably the water molecules in the narrower parts of the “ pores ” formed by flocculation. In flocculated systems the spin-spin relaxation rate 1 /T2 however will still be mainly determined by water molecules tightly bound to active sites on the particle surfaces. Flocculation will reduce the number of accessible sites and this will reduce 1/T2.Fig. 1 and 3 show that at low temperatures this effect is dominant causing a reduction of 1/T2 below the value expected for a given concentration. At high temperatures the slowing-down of molecular motion of water molecules in small pores which will lead to increased 1/T2 as it does l/Tl appears to be more important. The results given in this paper contrast with those reported previously in which large effects of polyvinyl acetate particles on l/Tl and 1/T2 of water protons were found even in much more dilute systems. There is some evidence that the surfaces of the particles used in the early work were porous with polymer chains extending from the surface into the water. The large effect of the particles on the water could well be due to the trapping of water molecules between such chains.Their anomalously slow flocculation rates could also be caused by such extruding chains through an entropic repulsion mechani~rn.~’ Also the solutions used in the early were concentrated by evaporation of water. We have now observed that this work J . CLIFFORD J . OAKES A N D G . J . T. TIDDY 183 causes flocculation at relatively low concentrations of solid. This may well account for the changes in slopes of the (l/T V/(l - V ) ) curves which were observed and which were attributed to long-range cooperative effects on structuring of the water between polyvinyl acetate latex particles. LAMELLAR PHASES The data given in table 2 and fig. 7 and 9 indicated that the relaxation rates of the water protons on the system are determined mainly by the ratio of COO- groups to water molecules in the system.There is no such relationship between the [decanol]/ [water] ratio and water proton relaxtaion rates. In these systems also eqn (1) should hold. If one assumes that the dominant effect on water mobility is binding ',,Or I000 - 800- r( !A mol caprylate/mol water FIG. 7.-The dependence of the spin-spin relaxation rate of water protons 1/T2 on the ratio mol caprylate/mol water in the lamellar D-phase of the sodium caprylate+ decanol+ water system. to COO- groups then the observed relaxation rate 1/T should vary linearly with the ratio (mol caprylate)/(mol water). Fig. 9 and table 1 2 show that this is so for water layer thicknesses of from 48 to 22 A. Over this range the effect of the surfaces on the spin lattice relaxation rate of the water protons can be accounted for in the same way as in the polystyrene latices-by an exchange averaging of water tightly bound to COO- groups with water with the same molecular rotation and translation rates as normal bulk water.Exchange with water bound to -0-H groups and with -0-H protons will also occur but it seems that this is not so important in these systems. 184 4. NUCLEAR MAGNETIC RESONANCE STUDIES 0 0 X 0 OX 0 0 0 O 0 I I I 0.0 5 0. I 0.1 5 mol caprylate/inol water FIG. 8.-The dependence of the spin lattice relaxation rate of protons l/Tl on the ratio rnol caprylatel rnol water in the lamellar D-phase of the sodium caprylate+decanol+water system. 0 H,O systems x D20 systems. 0 0 1 I I 0.05 0.1 0-15 mol caprylate/mol water FIG. 9.-The dependence of the spin lattice relaxation rate of water protons l/Tl W on the ratio mol caprylate/mol water in the lamdlar D-phase of the sodium caprylate + decanol +water system.J . CLIFFORD J . OAKES AND G . J . T. TIDDY 185 When water layer thicknesses of 17L$ or less are involved however there is a much greater effect on l/T1 for water protons. Fig. 9 and table 2 show that there is first a marked increase in the relaxation rate then a decrease as the thickness of the water layer is reduced with a maximum I/Tl at about 8.3 A. This is exactly what would be expected on the basis of the Bloemburgen Purcell and Pound theory of dipolar n.m.r. relaxation in liquids if the water molecules moved more and more slowly as the water layer thickness is reduced below 2OA. At the maximum value of l/Tl at 8.3 A the average correlation time for rotation of water molecules in the system will be 2 x s about three orders of magnitude longer than in ordinary water.This effect is much greater than is observed in a solution with the same concentra- tions of Naf and COO- groups in the form of sodium acetate even for water layer thicknesses of more than 20 A. From the gradients of (1 /T mol COO-/mol H20) curves molar relaxation rate enhancements A( 1 IT) can be calculated ; where A( 1 / T ) = (1 / T ) (molar solution) - (1 / T ) (water). These are shown in table 3. (The TABLE 3 system A( 1 /TI) at 20°C A( 1 /T2) at 20°C 1 s-1 mol-1 1 s-1 mol-1 5,200 A diam particles (not flocculated) - 17.4 890 A particles (not flocculated) - 71.4 lamellar D-phase of caprylate+ decanolf water systems ; water layer thickness 20 A 0.095 26.0 solution of sodium acetate 0.027 0.027 A(l/T,) for the polymer latices are too small to be measured at the concentrations of COO- groups attainable.) 1/T value for solutions of sodium acetate does not vary much with concentration from 0-8 M.) The relatively small effect of CH,COO- groups in solution on water proton n.m.r.relaxation times is presumably due to the effect of the rotation of the acetate ion + water complex on the correlation times determining the dipolar relaxa- tion rates of the water protons an effect which would not occur in COO- groups on surfaces. The spin-spin relaxation rates of water protons in the lamellar D-phase system shown in fig. 7 are also largely determined by the averaging between water molecules tightly bound to -COO- groups and -OH groups and the remaining water molecules.The spin-spin relaxation rates for the bound water molecules are much higher than the spin-lattice relaxation rates so that here the effect of water layer thickness on water structure is much less evident. Nevertheless some increase in slope of the (l/T2 mol caprylatelmol water) curve is evident as the thickness of the water layer is decreased. The difference between the 1 /T values given in table 3 for the polymer latices and those for lamellar phase are not surprising as systems with widely different concentrations of -COO- groups and -OH groups are being considered. CONCLUSIONS In the systems we have examined there appears to be no long-range effect of surfaces on water structure. For the polymer latices measurable effects on water structure only appear when pores are formed either as a result of the polymerization method or by flocculation.Normally the main effect is the binding of water mole- cules to charged surface groups although the modification of one or two layers of water by uncharged surfaces cannot be excluded. In general the effect of the 186 NUCLEAR MAGNETIC RESONANCE STUDIES modification of watcr structure by colloid particles on colloid stability for non-porous systems is likely to be small. Similar conclusions can be drawn for the lamellar phase that we have investigated. For systems containing layers of water more than 20 A thick all the effects observed can be explained in terms of the binding of a few water molecules to -COO- groups. Where the water thickness layer ie less than 20A there is a marked reduction in water molecular mobility as normal water structure ceases to exist and is replaced by a more rigid structure determined by the interaction of water with charged surface groups counter ions and hydrogen-bonding surface groups.Thus it is only for such relatively narrow water layers that special water structure effects need be allowed for when the properties of thin soap films are being considered. The authors thank Dr. D. Nicholls who made the polystyrene latices Dr. A. Lips and his colleagues who characterized them and Dr. B. A. Pethica for helpful discussions. J. Clifford and B. A. Pethica Hydrogen Bonded Solvent Systems ed. A. K. Covington 1968 B. V. Deryaguin Disc. Faruday SOC. 1966 42 109. G. A. Johnson S. M. A. Lecchini E. G. Smith J.Clifford and B. A. Pethica Disc. Furuday SOC. 1966 42 120. F. W. Cope Biophys. J 1969 303 9. T. M. Connor Trans. Faraday SOC. 1963 59 1574. D. E. Woessner J. Chem. Phys. 1963 39,2783. J. Clifford B. A. Pethica and W. A. Senior Ann. N . Y. Acad. Sci. 1965,125,458. J. Clifford and S. M. A. Lecchini SOC. Chem. Ind. Monograph no. 25 Wetting p. 174. T. H. Wu. J. Geophys. Res. 1964 69 1083. pp. 169-179. lo K. N. Lawson and T. J. Flautt J . Phys. Chem. 1968,72,2066. l 1 M. M. Dubinin Quart. Reu. 1959 9 101. l2 R. H. Ottewill and J. N. Shaw Kolloid 2. 2. Polymere 1967 218 34. l3 L. Mindell K. Fontell H. Lehtinen and P. Ekwall Acta Polytechn. Scand. 1968 74 1 . l4 D. Gill and S. Meiboom Rev. Sci. Instr. 1958 29 688. l6 A. L. Smith private communication. l7 D. E. Woessner J. Chem. Phys. 1961 35,41. E. L. Mackor J. ColloidSci. 1951 6 492. l 9 R. H. Ottewill private communication. J. R. Hansen and K. D. Lawson Nature 1970 225 542.

ISSN:0370-9302    

DOI:10.1039/SD9700100175

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Interlayer Water in Vermiculite Thermodynamic Properties Packing Density Nuclear Pulse Resonance and Infra-Red Absorption. BY J. HOUGARDY," J. M. SERRATOSA,~ w. STONE* AND H. VAN OLPHEN Laboratoire de Physica-Chimie Minkale Instituto de Edafologia y Biologia Vegetal and Exploration and Production Research Division of Shell Development Company P.O. Box 481 Houston Texas U.S.A. Received 16th March 1970 Water vapour sorption isotherms on sodium and magnesium vermiculite of high charge density were measured at 25 and 50°C. Heats of immersion at various stages of hydration were measured at 25°C. According to X-ray observations during the sorption process two discrete interlayer hydrates are obtained with one and two monolayers of water between the unit layers of the crystallites. The derived integral entropy of adsorption indicates a reduced freedom of motion of the interlayer water molecules compared with that for water molecules in the liquid state.Comparison of the apparent density of the hydrated clay with the calculated crystallographic density indicates that inter- layer water is slightly more densily packed than liquid water. However the bulk density of water in a sodium vermiculite suspension is normal for the water in excess of the two hydration layers. Nuclear pulse resonance results obtained on sodium vermiculite shows that water in the one- layer hydrate is organized and some hypotheses are presented regarding this organization. Water molecules in the two-layer complex show the same degree of orientation only below -65°C. Infra- red absorption spectra for hydrated flakes of sodium vermiculite were obtained as a function of angle between the (001) plane of the crystallites and the i.-r.beam. The results indicate orientation of water molecules in the one-layer hydrate at room temperature. The system expanding clay +water offers unique possibilities for the experimental study of the properties of water near a solid surface. In the expanded state the ad- sorbed water intercalated between the unit layers of the silicate represents a con- siderable volume fraction of the system ; the thickness of the intercalated water layers can be measured by X-ray diffraction; oriented flakes can be used for example in i.-r. studies affording the determination of the disposition of the water molecules. In suspension water properties can be determined at distances beyond the thickness of adsorbed water layers.Furthermore the solid surface is flat and its structure is reasonably well known as well as its charge density which is determined by isomor- phous ion substitutions within the unit layers. A high charge density vermiculite was selected as a particularly suitable mineral since it displays two well-separated discrete stages of hydration with one respectively two monolayers of water between the unit layers. Identical samples were used in the participating laboratories. Previously published data on the system are presented in summarized form only. * present address Laboratoire de Physico-Chimie Minkrale Heverlee-Louvain Belgium. present address Instituto de Edafologia y Biologia Vegetal Madrid Spain. present address National Academy of Sciences Washington D.C.U.S.A. 187 188 INTERLAYER WATER I N VERMICULITE CHARACTERIZATION OF THE SYSTEM The vermiculite sample was obtained near Llano Texas. Impurities were removed and the counter ions occurring in the natural clay (primarily Mg2+) were replaced by Na+ ions by exhaustive treatment with NaCI. The resulting Na-clay was washed with distilled water. The Mg-clay was prepared from the Na-clay by treatment with MgC12 solutions. The unit cell formula derived from chemical analysis is (si5. 28A12.,2)(A11 .32Mg4. 58)O2O(OH)4 -Naf.6(resp. - Mgi.i). Hence the unit cell weight is 793.6 (776.2 for Mg). The cation exchange capacity is 200 mequiv./100 g of dry clay (204 mequiv./100 g for Mg). The unit cell dimensions are a = 5.21 A b = 9.18 A c = 9.82 A (9.3 A for Mg).The area available per ion on exterior surfaces is 60 Hi2 (120 A2 for Mg) and in the interlayer space 30 A" (60A2 for Mg). The total unit layer surface area is 725 m2/g (741 m2/g for Mg). The surface density of charge is 26.7 p C/cm2. The total exterior surface area from argon desorption isotherms applying B.E.T. analysis is 3.5 m2/g (3.1 m2 for Mg) hence the total area of the interlayer space is 361 m2/g (369 m2/g for Mg). DENSITY OP WATER IN THE SYSTEM Pyknometric density determinations on the sodium vermiculite +water system have been reported by Deeds et al.' The apparent density of the clay was compared with the calculated crystallographic density. The observed small difference of these values would indicate an estimated 2.5 % greater packing density of the water mole- cules in the interlayer space assuming the normal density of water beyond two adsorbed water layers in bulk.The latter assumption is supported by determination of the apparent density of the clay with two pre-adsorbed layers of water using n- decane as the displacement liquid. This experiments yields a value which is identical with the apparent density of the clay in water. Since it is unlikely that abnormal densities would occur in n-decane surrounding the hydrated clay particle the water density would indeed be normal beyond about 5A from the surface according to these observations. These results contradict those of Anderson and LOW,^ who concluded from experiments involving differential displacement of water in a bentonite paste by mercury (comparing injected mercury volume increments with displaced water volume increments) that water densities are up to 3 % low in the range between 0 and 60 A from the clay surface.If their results were correct and would also apply to the vermiculite+ water system the apparent density of sodium vermiculite in water should have been about 20 % lower than that in the system prehydrated vermiculite+ n-decane. This difference would be far greater than the experimental error in the pyknometric data. THERMODYNAMIC PROPERTIES OF THE SYSTEM COMPOSITION OF THE HYDRATES Adsorption-desorption isotherms for water vapour at 25 and 50°C were determined and (001) spacings were measured at successive stages of hydration by van Ol~hen.~. From the two-step isotherms monolayer and two-layer coverages were derived from Langmuir plots. The monolayer hydrate of sodium vermiculite contains 2 molecules of water for each sodium ion which is positioned midway between the unit layers.In the two-layer hydrate each sodium ion is surrounded by almost 6 water molecules 3 in a plane above and 3 in a plane The results are summarized in table 1. J . HOUGARDY J . M . SERRATOSA W. STONE A N D H . VAN OLPHEN 189 below the midway sodium ion. In the Mg-vermiculite monolayer hydrate which is stable only below a relative pressure of 0.015 the water molecules are very loosely packed and at higher presssures the formation of the two-layer hydrate is favoured over the filling of vacant positions in the monolayer hydrate. In the two-layer clay TABLE 1 .-COMPOSITION OF HYDRATES (001) spacing molec. ratio H2O/ion area per molec. H2O Na-vermiculite one-layer hydrate 11.8 A 2.0 13.5 A2t t w o-layer hydrate 14.8 A 5.6-6.0 10.8-10.0 A2 Mg-vermiculite one-layer hydrate 11.6 A 3.5 17.0 A2 two-layer hydrate* 14.8 A 10.0 12.0 A2 14.8 A 11.0 10.8 A* * Two two-layer hydrates can be distinguished in the isotherm having identical X-ray spacings but tallowing 3AZ for the area occupied by the sodium ion.different packing densities for water. hydrate the water molecules are partially coordinated octahedrally around the Mg ions and partially arranged in a hydrogen bonded network according to Bradley and Serratosa.5 HYDRATION ENERGY Hydration energies were derived from the sorption isotherms. J/m2 for the first layer and 40 x For Na-vermi- culite these are (at 25°C) 105 x J/m2 for the second layer. These values represent the net energies of expansion consisting of the combined effects of adsorption electrostatic layer interaction and the van der Waals attraction between the unit layers.The electrostatic attraction between the cations and the negatively charged unit layers amounts to 4n.a2x/e in which c is the charge density 2x the separation of the unit layers and 8 the dielectric constant which will have a value between about 3 and 6. Neglecting the van der Waals attraction which will be small with respect to the electrostatic attraction the adsorption energy equals the sum of the net hydration energy and the electrostatic energy. For the formation of the monolayer hydrate the adsorption energy amounts to 240 to 375 x J/m2 which corresponds with 10-15.5 kcal/mol of sodium ion. For the formation of the two-layer complex from the dry clay the adsorption energy amounts to 21-32 kcal/mol of sodium ion.These estimates show that the hydration of the clay calculated on a per ion basis is considerably smaller than the ion hydration energy in bulk solution. HEATS OF IMMERSION Heats of immersion were determined for the clays at various stages of hydration. For Na-vermiculite the average heats of hydration amount to 3.75 and 1.9 kcal/mol of water for the first and the second layer of water respectively. For Mg-vermiculite these values are 8.0 and 3.5 kcal/mol of water respectively. ENTROPY OF HYDRATION Integral entropies of hydration were derived from isotherm and heat of immersion Because of lack of detail in the isotherms in the small monolayer hydration data. I90 INTERLAYER WATER IN VERMICULITE region for the Mg clay entropies could only be evaluated for the sodium clay.For both the one-layer and the two-layer adsorption processes the integral entropies are negative with respect to that of liquid water T(S-S,) is between -2 and - 3 kcal/ mol of water. These values apply to changes occurring in the entire system and include changes in the adsorbed phase the crystal phase and the ions. Since the latter two are likely to result in entropy gains the contribution to the entropy change by the water phase may be somewhat more negative than the net values. Hence the water molecules in the adsorbed phase will have a higher degree of order than that existing in the bulk liquid according to this interpretation. NUCLEAR PULSE RESONANCE Pulsed nuclear magnetic resonance experiments were carried out on powdered samples of sodium vermiculite.Particle diameters ranged from 7 to 0.5 pni. The Fe content of the sample was 1,710 p.p.m. as determined by flame absorption spectro- scopy. The one-layer and two-layer hydrates were prepared by equilibration of the sample with water vapour at relative pressures of 0.18 and 0.66 respectively. Meas- urements were made in the temperature range from 50 to - 180°C. ~~~ 0 + 5 0 C 1 0 +I0 c 0 - 1 5 0 C U \ I"; 0 5 0 100 0 5 0 100 t (w) FIG. 1 .-Free induction decay of powdered Llano Na-Vermiculite at various temperatures. Curve The shape of the free induction decay signal following a 90" pulse shows some interesting features. For the monolayer hydrate the signal shows a hump within the free induction decay irrespective of temperature (fig.1). The hump shifts towards the origin of the signal as the temperature is lowered from ambient to +5"C and then stays at approximately the same position for lower temperatures. This effect (reflect- marked (I) for monolayer hydrate marked (11) for two-layer hydrate. J . HOUGARDY J . M. SERRATOSA W. STONE AND H. VAN OLPHEN 191 ing the presence of a characteristic doublet in the n.m.r. absorption curve 6*7 ) is attributed to an averaged preferential orientation of the water molecules. This orientation effect which is observed at room temperature is enhanced upon lowering of the temperature and levels off below + 5°C. Due to the random orientation of the vermiculite flakes it was impossible to determine the orientation of the water intra- protonic vector with respect to the crystal faces.However in an oriented specimen of hectorite clay Woessner and Snowden have shown from analogous experiments that the intraprotonic vector has a tendency to be parallel to the (001) planes. Ex- periments on oriented flakes of vermiculite are in progress. n 2 W h" h E W 2G6 I 103p (K) FIG. 2.-Spin lattice relaxation times of powdered Llano Na-Vermiculite at various temperatures. e two-layer hydrate -lines drawn show positions of the two minima - - - - - - calculated combina- tion of the two relaxations 0 one-layer hydrate. For the two-layer hydrate of vermiculite the decay of the signal is monotonic for temperatures between 20 and -60°C. Below this temperature a hump is again observed (fig. 1). It is concluded therefore that for the two-layer hydrate the orientation effect is obtained at a much lower temperature than for the monolayer hydrate.Apparently the presence of a larger amount of water prevents to a large extent the preferential orientation. In order to gain information about spin mobility spin lattice relaxation times TI were measured i n the same temperature range. The data for TI as a function of temperature for the one-layer and two-layer hydrates-corrected for the influence of constitutional hydroxyls and paramagnetic centres are shown in fig. 2. 192 INTERLAYER WATER I N VERMICULITE For the two-layer hydrate a minimum is observed around +38"C beyond which Tl levels off until it increases abruptly at - 70°C. This behaviour can be explained by postulating the existence of a second minimum situated at around -60°C.This minimum would be the same as that observed for Na-montmorillonite by Touillaux et aZ.,9 who proposed proton diffusion as the relaxation mechanism. In each case an activation energy of 6 kcal is obtained. The minimum at high temperature could perhaps be attributed to a diffusion of the water molecules (E = 4 kcal). For the one-layer hydrate these two minima also seem to be present but they are less pronounced. INFRA-RED ABSORPTION One-layer and two-layer hydrates of Na-vermiculite were examined as oriented aggregates and the infra-red spectra were recorded for two angles of incidence 0 and 40". In the recorded spectra the relatively sharp peaks at the higher frequencies which are produced by the structural OH groups of the silicate layers are easily distinguished froin the very broad absorption bands at lower frequencies which are due to the interlayer water.The spectra of the two-layer hydrate of Na-vermiculate (fig. 3) shows absorption at 3,675 cm-l (due to structural OH groups) and a very broad band with a maximum cm-l FIG. 3.-Infra-red absorption spectra of an oriented aggregate of the two-layer hydrate of Na-Vermi- culite at two angles of incidence (a) 0" ; (b) 40". at 3,415 cm-l which corresponds to the stretching vibrations of the interlayer water. There is no appreciable change in the intensity of this band with the angle of incidence indicating that there is no preferential orientation of the interlayer water molecules. J . HOUGARDY J . M . SERRATOSA w. STONE AND H . VAN OLPHEN 193 For the one-layer hydrate (and at lower water contents of the clay) the maximum of the band corresponding to the stretching vibrations of water molecules moves to higher frequencies i.e.3,450 cm-1 (fig. 4). The band associated with the deformation Do 3600 3400 3200 cm-' FIG. 4.-Infra-red absorption spectra of an oriented aggregate of the one-layer hydrate of Na- vermiculite at two angles of incidence (a) 0" ; (b) 40". vibration which appears at about 1,600 cm-l also shows a shift although in the direc- tion of lower frequencies (not shown). Both displacements indicate weakening of the hydrogen bonding of the water molecules and therefore that the association ion- dipole is the more prevalent. For inclined incidence (fig. 4 curve (b)) a significant increase of intensity of the water band is observed indicating that for these low water contents the molecules are not randomly disposed but do adopt a preferential orienta- tion in the interlayer space.The authors thank Mr. C . T. Deeds and Mr. W. K. Lumb for their assistance in the isotherm X-ray and density experiments Dr. K. E. Manchester for measuring the heats of immersion Prof. J. J. Fripiat for stimulating discussions on the n.m.r. work. J. Hougardy is indebted to IRSIA for a Ph.D. grant. C. T. Deeds and H. van Olphen A h . Chem. Series 1961,33,332. D. M. Anderson and P. F. Low Proc. SoiZ Sci. SOC. Amer. 1958 22 99. H. van Olphen J. Colloid Sci. 1965 20 822. H. van Olphen Proc. Int. CZay Con$ Tokyo (Israel Universities Press Jerusalem) 1969,1,649. W. F. Bradley and J. M. Serratosa CZuys CZay Minerals 1960,7,260. J. Graham G. F. Walker and C. W. West J. Chem. Phys. 1964,40 540. ' M. Dupont Thhe de Doctorat (Grenoble 1965). * D. E. Woessner and B. S. Snowden Jr. J. Chem. Phys. 1969 50 1516. R. Touillaux P. Salvador C. Vandermeersch and J. J. Fripiat Israel J. Chem. 1968 6. 337. SP 1 -G

ISSN:0370-9302    

DOI:10.1039/SD9700100187

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Neutron Scattering Studies of Hydrated Layer Silicates B Y s. OLEJNIK G. c. STIRLING AND J. w. WHITE Oxford University Physical Chemistry Laboratory Oxford England Received 17th June 1970 Slow-neutron scattering spectrometry has been used to investigate the dynamics of interlamellar water molecules in layered silicate clay minerals. Experiments on hydrated lithium vermiculite give information in three regions. At energy transfers between 200 and 800 cm-l vibrational bands associated with hydrogen motions in the silicate lattice and in the water are distinguished. In the low-frequency region ca. 15-1 50 cm-' a broad band of scattering characteristic of water emerges as the water content of the clay increases. Line-broadening effects arising from small energy transfers are indicative of diffusional processes in the water layers but improved technique is required to make this certain for the thinnest films.The data are analyzed with particular reference to the thickness of the interlamellar water layers and indicate significant changes in the water structure for water thicknesses of ca. 1-2 molecular layers. Similar results obtain in experiments with sodium vermiculite and with lithium and sodium montmorillonite. The hydration layer confined between the silicate sheets in certain clay mineral structures provides an excellent model system for the study of thin aqueous films. In favourable cases water layers can be established under well-defined conditions with thicknesses readily variable from zero to some 100 A. In vermiculite and montmorillonite the silicate sheets are separated by hydration layers containing exchangeable cations.The nature of the water in these layers has been the subject of considerable speculation.' It is generally conceded that the silicate surface confers a certain degree of order on the water structure but the causes and range of this ordering are matters of some debate.2* Various techniques have been employed to study the dynamics of water molecules in the hydration layer. N.m.r. line-shape observations on vermiculite and montmorillonite indicate that proton mobility is dependent on the water content of the clay.4 Pulsed n.m.r. on montmorillonite + H 2 0 and + D,O systems has been used to show that the interfacial water is preferentially ~ r i e n t e d . ~ ~ Self-diffusion of the water protons in hydrated vermiculite has been studied by pulsed spin-echo n.m.~-.~ Yet these and other experiments have not led to a full understanding of the interlamella water structure and in particular the relative effects of the charged surfaces and dissolved cations in producing order have to be evaluated.Neutron scattering spectrometry is well suited to study the dynamics of water in clays since the silicate structure remains essentially " transparent " to the neutron beam owing to the low scattering cross-section of aluminosilicate compared to that of hydrogen. Additionally the spectra record not only the frequency spectra of molecular modes but also the intermolecular and diffusive motions for included water molecules. The last are most strongly altered by structuring fields and can be studied here because of the high energy resolution available.The neutron results can be used in two different ways. Relatively large energy transfers up to about 800cm-l arise from vibrational and rotational motions of hydrogen groups. In some respects this inelastic scattering is analogous to Raman spectroscopy but has the advantage that hydrogen motions are selectively observed 194 S . OLEJNIK G . C . STIRLING A N D J . W . WHITE 195 and without the restrictions of optical selection rules. Secondly very small energy transfers (ca. 0.1-10 cm-I) can be measured with neutrons and coupled with the angular dependence of scattering they provide information on diffusional behaviour of the scattering nuclei in this case the water protons. We have shown that the neutron method can be used to determine diffusion coefficients in water and aqueous solutions with comparable accuracy to move conventional technique^.^ Naumann Safford and Mumpton measured the neutron inelastic spectra for a range of layer silicate minerals,l0 with particular reference to OH motions in the 200-1,000 cm-l region.In some preliminary experiments with a line-source incident neutron beam we investigated water motions in lithium vermiculites with relatively large water spacings and showed how the self-diffusion coefficient of the water varied with the clay water content.’l In the present experiments we concentrate on thinner films mainly in the region of 1-2 molecular layers and extend the work to some other minerals. THEORETICAL BACKGROUND For an incident neutron spectrum with a narrow energy spread the observed intensity distribution of the scattered neutrons from an incoherent scaiterer like hydrogen as a function of the energy transfer co and the solid angle of scatter LR is identical to the differential scattering cross-section a20/dR80 - - bzk -!-J[exp (i(Q .r-cot)) . Gs(r t ) dr dt. a20 anam ko2n Here b is the scattering length of the nucleus; k ko are the scattered and incident wave vectors of the neutron respectively Q = k -ko and is the momentum transfer in the collision expressed in radians/s and G,(r,t) is the space-time autocorrelation function describing the motion of the atom containing the scattering nucleus. G,(r,t) is the chance of finding a proton at r at time t if it was at the origin at time t = 0. It is defined by eqn (2) Characteristics of G,(r,t) have been discussed by Egelstaff.” The functional relation- ship between r and t may be simply sinusoidal as for a vibrator or it may have a complicated form as for diffusive motions in liquids.For liquids a number of models have been proposed to describe the molecular motions and hence G,(r,t).13 Using these it is possible to analyze the shape and angular dependence of the scattering to obtain the self-diffusion coefficients associated with a particular scattering centre. Models have been discussed e.g. by SjOlander.l4 For a molecule undergoing simple Fick’s law diffusion the differential scattering cross-section is given by eqn (3) a20 DQ2 = const. anam ( D Q ~ ) ~ +a2’ (3) where D is the self-diffusion coefficient defined by Fick’s Law. This model is adequate for long-time behaviour (i.e. small values of Q and o) and gives rise to an energy broadening which increases with the momentum transfer squared AE = 2A0Q2.(4) 196 NEUTRON SCATTERING OF LAYER SILICATES EXPERIMENTAL Young River (Western Australia) sodium vermiculite was treated following the method of Posner and Quirk," to produce lithium- and sodium-exchanged materials. After treatment films approximately 4 cm x 5 cm in area were prepared by sedimentation and drying of a carefully dialyzed suspension. They were equilibrated at known constant humidities to yield specimens of different water contents and hence water layer thicknesses. For neutron scattering experiments samples were quickly sealed in aluminium casettes. Sample weights were measured before and after an experimental run which typically took about 12 h.During this time the sample was surrounded by an atmosphere of dry helium in the sample changer of the spectrometer. All samples were prepared to give a total scattering of between 7 and 9 % of the incident beam intensity. Basal plane spacings were meaured by X-ray diffraction immediately after recording a neutron spectrum. The clay specimen was transferred to a glass slide and sealed under a thin Polythene film to prevent any change in moisture content. Spacings were measured with a Philips PW 1050 diffractometer using Ni-filtred Cu radiation. Samples were ignited at 310°C and weighed to give water contents. Neutron spectra were measured on the 6H time-of-flight spectrometer on the reactor DIDO at A.E.R.E. HarwelI.16 Beryllium-filtered neutrons are monochromated by a phased chopper system and energy spectra of the scattered neutrons measured at nine angles between 18 and 90" to the incident beam direction.The incident neutron energy chosen for the present experiments was 4.60 meV (4.22 A) with resolution 0.55 meV (full- width at half-maximum). All experiments were carried out at ambient temperature 23 f1"C. Experimental data are normalized against a vanadium standard run at the same time as the sample and corrected for background and counter efficiency. The corrected data are displayed as " time-of-flight " spectra where corrected counts are shown against the scattered neutron reciprocal velocity ( p m-l). Inelastically scattered neutrons occur as an energy- gain spectrum (i.e. towards lower reciprocal velocities) owing to the low energy of the incident neutron beam.To obtain AE the energy broadening (taken to be Lorentzian following eqn (3)) and the incident spectrum (Gaussian) are deconvoluted using standard tables. Slopes of (AE Q') plots may then be used to estimate D. For quasi-elastic scattering km ko and Q = (47114 sin (0/2) where II is the neutron wavelength and 8 is the angle of scatter. For water and aqueous solutions this is a reliable procedure provided slopes of AE against K~ are measured at sufficiently low Q value^.^ RESULTS Fig. 1 shows time-of-flight spectra measured at 90" scattering angle for neutrons scattered from (a) lithium vermiculite dried at 310°C and (b) liquid water. In these spectra the elastic peak is centred at 1,065 ps m-l with neutron energy-gain spectra deployed to the left. Energy transfers are indicated in cm-l.The spectra can be conveniently described in three parts (i) MOLECULAR MoDES.-These occur mainly in the region above 200 cm-I energy transfer and in the present experiments up to about 800cm-l (energy levels above this are not observed owing to the low populations of upper levels arising from the Boltzmann distribution). The scattering results from hydrogen motions which in the dry clay reflects vibrations of OH groups in the clay structure. These bands correspond to those observed in infra-red and Raman spectroscopy and have been discussed for neutron spectroscopy by Naumann et al. O A particular advantage of the neutron method is the selective scattering by hydrogen which can often facilitate the assignment of complex molecular spectra. S . OLEJNIK G .C. STIRLING AND J . W. WHITE 197 In the liquid water spectrum the band at ca. 480cm-I arises from hindered torsional oscillations of the water molecules. The position and structure of the band is significantly affected by the presence of various ionic solutes,17 thus providing information on the water structure in electrolyte solutions. The hydrogen-bond stretching frequency which occurs at 175 cm-l in optical spectra has a low neutron scattering intensity and was not resolved in the present experiments. approximately 15 and 150 cm-l is moderately intense in the neutron spectrum of liquid water (fig. l(a)). A broad spectrum of vibrations in this region is characteristic of liquids l8 and may arise from strongly damped intermolecular vibrations and torsions. For H20 at 23°C this spectrum has a broad maximum at ca.60cm-I. For D20 measured under the same experimental conditions the maximum is at (ii) LOW-FREQUENCY INELASTIC ScATTERING.-The frequency region between energy transfer (cm-I) neutron reciprocal velocity (micro secs. metre-') FIG. 1.-Neutron time-of-flight spectra angle of scatter = 90". (a) Lithium vermiculite dried at 310°C ; (b) water. ca. 40 cm-l. The peak position in H20 is almost independent of the scattering angle (and hence momentum transfer) but in D20 a noticeable shift of order 10 cm-' to lower frequencies occurs at high scattering angles.lg The light water scatters incoherently and the spectrum is a density of states curve weighted by (u.Q)~ where u is the hydrogen amplitude. The D20 spectrum is more akin to the " quasi-phonon" spectrum from polycrystalline and coherent scattering liquid metals.20 It supports the analysis given to the hydrogen spectrum relating it to cooperative intermolecular modes.(iii) QUASI-ELASTIC scAmmNG.-Small energy transfers arising from translational motions of the scattering nuclei lead to a broadening of the incident neutron energy spectrum as described earlier. In fig. l(a) the elastic peak has the same energy width as the incident neutron beam indicating that the scattering nuclei in the dry clay are fixed in the lattice. In fig. 1 (b) for liquid water there is a relatively large broaden- ing and it is from measurements in this quasi-elastic region that diffusional details can be obtained. To exploit these characteristics we have obtained spectra for a series of lithium vermiculites containing water spacings of different thicknesses.A point of some concern has been the swelling characteristics shown by the clay specimens. We have found that vermiculite films equilibrated for some time in a constant humidity atmosphere yield homogeneous samples with basal spacings up to about 15A (interlamellar spacing about 5~4 or two molecular layers of water). For larger 198 NEUTRON SCATTERING OF LAYER SILICATES spacings it is necessary to use direct wetting techniques. After wetting specimens were sealed in the aluminium casettes used for the neutron measurements. However subsequent measurements of total water contents and X-ray spacings showed that swelling behaviour could be erratic and uniform swelling was not always achieved. This effect has been noted before,21* 2 2 and requires that all specimens be carefully monitored both for water content and X-ray spacing.Because of the non-uniformity in swelling behaviour we do not report here any data on clay systems with large water contents. energy transfer (cm-') E Y .d 10 - 8 - b - 4 - 2- J t I ' r J I 0 5W 1000 ISW I neutron reciprocal velocity ps m-l clay = 0.079 ; (c) g HzO/g clay = 0.135 ; ( d ) water. FIG. 2.-Inelastic neutron time-of-flight spectra angle of scatter = 90". (a) dried clay ; (b) g H20/g In fig. 2 the inelastic spectra at 90" scattering angle for some homogeneous lithium vermiculites are shown together with those of the dry clay and water. Descriptions of these specimens are included in table 1. There is a continuous gradation in the spectra as the water content of the clay increases.Molecular modes associated with OH motions in the clay which occur between 200 and 800cm-' (fig. 2(a)) are progressively submerged as the water spectrum becomes more dominant at higher water contents. In the lower frequency region 15-1 50 cm-l the characteristic scattering of liquid water also builds up significantly by the time the interlamellar layer is about two molecular layers of water thick. Further evidence for incipient water behaviour in these thin layers can be obtained from quasi-elastic line broadening measurements. S . OLEJNIK G . C. STIRLING AND J . W. WHITE 0 6 05 0.3 0 2 - 0.1 199 11/1141’ - - I/ I I I I TABLE 1 .-DIFFUSION DATA FOR LITHIUM VERMICULITE SAMPLES water content D sample g HzO/g clay d(001) 8 cm2 s-1 X 105 a 0 dry clay 0 b 0.079 12.75 0 C 0.114 13.3 0.05 d 0.135 14.05 0.1 1 e - water 2.14 Fig.3 shows line broadenings measured for a number of lithium vermiculite samples as a function of momentum transfer squared. At present these results cannot be taken as complete evidence of liquid behaviour in the thinnest layers since a problem with samples having low water contents is the relatively low signal/ background ratio where “ background ” refers to scattering from the clay lattice. There is negligible line broadening from this scattering (see fig. 3(a)) and this may well modify broadening effects arising from H,O scattering. Self-diffusion co- efficients derived from data in this region are likely therefore to represent minimum values. At higher water contents the scattering is almost exclusively from the interlayer water and the problem does not arise.Table 1 gives values of D derived directly from the plots shown in fig. 3 together with details of the clay specimens. Similar results are given for some related systems in table 2. 200 NEUTRON SCATTERING OF LAYER SILICATES TABLE 2.-DIFFUSION DATA FOR VERMICULITE AND MONTMORILLONITE SAMPLES water content D material g HZO/g clay d(OO1). A cm2 s-1 x 105 Na-vermiculite 0.080 13.71 0.06 Na-vermiculite 0.111 13.92 0.06 Na-vermiculite 0.185 14.48 0.17 Li-montmorillonite 0.230 14.5 0.19 Na-montmorillonite 0.228 14.53 0.25 DISCUSSION We have demonstrated some of the different ways in which neutron scattering can be applied in the study of H motions in clay minerals with particular reference to water dynamics in the thin films in layered silicates.A unique feature of the neutron method is the time-scale of observation ca 10-12-10-11 s which means that self-diffusion coefficients can be estimated independently of macroscopic imper- fections in the clay structure. In conventional “ long-time ” techniques (10-3-10-2 s and longer) allowance has to be made for pores and cracks in the structure which can modify the diffusion coefficient.’ In films about 1-2 molecular layers of water thick water self-diffusion coefficients are an order of magnitude lower than those in bulk water. Within the limits of the present experiments we are unable to detect significant differences between water in vermiculite and montmorillonite containing lithium or sodium as the exchangeable cations. Broadly the self-diffusion coefficients appear to vary as the interlamellar spacing even within this relatively restricted range.Future work will concentrate on the various factors influencing these values. In this work we have identified various problems involved in the study of water dynamics in hydrated clay minerals. Foremost is the necessity for careful charac- terization of the sample under investigation a point stressed e1sewhere.l. An important feature is the homogeneity of the sample i.e. is all the absorbed water distributed between uniformly expanded silicate sheets? If not a fraction of the water present may display nearly bulk properties which may lead to spurious inter- pretation of experimental results. Delicate experimentation may be required to distinguish different effects in such cases. In the present work we have taken care to check overall water contents and X-ray spacings and have reported only those results obtained for homogeneously swelled specimens.In these experiments it has been possible to measure quasi-elastic line broadenings corresponding to self-diffusion coefficients as low as 0.05 x cm2 s-l. At this level the water content of the clay is sufficiently low that appreciable scattering from OH motions in the clay can be incorporated in the elastic peak. This scattering contributes an unbroadened component to the peak resulting in a decreased value of D for the interlamellar water. Future experiments with very small spacings will be required to account for this factor. We thank Prof. J. P. Quirk for the clay samples and the Dept. of Soil Science Oxford for advice and facilities to prepare samples.J. Graham Reu. Pure A&. Chem. 1964 14,81. P. F. Low A&. Agronomy 1961,13,269. R. T . Martin Clays and Clay Minerals Proc. Nat. Conf. 1962,9,28. J. Graham G. W. West and G. F. Walker J. Chem. Phys. 1964,40,540. D. E. Woessner and B. S. Snowden J. Colloid Interface Sci. 1969 30 54. D. E. Woessner and B. S. Snowden J. Chem. Phys. 1969,!50 1516. B. D. Boss and E. 0. Stejskal J. Colloid Interface Sci. 1968 26 271. S . OLEJNIK G . C. STIRLING AND J . W. WHITE 201 P. A. Reynolds and J. W. White Disc. Furuday SOC. 1970 48 131. G. C. Stirling and J. W. White unpublished. 1966 14 367. R. J. Hunter G. C. Stirling and J. W. White unpublished. l o A. W. Naumann G. J. Safford and F. A. Mumpton Clays and Cluy Minerals Proc. Nut. Conf. l 2 P. A. Egelstaf€ An Introduction to the Liquid State (Academic Press London 1967). l3 G. H. Vineyard Phys. Rev. 1958,110,999. l4 A. Sjolander ThermuZ Neutron Scattering ed. P. A. Egelstaff (Academic Press London 1965) l 5 A. M. Posner and J. P. Quirk Proc. Roy. SOC. A 1964 278 35. l6 L. J. Bumce D. H. C. Harris and G. C . Stirling U.K.A.E.A. Report AERE-R 6246 (H.M.S.O. l7 G. J. Safford P. S. hung A. W. Naumann and P. C. Schaffer J. Chem. Phys. 1969,50,4444. * K.-E. Larsson Neutron Inelastic Scattering (Copenhagen 1968) (I.A.E.A. Vienna 1968) chap. 7. 1970). vol. 1 p. 397. p. 463. l 9 B. K. Aldred and J. W. White unpublished. 'O S. J. Cocking Neutron Inelastic Scattering (Copenhagen 1968) (I.A.E.A. Vienna 1968) vol. 1 " W. G. Garrett and G. F. Walker Clays and Chy Minerals Proc. Nut. Con$ 1962,9 557. 22 K. Norrish and J. A. Rausell-Colom Cluys and Clay Minerals Proc. Nat. Conf 1963,10 123.

ISSN:0370-9302    

DOI:10.1039/SD9700100194

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

GENERAL DISCUSSION Prof. A. Watillon (Universite‘ Libre de Bruxelles) said In the first part of his paper Dukhin considers the presence of a stagnant layer at the solid-liquid interface. This idea is supported by Wijga and implicitly by Mossman and Mason.2 More recently at a glasslwater interfa~e,~ based on the same assumption we interpreted simultaneous measurements of [ potentials and of surface conductance. For K-k Ba2+ and La3+ ions used as counterions we obtained for the stagnant layer a thickness in the range 60-80A. This fact suggests that the proposed double-layer model for glass seems to be reasonably valid. In the last part of his paper Dukhin proposes an interpretation of the relaxation of the dielectric constant of colloid suspensions based on his theory of double-layer polarization combined with Maxwell-Wagner relaxation theory of heterogeneous systems.He considers that this approach could interpret experimental results such as those of Schwan on polystyrene latex suspensions. In our laboratory DeBacker observed relaxation phenomena on Pyrex spheres (diam. = 40pm) in very dilute electrolyte solutions. Taking into account their larger size and the lower ionic strength this system can be compared with Schwan’s polystyrene latices. DeBacker calculated on the IBM 7040 computer the complex dielectric properties of the system against frequency. Therefore he used the Maxwell equation6 twice first to deduce the admittance of the particle embedded in its double layer secondly to obtain the admittance when the particle surrounded with its atmosphere is embedded in the dispersion medium.The calculations fit the observed data well. Finally we wonder if polarization effect is large enough to explain the major part of relaxation processes in colloid systems. Prof. S. S. Dukhin (Inst. Colloid and Water Chem. Kiev) (communicated) In reply to Watillon in the Dukhin-Shilov theory of concentration polarization of a thin double layer of spherical particles allowance is made for the appearance of difference in ion concentration beyond the double layer due to tangential flows of the diffuse layer ions and ion exchange between the double layer and contiguous volume of electrolyte. The effective relaxation time of this polarization process is z = a2/2D since this time is required for uniformity of concentration through diffusion at a distance of the order of a.If the frequency is such that cuz is very great the concentration difference does not appear during a single cycle and the concentration mechanism of polarization is no longer the determining factor. With very large co O’Konski’s theory holds true ; in this the effect of the double layer is taken into account through the surface conductivity. Although Watillon does not mention the frequency used by DeBacker it may be presumed that coz was very high in his experiment since large particles with diameter 2a = 40 pm were used i.e. z = a2/2D- 1 s was very large. These experi- ments cannot therefore be a test of our theory. P. W. 0. Wijga Thesis (Utrecht 1946). A. Watillon and R. DeBacker J. Electronal. Chem. 1970 25 181. P. H. Schwan G. Schwarz J. Maszuk and H. Pauly J. Phys.Chem. 1962,68,2626. R. DeBacker and A. Watillon to be published. J. C. Maxwell A Treatise on Electricity and Magnetism (Constable and Co. 1891). C. T. O’Konski J. Phys. Chem. 1960,64,605. 202 * C. E. Mossman and S. G. Mason Can. J. Chem. 1959,37 1153. GENERAL DISCUSSION 203 Prof. H. Pfeifer (Karl-Marx Universitut) (cornmimicated) The important state- ment in the paper by Clifford Oakes and Tiddy that for the lamellar phase the average correlation time of water molecules at 8.3 A will be about three orders of magnitude longer than in ordinary water depends only on one point in fig. 9. (The last circle on the right-hand side.) Since these relaxation rates (Tid) were calculated from measured rates which are not very different (fig. 8) the error may be very high and the decrease of Ti$ for a thickness of water layer less than 8.3 A questionable.Mr. J. Clifford (Unilever Rex. Port Sunlight) said In reply to Pfeifer in presenting our paper it was pointed out that subsequent experiments had shown that the errors involved in calculating the final point in fig. 9 were indeed large and that the figure of 2 x s for the average correlation time for rotation of water molecules for a layer of thickness of 8.3A should be regarded as only very approximate. This does not affect the general conclusions in any way. Dr. G. Peschel (Universitiit Wurzburg) said With regard to the paper by Clifford et al. in a former paper dealing with nuclear magnetic resonance relaxation times of water protons in water layers near polyvinylacetate the authors found a molecular long range effect for the water molecules which might be due to the highly porous structure of the polyvinylacetate used since water molecules might be trapped in the pores and probably lead to a pattern of surface centres which can form hydrogen bonds with adjacent water layers but which do not fit into the water structure.In this way a highly disturbed surface zone may be produced. The polystyrene used in the present paper has a much smoother surface whch should not give rise to a significant perturbation of adjacent water layers because the potential wells of the surface are less pronounced so that most of the vicinal water molecules can easily arrange in such a way that the mismatch between surface zone and bulk liquid is negligibly small. Would the authors’ explanation tend in this direction? Dr.A. L. Smith (Liverpool Polytechnic) said The flocculation kinetic results cited in ref. (16) of the paper by Clifford Oakes and Tiddy refer to narrow size distribution polystyrene latex dispersions of particle diameter 0.37 pm flocculated in excess electrolyte. The method of particle counting in a flow laser ultra-microscope was used. Some variation of derived second-order rate constant with initial particle concentration was found and the rates quoted were obtained by extrapolation to zero particle concentration where the Smoluchowski theory of rapid flocculation is most likely to be valid. At 25°C the particle counting method gave a rate 50 % of the Smoluchowski figure. The temperature coefficient of the second-order rate constant was found to be equal within experimental uncertainty to that predicted by the Smoluchowski expression 4 kT/3r for the extrapolated zero particle concentra- tion rates but rose significantly at high particle concentrations.Sols not extensively dialyzed or passed through an ion-exchange column further showed a variation of flocculation rate with particle charge even in 100 mol MgS04 aq. with maximum rate near the zero point of charge. This effect is presumably due to the presence of residual adsorbed layers of surface-active material and can be reproduced by the addition of such material to dialyzed sols. Prof. J. Th. G. Overbeek (University of Utrecht) said At the Faraday Society Discussion at Nottingham in 1966 Deryaguin pointed out that Smoluchowski’s G. A. Johnson S . M. A. Lecchini E. G. Smith J. Clifford and B. A.Pethica Disc. Faraduy SOC. 1966 42 120. 204 GENERAL DISCUSSION theory does not take into account that the mobility of two particles when they are close together is smaller than when they are far apart. The remark has been worked out by Deryaguin and with more precision by Honig (Philips Eindhoven) and Roe- bersen and Wiersema (Utrecht). The effect is so large that the mobility becomes proportional to h the distance of separation of the spherical particles and therefore they would never touch (h = 0) unless the van der Waals attraction pulled them together. The final result taking both the hydrodynamic and the van der Waals effect into account is that for any acceptable van der Waals-Hamaker constant rapid flocculation is slower than the Smoluchowski rate by about 10-50 %. Dr.E. Willis (Unilever Res. Port Sunlight) (communicated) Mr. A. Lips and I have studied the flocculation of a 126 nm diam. polystyrene latex in excess electrolyte at 25°C by light scattering. The particle concentration was 1 x lo9 ml-l. Using a model which describes the scattering properties of aggregates in terms of optical interference between their constituent primary particles we obtained agreement of better than 5 % between the theoretical and the measured (intensity time) curves for a range of scattering angles between 30 and 135" for times up to the half-life. We found the second-order rate constant to be 68 % of the Smoluchowski value in good agreement with Smith's result. It would appear from these results and from Overbeek's comments that bound water layers make little contribution to the colloid stability of polystyrene latices.Dr. M. L. Hair (Xerox Corporation Rochester U.S.A.) said In fig. 4 Hougardy et al. present an infra-red spectrum (b) which clearly shows the presence of two different types of hydroxyl species at 3,400 and 3,550cm-l. This spectrumis attributed to a one-layer hydrate of Na-vermiculite containing oriented water molecules. In fig. 3 the infra-red spectrum of the two-layer hydrate shows only a single band at 3,400 cm-l and assigned to non-oriented water. I would ask the authors whether the two bands observed with the one-layer hydrate are in fact due to a mixture of oriented and non-oriented water or is there an alternative explanation ? The authors comment that the deformation band due to the water which appears at about 1,600 cm-I is shifted to lower frequencies.Does this remain as a single band or are two bands observed due to two different species? Dr. J. M. Serratosa (Inst. Edafologia y Biologia Vergetal C.S.I.C. Madrid) said In reply to Hair the presence of two bands in the region of the OH stretching vibrations does not necessarily indicate the existence of two different species of water. In fact a water molecule has three normal modes of vibration symmetric stretching vl bending v2 and asymmetric stretching v,. In many cases the absorption bands are broad and the v1 and v frequencies cannot be differentiated. In the one layer complex we believe that only one kind of water molecule exists. The water molecule either maintains its symmetry or alternatively the two OH bonds of the molecule are associated by hydrogen bonds of different energy.This alternative precludes one from reaching a definitive conclusion on the orientation of the molecules in the interlayer space. Only one band corresponding to the bending vibration v2 is observed which indicates one species of water molecule present. Prof. H. Pfeifer (Karl-Marx Universitat) (communicated) It would be of interest to know how large is the influence of constitutional hydroxyls and paramagnetic centres on spin-lattice relaxation times Tl and in what manner van Olphen Stone Hougardy and Serratosa were able to correct this influence (error of this method). GENERAL DISCUSSION 205 I would also ask if the authors have some opinion about the observed change in relaxation times in going from two-layer hydrate to one-layer hydrate (fig.2). It seems that the interaction energy is lower (decreased intermolecular interaction) and that the correlation time is unchanged (no shift of the minima-which are said to be present in the one-layer hydrate-but less pronounced). The latter statement would be in contrast with the experimental results of the free induction decay measure- ments ('' it is concluded therefore that for the two-layer hydrate the orientation effect is obtained at a much lower temperature than for the monolayer hydrate ") and of their infra-red absorption spectra. Dr. W. Stone (Inst. Sci. de la Terre) said In reply to Pfeifer all T1 measurements show a plateau at low temperature. This effect was considered as resulting from paramagnetic impurities. In these preliminary experiments the low-temperature relaxation rate was then simply subtracted from the observed rates.The same procedure was utilized to correct for the constitutional protons. These corrective factors represent approximately 10% of the observed rates in the region of T1 mini- mum. Concerning the slightly lower Tl values observed in the two-layer hydrate case this fact is not in contradiction with the new results obtained on the oriented flakes. Preliminary results on oriented flakes of Na-vermiculite by pulsed n.m.r. show (i) that for the two-layer hydrate an organization of the water molecules exists even at room temperature. This result agrees with the entropy calculations. The observed free induction decay is angle dependent (i.e. varies as the angle between the c-axis and the direction of the large magnetic field is changed).By examining this dependence it can be concluded that for temperature above O"C the H-H vector of the water molecule is axially symmetric about the c-axis. The angle 6 between these two vectors is approximately 50". It is also found that at low temperature (around - 50°C) the organization of the water molecules changes the symmetry axis is no longer the c-axis. (ii) That for the one layer hydrate the H-H vector of the water molecules is also axially symmetric about the c-axis (9 again about 50") and that this symmetry is retained whatever the temperature. (iii) That the spin-lattice relaxation times appear to be slightly angle-dependent. Dr. R. G. W. Anderson and Dr. J. W. White (Phys. Chern. Lab. Oxford) said To complement the studies of water diffusion near vermiculite surfaces we present some neutron inelastic scattering measurements of water diffusion in aqueous sols made from fumed silica.This system is favourable for the neutron method because the very low scattering from silica compared to water (scattering cross-section Si02 = 12x cm2) allows the water dynamics alone to be observed with €ugh signal-to-background. For a 5 % mass/mass H20/Si02 dispersion the water scattering is still about 5 times that from silica. Six mass ratios from pure water down to 5 % have been studied. Values of the water diffusion coefficients as a function of coverage have been measured from the quasi-elastic neutron scattering and evidence for jump diffusion at low ratios H20/Si02 has been found. The inelastic scattering spectra reveal some changes in the vibration frequency spectrum of water near 200 cm-l associated with the adsorption.on the DIDO reactor at A.E.R.E. Harwell. For all spectra the scattering was less than see S. Olejnik G. C. Stirling and J. W. White this Discussion. L. Bunce D. H. C. Harris and G. C. Stirling U.K.A.E.A. Research Report. no. 26246 (H.M.S.O. 1970). cm2 H20 = 170x Measurements were made at 23°C with the 6H cold neutron spectrometer 206 GENERAL DISCUSSION 5 % to minimize multiple scattering events and from the straightness of the Debye- Waller plots (fig. 2b) incoherent scattering from hydrogen was dominant. The whole experiment was designed to estimate the maximum structure-forming effect of silica on water and unannealed Cabosil HS-5 was used. This material has a quoted particle size of ca.70 A (700 nm) and a B.E.T. surface area of 325+25 m2 g-'. The results of Clifford and Lecchini indicate that a large structuring effect could be expected for this type of unannealed material. Samples were prepared by mixing weighed quantities of water and Cabosil for several hours in a ball mill the resulting gel was transferred quickly to thin-walled aluminium sample containers which were then sealed to avoid changes in water content. Samples with mass/mass H20/Si02 greater than about 60 % were seen to be efflorescent by allowing them to stand in a sealed glass vessel for several days when water distilled out. The composition of all samples was checked by ignition and weighing. 2.5- LO - (a) 0=18" 15 - 4 * . .. .. .. . . . . I;O - 0 . . E .gj 0.5 - ." c o I I I . . I neutron time of flight ps m-' FIG.1.-Neutron time of flight spectra for (a) mass/mass(HzO/SiO2) = 30 %; (b) Cabosil outgassed at 500°C. Neutron time of flight spectra taken at scattering angles 0 = 18 and 90" to the incident beam direction are shown in fig. 1 for 30 % by mass water in Cabosil and for a Cabosil sample outgassed at 350°C and measured in an aluminium can. The spectrum of the 30 % water gel resembles that of liquid water (ref. (l) fig. l(b)). The water torsion band near 480 cni-l (ca. 300 p s m-l) and the hindered translations near 60 cm-1 appear strongly. The smaller quasielastic broadening (peak near 1100 p s m-l) for the gel is alone evident. The intensities in the spectra for the gel J. Clifford and S. M. A. Lecchini Wetting (S.C.I. Monograph no. 25) 1967 p. 174. GENERAL DISCUSSION 207 and the silica are separately normalized to vanadium.To siinulate the SiO contri- bution to the gel spectrum the ordinate in fig. l(b) should be divided by 30. The contribution is thus negligible. At the lowest water content studied (6 %) the torsion and translation peaks shifted to 460_+20 and 50+ 10 cm-1 respectively from ca. 480+20 and 58f10 cm-l in liquid H20. The hindered translations are now very broad. In fig. 2 we show the angular dependences of the quasielastic scattering peak widths and intensities. The squared momentum transfer in the scattering collision Q2 is related to the scattering angle 0 and neutron wavelength A by Q2 = (16n2/A2) sin2 (8/2) (1) and for moleculcs undergoing continuous diffusion obeying Fick's Law the line broadening AE is given by AE = 2hDQ2 (2) where D is the diffusion coefficient of the scattering molecule.By increasing the scattering angle and hence Q the neutron is allowed to observe the diffusion on a progressively shortening time scale between 10-I' and s. It can be seen from fig. 2(a) that at long observation times (small Q2) an almost linear relation between AE and Q2 is obtained as expected from eqn (2). For pure water the value of D found on the longest neutron time scale agrees with the macroscopic value from tracer and nuclear magnetic resonance spin echo studies.l For the gels the limiting D values are lower and depend on the water/silica ratio (see fig. 3). This behaviour is analogous to that of some dissolved ions such as Li+ which reduce D for Others (e.g. I-) may increase it but the diffusion constants measured macroscopically and by neutrons agree in both classes of solutions.We can therefore use the neutron measurement of D at least empirically as an indication of the extent of the water structure-forming tendency of a solute. Fig. 2 and the concentration dependence of D (fig. 3) indicate a continuous decrease in water diffusion constant as a function of increased silica content in the gels. At the same time the curvature at large Q values in fig. 2(a) shows that as the concentration of silica increases the water diffusion becomes less continuous and more like jump-diffusion. Again the water behaviour is analogous to water in strong salt solutions at low temperature^.^ A second measure of the restriction of water motions at high silica concentrations is provided by fig.2(b). For an incoherent scattering solid the elastic scattering intensity is related to the momentum transfer Q by the Debye-Waller factor exp [-2W] = exp -[Q2(u2)] (3) where <u2) is the mean square vibrational amplitude of the scattering atom. The log (peak area) is linear in Q2 with a slope proportional to this mean square vibrational amplitude. For a liquid where the atomic displacements are not bounded in time this formula is not correct but Debye-Waller plots still give a qualitative measure of the net displacement on time scales with an upper bound defined approximately by the observation time of the neutron. G. C. Stirling and J. W. White Mol. Phys. to be published. G. J. Safford P. S. Leung A. W. Naumann and P. C. Schaffer J . Chem. Phys. 1969,50,4444. J. W. White Deutsche Bunsen Gesellschajl Proc.(May 1971). G. J. Safford Deutsche Bunsen Gesellschaft Proc. (May 1971). (momentum transfer)2 Q z k 2 mass % water D (cm2 s-l) A 100 2.2 x 10-5 x 0 1.7 x 10-5 e 30 LOX 10-5 (4 I 2 3 (momentum transfer)2 Q 2 k 2 Debye-Waller plot D. W. factors 0 3 0 % m/m H20/Si02 0.33 x ,60 % m/m H20/Si02 0.42 0 8 0 % m/m H20/SiO2 0.45 (6) Fro. 2.-Quasi-elastic scattering dependence on squared momentum transfer and on water content A 100 % H20 ; m/m H20/Si02 ; 0 80 % ; x 60 % ; e 30 %. (a) Energy broadening as a function of Q2 ; (b) " Debye-Walier " plots of the quasi-elastic intensity. GENERAL DISCUSSION 209 Mathematically the Debye-Waller factor for a solid is a frequency integral over the whole displacement (vibration and rotation) frequency spectrum from cu = 0 to w = C O ~ eqn.(4) where N is the number of atoms of mass M and Tis the absolute temperature of the sample. For neutron incoherent scattering from hydrogenous substances the hydrogen contributions to eqn (4) usually dominate. Also in practical measure- ments the lower limit of the integration in eqn 4 is set by the finite and short time of the neutron-molecule scattering event. This time z may be calculated approxi- mately for a solid from the neutron velocity and for a liquid by the method of Larsson and lies between 10-l1 and s for thermal neutrons. The lower integration limit is of the order co = 27c/z. For most solids the frequency spectrum between w = 0 and say w = 10'' radians s-I is insignificant and an error in 2Wproduced by integrating from co = 10'' rad s-1 to infinity is small.For a liquid the Debye-Waller factor is given by the same formula as above but an appreciable fraction of the translational and rotational modes may lie at frequencies below 0,. The Debye-Waller factor is sensitive to solute effects in a different way to the quasi-elastic broadening since the frequency width and the area of the spectrum are both involved. Provided no modes are shifted from above tor the Debye-Waller can be used to compare the vibration amplitudes as for solids. Close resemblance between the spectra in fig. 1 and those of water supports the evidence of fig. 2(b) that decreasing the water content of the gels from 80 to 30 % decreases the mean square displacements by about a factor of two. This behaviour correlates with the increased importance of jump diffusion rather than continuous diffusion in the system.Unannealed Cabosil and dissolved ions both induce structure in water. Their effectiveness and hence the range of the induced water structure may be compared first but considering the Cabosil as a normal solute and measuring the concentration dependence of the water diffusion coefficients for the two solutes. This reveals the mass effectiveness without regard to how many silica molecules are actually attached to water molecules. By using the specific surface area of the material (300 m2 g-l) and the area of a water molecule (9 A2)3 it can be shown that at surface saturation there are nearly 3 silica molecules to one of water. The number of effective (surface) silica molecules was therefore chosen as one third the number of moles of silica present.Fig. 3 shows the dependence of the water diffusion coefficient on (a) total mol ratio of Si02/H20; (b) effective mol ratio of Si02/H20 (one third (a)); and (c) mol ratio LiCI/H,O. When account is taken of the number of silica molecules exposed to the water silica has a stronger structure making effect on water than lithium chloride in accordance with the suggestion of Pethica. Unless some mechanism for proton transport other than centre of mass diffusion is operating here these data suggest by analogy with lithium chloride solutions that water- ordering occurs near silica surfaces out to several molecular layers but probably not for many 100 A. I. I. Gurevich and L. V. Tarasov Low Energy Neutron Physics (North Holland 1968) p. 145. K. E. Larson Thermal Neutron Scattering ed.P. A. Egelstaff (Academic Press 1965) chap. 8 p. 348. C. K. Hersh Molecular Sieves (Reinhold 1961) p. 49. 210 GENERAL DISCUSSION (number of molecules of S/number of water molecules) x lo2 (a) x S = total silica molecdes ; (b) 0 S = (total silica mOleCUleS)/3 ; (c) 0 S =; total lithium chloride (Stirling and White ") FIG. 3.-Comparison of the effect of Si02 and LiCl on the diffusion coefficient of water (a) total SO2 ; (b) surface SO2 in contact with water ; (c) lithium chloride. Dr. A. M. Hecht and Dr. E. Geissler (Lab. de Spectrome'tric Physique,* Grenoble Gare) said With regard to the paper by Olejnik et al. although there exists a large body of work devoted to clay+water systems apart from the results of neutron scattering experiments we have little detailed knowledge of the structure and motions of water molecules adsorbed on clay interfaces.We therefore undertook a series of pulsed and continuous wave measurements of proton magnetic relaxation times of water adsorbed in oriented films of synthetic fluorine-substituted sodium montmoril- lonite. The synthetic clay has the double advantage of not possessing structural protons since the hydroxyls are replaced by fluorines and of containing a small (5 10 p.p.m.) concentration of paramagnetic ions. Two samples of different hydra- tion were examined (I) containing a monomolecular layer and (11) containing a bimolecular layer of water between the clay platelets (hydrated respectively at 20°C with relative humidity 0.4 and at 27°C with relative humidity 0.8). The n.m.r. lineshape between 220°K and room temperature consists of a narrow central line flanked by a doublet.The latter is generated by the water molecules in a rotational mode around the c-crystal axis of the clay plates. With increasing temperature the central line increases in intensity at the expense of the doublet the variation being described by the activation energies 6.9 kcal/mol (sample I) and 9-10 kcal/mol (sample 11). The central line is Lorentzian characterized by a phase coherence time z? for the spins which remains approximately constant at 600 p s in the temperature ange 220-300 K. The shape of the line indicates the presence of motional narrowing Am of the static magnetic interaction between protons. The characteristic frequency of such a movement would be i/z = ~ ~ 2 ~ 107 s-1.In order to identify the various kinds of movement prevailing in the water layer we measured the proton spin-lattice relaxation time z,(T). The relaxation curves * Laboratoire associC au C.N.R.S. A. M. Hecht and E. Geissler J . ColZoid Interface Sci. to be published. GENERAL DISCUSSION 21 1 show a single z1 at each temperature which is practically independent of the orientation of the sample in the magnetic field. The measurement of z1 was also carried out as a function of the resonance frequency between 2.0 and 11.3 MHz. The minimum values of z1 at 11.3 MHz were 9.5 ms (sample I) and 9.0 ms (sample 11). Different activation processes were distinguished with cori elation times as follows SAMPLE I z = 2 x 10-14 exp (6.9/RT) s and zcg = 5 x exp (4.9/RT) s ; zCi = 2 x 10-14 exp (6.6/RT) s.SAMPLE I1 In both samples the relaxation time z1 shows a magnetic field dependence not yet understood. A lower energy activation process was also observed in sample 11 but its parameters have not been determined with accuracy. The pre-exponential time factor in z, is in good agreement with the rotational frequency of about 60cm-I seen in the neutron spectra of Olejnik Stirling and White and is not much different from the rotational period of the free water molecule. We associate this mode with the rotation responsible for the doublet in the C.W. spectrum. The activation energy in z, is too weak for a normal molecular diffusion process and the pre-exponential factor is short for a rotational motion. As vibrations hardly contribute to the spin-lattice relaxation we consider proton jumps between molecules as a possible mechanism.In such a movement the dipolar coupling time-averages to zero resulting in a narrow central line. It is notable that the activation energy for the intensity of the central line is identical to that of z for sample I (though not for sample 11 where more complex movements are expected to exist). If the central line is generated by such proton jumps the shortness of the pre-exponential time factor suggests that the movement is coherent over an area involving several hundreds of protons. Dr. B. A. Pethica (Unilever Res. Port Sunlight) said The neutron scattering data presented in the paper by Olejnik Stirling and White relate to low levels of hydration of vermiculite and show strong localization of the inter-lamellar water. The behaviow of water in larger amounts is of even greater interest in relation to the behaviow of liquid layers near a solid wall.The extra data provided in discussion concerning the properties of water in Aerosil+ water mixtures is consequently im- portant. It showed that the effect per mol of Si02 in reducing water diffusion is much less than that of LiCl. However if allowance is made for the fact that only some few percent of the Si02 groups are in contact with the water the effect of the Aerosil in restricting water diffusion is seen to be very great. Probably the Aerosil was un-annealed and it would be expected as mentioned in the introductory paper that annealed Aerosil would have a much smaller effect on the water for the same surface area in contact. If this is the case it would provide important evidence in judging data obtained on silica surfaces in general.Dr. M. M. Breuer (Unilever Res. Lab. Isleworth) said Some experimental results obtained by Dr. C. B. Baddiel Mrs. S. G. Clode and myself also suggest that solid surfaces affect the nature of liquid layers in their near vicinity and that these effects do not extend beyond a few molecular layers. We studied the molecular structure of adsorbed H,O and D,O on oriented films of synthetic polypeptides using polarized infra-red radiation. In particular we measured the dichroism of the i.-r. absorption 212 GENERAL DISCUSSION band at v = 3,500cm-1 of H 2 0 and 2,500cm-i of D20 molecules adsorbed on poly-L-alanine films in which the polypeptide molecules were in the a-helical conforma- tion and the helices were aligned in oriented parallel arrays.As the only polar groups present in the system capable of binding H20 (D20) molecules were the peptide linkages which in turn were aligned almost parallel to the helix axes it was possible that the H20 (D20) molecules would exhibit considerable dichroism. In fact we found at low relative humidities (r.h.) (0.30 %) i.e. within the monolayer domain that the band at 3,500 cm-1 (2,500 cm-l) showed some dichroism (1.3-L). At higher r.h. values however the dichroic ratio decreases which may be attributed to an increase in the water adsorbed (the overall H20 (D20) intensity increased) whilst the proportion of ordered molecules remained the same. It is difficult to judge accurately the central position of these broad bands but there appears to be little or no change in the absorption band frequency (_+ 10 cm-') compared to that of liquid H20 (D20).These results suggest that the structure of the first layer of H20 (D,O) is affected by the orientation of the peptide groups at the polypeptide surface most probably through interactions of the dipole moments of the peptide bonds with those of the water molecules i.e. C=O and N-H with H20. On the other hand the ordering effect of the polypeptide surface on subsequent layers of H20 (DzO) appears to diminish rapidly with distance from the interface. Finally I should mention that simultaneously with OUT work at the University of Edinburgh Dr. B. Malcolm also obtained similar results (except he also claims his spectral data indicated more than one species of H20) with oriented polypeptide films which were however prepared by a different technique and were of much lower dichroic rati0.l B. Malcolm Nature 1970 227 1358.

ISSN:0370-9302    

DOI:10.1039/SD9700100202

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Measurement of Viscosity of Liquids in Quartz Capillaries BY N. V. CHURAYEV V. D. SOBOLEV AND Z. M. ZORIN Dept. of Surface Penomena Institute of Physical Chemistry The Academy of Sciences of the U.S.S.R. Moscow. Received 30th April 1970 A method is developed for iiieasuring radii of microcapillaries and the viscosity of liquids in them. The viscosity of water benzene and carbon tetrachloride was measured in quartz capillaries of radius Y == 0.5-0.04 pm. The viscosity of water in such capillaries is elevated (by 40 % in capillaries 0.04 pm radius) but the viscosity of non-polar CC1 and benzene remains normal. The temperature dependence of the increased viscosity of water is studied ; the viscosity becomes normal at t = 60-70°C When water is drawn into a capillary with " dry " walls the wetting angle differs from zero.In these cases the contact angle is not constant but depends on the rate of entry of the water. Owing to the action of surface forces the structure of liquids in fine capillaries may differ from their structure in bulk. Investigation of these structural changes yields information not only concerning the range of action of surface forces but also about the structure of the liquids. We studied the viscosity of water and non- polar liquids in capillaries from 1 to 0.05 pm in radius drawn from quartz glass containing over 99.99 % SiOz. This material completely eliminated the effect of leaching and dissolving of the capillary walls and interpretation of the experiments was not complicated by pore geometry. The first task when using fine capillaries is to determine their radius to a sufficient accuracy.Microscopic methods are useless for radii below 1 pm owing to wave limitations. Electron microscopy may give satisfactory results,' but this method involves rigid requirements as to quality of the end-cut because the capillary hole is difficult to distinguish from surface irregularities. The radius can be determined in simple manner from the capillary pressure as the pressure of air compressed by a liquid is sucked into the capillary which is sealed at one end.2 However since the liquid advances along an unwetted surface the contact angle 8 and the radius r are unknown. More reliable results can be obtained by calibration of the capillary by thermal expansion of the liquid in a communicating ~ a v i t y . ~ However this method is laborious and the accuracy is not better than The method used here for measuring capillary radii was that of measuring the pressure of nitrogen Po needed to form and detach a bubble from the capillary end immersed in a liquid.Then the radius yo is found from the equation 10-15 %. ro = 2a/P,. (1) The major advantage of the method is that tabulated values of surface tension cr can be used in the calculations since bubble formation occurs in the bulk of the liquid outside the sphere of influence of surface forces. The detaching pressure recorded corresponds to the minimum curvature radius of the meniscus thus eliminating the difficulty-determined value cos 8 from the calculations. The capillary radius measured by this method differs from the actual r by the thickness 21 3 214 VISCOSITY OF LIQUIDS I N CAPILLARIES h of the film coating the capillary surface r = ro + 11.However a similar correction must be introduced when other methods are used (except for microscopic ones). It was designed not only for determining capillary diameters but also for measuring the viscosity and surface tension of the liquids contained in the capillaries. An empty sealed quartz capillary (1) is cemented with an epoxy resin into the stopper of a high-pressure chamber (3). After cementing the capillary ends are trimmed one of them being inserted in the ampoule (2) containing the liquid under study and the other into the high-pressure chamber. The high-pressure chamber communicates with a pneumatic system by means of which the gas pressure in the chamber can be changed rapidly within a range of up to 100 atm.Pressure is measured by a set of standard (inter- changeable for different ranges) gauges (4) of 0.3 % class accuracy. The accuracy of the pressure reading when working within the latter two-thirds of the gauge scales was not worse than 1 %. Fig. 1 is a schematic diagram of the experimental apparatus. FIG. 1 .-Diagram of unit for investigating liquids in microcapillaries. The movements of the meniscus of the liquid filling the capillary were observed with a long-focus microscope (1OOx) (5) by the dark background method. A micrometer eyepiece was used for measuring meniscus travel (to _+1 ,urn). The time of travel was measured with a stopwatch. The capillary and the ampoule containing the liquid under study were controlled thermostattically at the assigned temperature to within 50.1".The temperature of the liquid was measured with a differential thermocouple (6) the tip of which was located near the capillary. The entire unit was arranged on the movable stage of a reading microscope by means of which the position of the meniscus in the capillary could be read off the scale (7) of a micrometric screw to within & 10 pm. When the free end of the capillary comes in contact with the liquid in the ampoule the liquid is sucked into the capillary if the gas pressure in the capillary P<Pc (Pc being the capillary pressure) or is forced out if P>Pc ; when P = Pc the meniscus stops. The capillary pressure Pc was registered on approach to equilibrium from both directions (at P>Pc and P<P,) the advancing meniscus always moving along a wetting film preformed by the movement of the receding meniscus.Equality of the measured values of P showed that under these conditions there was no wetting hysteresis. The effect of the limiting shear stress of the bulk water ( ~ = 1 0 - ~ d y n e / ~ r n ~ ) ~ was much smaller than the possible error of pressure measurement and therefore undetectable. At a pressure slightly exceeding the capillary pressure the liquid is forced out of the capillary slowly so that the formation and detachment of a gas bubble from the capillary end can be observed and the minimum pressure Po needed for this measured. N. V. CHURAYEV V. D. SOBOLEV AND Z. M. ZORIN 215 With Po constant the following events were observed as the bubble size increased the pressure inside the bubble decreased the rate of its growth being determined by the resistance of the capillary to gas flow.At the moment of bubble detachment this pressure is smaller than P, and therefore immediately after detachment the liquid is drawn into the capillary. However since the pressure Po in the system is kept constant the liquid column is gradually forced out of the Capillary and the events are repeated. When working with a capillary of r<O.1 pm the bubbles are difficult to see but the periodic intake and outflow of the liquid from the capillary is readily observed and this occurs only when the pressure rises to P = Po. Hence the applicability of the method is restricted only by the possibility of observing the meniscus in the capillary. The results of measurement of P and Po in the same capillaries for different liquids are presented in table I .TABLE 1 .-MEASURED VALUES OF CAPILLARY PRESSURE Pc PRESSURE OF AIR BUBBLE DETACH- experiment no. 1 2 3 4 5 liquid water CCl water CCI water CCI benzene mercury water CCI benzene water CCI benzene MENT Po AND CAPILLARY RADIUS Yg P c PO Q atm atm t "C dyne/cm 2.82 1.06 6.05 2.23 8.21 2.99 3.24 36.4 37.5 13.14 14.06 35.4 13.14 14.04 2.82 1.06 - - 8.21 2.99 3.24 - - 13.14 14.06 - - - 19.5 19.5 21.8 21.8 18.9 19.0 19.0 19.1 20.4 20.5 20.8 22 22 22 72.7 26.8 72.6 26.7 72.9 26.8 28.95 323* 72.72 26.75 28.85 72.6 26.7 28.9 r0 Arolro ium % 0.525 0.77 0.521 0-247 0.81 0.245 0.181 0.182 0.181 0.0394 0.0416 1.7 0.0414 0.041 8 0.041 5 0.72 0.041 8 0.55 * B cos 0 Since P was measured by the receding meniscus i.e. under conditions of complete wetting 5 * the data given are evidence of equality (to the accuracy of measurement) of the surface tension of the liquids in the capillaries (of the radii indicated in table 1) and in the bulk.The capillary radii were calculated by eqn (1) from the pressures Po measured for different liquids and the bulk values of 6. Experiments with different liquids gave closely agreeing values for y o the average deviation not usually exceeding 1 %. With fine capillaries (r<O.l pm) allowance must be made for a systematic error 0 = (h/ro) 100 % where h is the thickness of the adsorption film. If taking into account undersaturation due to concavity of the meniscus the adsorption film thick- ness is taken to be 10-15&' the possible error of determination of the capillary radius even for the finest of the capillaries studied is not more than 3-4 %.If the capillary has a slight taper the " profile " of the channel r(x) can be found or constancy of Y over the length of the capillary can be verified by measuring the pressure P and the corresponding coordinate of the meniscus. Poiseille's law for a liquid column of length Z moving at a velocity ZI in a cylindrical capillary of radius r is v = r2(Pc-PI)/8qZ (2) 216 VISCOSITY OF LIQUIDS IN CAPILLARIES where q is the viscosity of the liquid P is the capillary pressure and PI is the air pressure near the meniscus. The pressure PI is not equal to the pressure recorded by the gauge because there is a pressure drop due to movement of the gas column in the capillary. To determine PI we make use of the equation of gas movement in a capillary for Knudsen numbers K = 0.001-0.1 v = r2(P-P,)(I +4t/r)/Sq*Z* (3) where P is the gas pressure in the chamber q* is the viscosity of the gas I* is the length of the gas column and 5 = 1.381 is the coefficient of slip ( A being the free path of the gas molecules).Eqn (3) is applicable under the conditions of our experi- ments since for capillaries from 10 to 0.01 pm in radius K = 0.003-0.03. Simul- taneous solution of eqn (2) and (3) gives Zlv = r2(Pc-P)/8q (4) Z1/Z = 1 +q*Z*/qZ(l+45/r). ( 5 ) where Il is the effective length of the liquid column Usually Z>Z*/3. Then the difference between Zl and I is not more than a few per cent. Since the second term of eqn (5) is small compared to 1 and as the viscosity of air depends little on the pressuie the dependence of vZ on P (at P = const.) should be practically linear.All our experimental data obtained are in good accord with this assumption.* Fig. 2 illustrate one of the (vZ, P ) dependencies for water (curve 2) and fig. 3 for CC14 (the graphs are plotted as v against P at I = const). The intersection of the graphs with the abscissa axis corresponds to the capillary pressure P, which coincides with Po since the measurements are made with a receding meniscus or with a meniscus advancing on a part of the capillary coated with a wetting film. The microscope is set up a definite distance Z from the end of the capillary. The travel of the meniscus (receding at first and then advancing along the previously traversed section) are observed in the field of view of the fixed microscope at different pressures P .The length of the section of observation was 0.4 mm for large u and 0.02 mm for the smallest velocities of movement. Then the same measurements were repeated at a different value of 1. The fact that the (v P ) graphs intersect at one point (fig. 3) is evidence that the radius was constant over the length of the capillary. Using eqn (4) the coefficient of viscosity q of the liquids in the capillaries could be determined from the angular coefficient of the (v P ) graphs. The P values were found from the points of intersection of the graphs with the abscissa axis. These values were also usel" for determining the capillary radius. The results of measurement of the viscosity of water (points l) CC14 (points 2) and benzene (points 3) at tz20"C are presented in fig. 4. In this graph the ordinates are relative viscosity q/qo (qo being the viscosity of the liquid in bulk) and the abscissae are capillary radii.It is evident from the data obtained that no changes in viscosity occur for non-polar CC14 and benzene with decrease of r down to 0.05 pm. For polar water the supermolecular structure of which varies under the action of surface forces the viscosity rises with decreasing capillary radius. The difference of the viscosity from the bulk value becomes perceptible at rg0.5 pm and reaches ca. 35-40 % when r is decreased to 0.05 pm. It is noted that the data obtained can be interpreted not only as increase of the viscosity of water in fine capillaries. If it is assumed that the properties of the liquid * Except for travel of the advancing meniscus of polar liquids along a dry surface (fig.2 curve 1). P atm FIG. 2.-Dependence of product uZl on gas pressure in chamber P for movement of a water column along dry (l) and prewetted (2) surfaces of a quartz capillary (r = 0.045 pm t = 22.2"C I = 11.26 mm). P atm FIG. 3.-Dependence of velocity of movement v of a CC1 column in a quartz capillary (r = 0.051 pm t = 19.5"C) on gas pressure in chamber P at I = 11.55 mm (l) 25.11 mm (2) and 37.52 mm (3). 21 8 \ 1 - 4 P L - - - - o - - - * - - - - L l 1 1 VISCOSITY OF LIQUIDS I N CAPILLARIES 1.5 0 F F I I 0 25 5 0 7 5 FIG. 5.-Temperature dependence of relative viscosity of water for capillaries for radius r = 0.17 pm (l) r = 0.046 prn (2) and r = 0.050 pm (3). . 1 ; 0 2 ; 0 3 * For boundary layers of water T~ = 100 dyne/cm2. Calculations show that the gradients of pressure used in our experiments were insufficient to put them in motion.N. V . CHURAYEV V . D. SOBOLEV AND Z . M. ZORIN 219 We now consider the case where the liquid meniscus advances along the surface of a dry capillary (fig. 2 curve 1). These observations were made as follows. First the velocity of entry of the liquid into the dry capillary was measured at atmospheric pressure (Po = 0) and then at gradually increasing P values (but with P - 0 ) . Under all conditions the meniscus travels only in one direction always advancing on the unwetted surface of the capillary. After the movement of the column stopped the pressure was again lowered to atmospheric and the cycle was repeated on another section of the capillary. Coniparison of curves 1 and 2 (fig.2) shows that the velocity of the meniscus along the dry surface is much lower than along the wetted surface. This can be attributed to the effect of wetting hysteresis. During travel of an advancing meniscus a finite contact angle 8 forms between the meniscus and the substrate surface.4* For 0 = const. the graph of ul against P should be linear and should intersect the abscissa axis at P = (20 cos 0)/r. However as shown in fig. 2 at small velocities of the liquid column the (ul, P) dependence deviates from a straight line. This can be associated only with the decrease in 8 when the velocity of the meniscus is lowered. Fig. 6 shows the values of cos 8 calculated from eqn (4) under the assumption that P = (2 S cos 8)/r plotted against the meniscus velocity v for water (curves 1-4) and CCI (curve 6).With CC14 as investigations have shown 8 = 0 for travel along a surface both dry or wet. For water at u>5 pm/s the contact angle differs from 0 and remains practically constant in value. At flow velocities u<5 pm/s cos O+ 1 with u+O this effect being more pronounced in fine capillaries. 0 4 0 0.8 0.7 O* 6 0.3 1 5 10 15 5 0 25 30 v pmls FIG. 6.-Cos 0 against veloctiy v of meniscus advancing along unwetted capillary surface ( t = 20°C). Water r = 0.181 pm (l) r = 0.049 pm (2) Y = 0.054 pm (3) r = 0.5 pm (4). Water along adsorbed film r = 0.22 pm (5) ; carbon tetrachloride r = 0.051 pm (6). The dependence of 8 on u cannot be attributed to any known purely hydrodynainic causes,6* l o because the latter change the shape of meniscus perceptibly only at velocities of v b 2 - 3 cm/s i.e.much greater than those used in our experiments. In 220 VISCOSITY OF LIQUIDS I N CAPILLARIES all probability this phenomenon is related to the conditions of formation of an absorbed film near the advancing meniscus and to specific meniscus-film interactions (formation of a transitional zone l1 and heat effects of film wetting). To verify this statement we measured cos 8 on suction of water into a capillary whose walls had preliminarily been coated with an adsorbed film of water. The film was formed as follows. A water column 10-15 mm in length was drawn into one end of capillary 50-70mm long after which the capillary was sealed at both ends and placed in a thermostat. The decrease in length of the column in time was observed with a comparator until equilibrium was established.If it is assumed that at equilibrium the water evaporating from the column coats the free walls of the capillary with a uniform film the thickness of the film can easily be calculated from the capillary radius.' After cementing this capillary in the unit the part filled with water was broken off and the free end was brought in contact with the ampoule (2) (fig. 1). Measurements were made by the method described above. Curve 5 of fig. 6 is a plot of cos 8 against velocity o in a capillary of radius r = 0.22 pm coated with an adsorbed film 15 A thick. It is evident from the graph that in this case the wetting angle 8 is independent of the velocity and cos 8 = 0.96. Further investiga- tion is necessary. The results of the above experiments show that reliable measurements of the viscosity of liquids in capillaries can be made only if cos 8 = const.It is evident from fig. 2 that this can be accomplished either with the meniscus moving along a wetted surface (curve 2 cos 8 = I) or along a dry one but only at high velocities (curve 1 vl 3 7 ~ 1 0 - ~ cm2/s). The slope of the linear part of the graphs is the same in both cases which gives the same values of viscosity of the liquid. However the first method is preferable since the capillary pressure P is determined from the intersection of the graph with the abscissa axis. The linear course of the (vZ, P) or (0 P) graphs at D O and v<O are convincing proof that the computational basis of the method is correct. V. D. Sobolev and Z. M. Zorin Research in Surface Forces (Moscow 1967) vol. 3 p. 36. N. N. Fedyakin Kolloid Zhur. 1962 24,497. N. N. Fedyakin Dokl. Akad. Nauk. S.S.S.R. 1961 138 1389. Zhur. Fiz. Khim. 1962 36 1450. N. F. Bondarenko Dokl. Akad. Nauk. S.S.SR. 1967 177 383. W. Rose and R. W. Heins J. Colloid. Sci. 1962 17 39. G. Friz 2. angew. Phys. 1965 19 374. B. V. Deryaguin and Z. M. Zorin Dokl. Akad. Nauk S.S.S.R. 1954,98,93; B. V. Deryaguin 2. M. Zorin Proc. 2nd Int. Congr. Surface Activity 1957 2 145. M. Devienne Frottement et kchanges Thermiques duns les Gaz Rarefies (Paris 1958). B. V. Deryaguin A. S. Titiyevskaya and B. Kh. Vybornova Kolloid Zhur. 1960 22 398. B. V. Deryaguin Zhur. Fiz. Khim. 1940 14 137. lo V. Ludviksson and E. N. Lightfoot A.I.Chem. Eng. J. 1968 14 674.

ISSN:0370-9302    

DOI:10.1039/SD9700100213

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Preliminary Studies of Thick Surface Films BY A. J. SMITH AND A. CAMERON Lubrication Laboratory Mechanical Engineering Department Imperial College of Science and Technology Exhibition Road London S.W.7 England Received 9th April 1970 A new technique is described for studying the existence of a thick viscous lubricant film adjacent to the surface of a metal. It uses mercury instead of a solid plate to displace a hydrocarbon from the metal surface thereby avoiding spurious effects due to dirt or surface asperities. The capacitance between the solid metal and the mercury is used to indicate the film thickness. A special circuit enables the potential difference applied to the surfaces to be reduced to the order of microvolts. A thick film is formed when there is a surfactant present in the hydrocarbon which reacts chemically with the metal.The soap so formed appears to enmesh the hydrocarbon near the surface forming a grease layer some 103-104 8 thick. Preliminary results show the effect of carrier and surfactant matching. The addition of polar molecules to non-polar lubricants improves their frictional qualities. Since the work of Hardy and Bowden it has been assumed that the action of the additives is due to the formation of a close-packed layer one molecule thick on the lubricated surface. There has however been a continuing series of reports that these mixtures can form a “thick” film of semi-fluid material adjacent to the metal surface. Many workers 3-10 have investigated this problem by studying the behaviour of liquids squeezed out from between plane surfaces and some have variously detected gross departures from Newtonian flow at film thicknesses ranging from 200A to 10,000 A.Studies of flow through capillaries lo* have yielded similar results with the upper limit of anomalous film formation reported as high as 3 x lo6 A. Although some of these studies have been conducted with scrupulous care and have been of a convincing nature others have been successfully criticized on the grounds that solid dust or colloidal particles or asperities on the approaching surfaces may have affected the experimental results. Notwithstanding the fact that other experiments conducted on one surface only 12-18 have tended to support the findings of these squeeze-film experiments the squeeze-film technique generally has become discredited on the grounds that it is fundamentally susceptible to spurious interference due to dirt or asperities.In this report the possibility of dirt or surface asperities keeping the squeeze surfaces apart has been avoided by displacing the fluid by mercury which might be expected to flow around any such protruberances. The film thickness was deduced from measurements of capacitance. EXPERIMENTAL Two forms of the same basic apparatus have been used these will be designated the ball apparatus and the plate apparatus. The plate apparatus (fig. 1) consists of a vertical glass plate A on whose surface are deposited by vacuum evaporation a number of rectangular areas of metal B. Holes bored through the plate and plugged with a conducting epoxy 221 222 THICK nA / SURFACE FILMS /- E U FIG. 1 .-Plate apparatus.A glass plate ; B metallized areas ; C electrical connections ; D glass chamber ; E mercury reservoir ; F glass cock. resin permit electrical connections C to be made to the metal film from the reverse side of the plate. The metallized side of the plate covers the open face of a glass chamber D connected by flexible PTFE tube to a mercury reservoir E. By opening the cock F and raising or lowering the reservoir the level of the mercury may be controlled. A wire passing into the reservoir makes electrical connection with the mercury. The apparatus was used in the following manner. The mercury in the chamber was raised to cover the metallized surface and the electrical circuit from the mercury to the metal film checked. The mercury was lowered exposing some of the plates and sufficient lubricant was added to the chamber through an opening in its top to cover them.The oil was left in contact with the metal for a measured period of time. The mercury level in the chamber was then raised to cover the metal plates and the capacitance between a plate and the mercury was measured as a function of time. In the ball apparatus (fig. 2) the metallized glass surface is functionally replaced by a steel ball A suspended by an insulated steel wire from a stopper B. A guard coil of steel -1-1 ATT EN UATOR -1 w AMPLl Fl E R FIG. 2.-Ball apparatus. A steel ball ; B stopper ; C Guard coil ; D mercury pool ; E test tube; F G electrical connections. wire C encircles the ball and makes contact with the mercury pool D contained in a test tube E. This apparatus was used in a similar manner to the plate apparatus the stopper being raised or lowered to effect the relative motion of ball and mercury.In order to ensure that the electrical measurements made at the terminal pair F G would not be affected by the pres- ence of a film of lubricant on the guard coil care was taken that the lower end of the coil A . J . SMITH AND A . CAMERON 223 was at all times in contact with the mercury both before and after the addition of the lubricant to the apparatus. By immersing the tube of the ball apparatus in a heated bath exppriments could conveniently be carried out over a range of temperatures. CLEANING.-A~~ glass parts of the apparatus were cleaned in chromic acid and rinsed in distilled water and acetone immediately before use. Metal parts including the metallized glass plate were cleaned by ultrasonic agitation in an alkaline detergent solution (ARDROX 161 8) for a period of 15 min.followed by similar periods of agitation in an emulsifier solution (Imperial Smelters ISCEON 11 3s) and in pure tetra-chloro-tetrafluoro-ethane (Imperial Smelters Fluorisol). The parts were rinsed in distilled water between cleaning stages. In all the cleaning procedures spreading of a film of distilled water on the surface was used as the criterion of cleanliness. MATERIALS The mercury used in the experiments was trebly distilled in uacuu and stored in polythene bottles prior to the tests. The hexadecane and benzene which acted as solvents (i.e. the test lubricants) were supplied by Koch-Light (Purissimus grade) and Hopkins and Williams (A.R.) respectively.The solvents were purified by passing them through glass columns 700mm long by 20 mm diam. packed with activated silica gel. The solvents were left in contact with the adsorbant for a minimum period of 24 h. The organic acids and amines used were also supplied by Koc h-Li g h t (Purissimus) . CAPACITANCE MEASUREMENTS The electrical capacitance of the mercury-oil film-metal surface system was measured in order to deduce the thickness of the oil film which acted as the dielectric of capacitor. A modified Wayne-Kerr Universal Bridge type B221 in conjunction with the Auto-balance Adaptor AA221 was used in making these measurements. This bridge operates on the transformer ratio arm principle l9 and unmodified presented to the test capacitor over the range of capacitance of interest a peak voltage of 70 mV.Under the experimental condi- tions this voltage was sufficient to cause dielectric breakdown of the oil film thereby invali- dating the measurements. The voltage applied to the oil film was therefore reduced by inserting an attenuator between the E terminals of the bridge and the capacitor and an ampli- fier between the capacitor and the I terminals of the bridge. Fig. 3 and 4 are much simplified ‘ X .L - FIG. 3.-Unmodified bridge. circuit diagrams of the unmodified and modified bridges respectively. The original bridge applied a voltage Yx to the unknown admittance through which passed a current i,. Adjust- ing the standards G and B brought it into balance with ix. By inserting the attenuator R1 and R2 the voltage across Y was reduced to V,IA. The resulting current iJA was boosted 224 THICK SURFACE FILMS by the amplifier to give the output current ix B/A.By setting B equal to A the bridge could be used in the same way as before modification. In practice A and B were each set at 1,000 thereby reducing the peak voltage applied to the oil film to 70pV. This voltage was sufficiently small to avoid breakdown of oil films above 1,OOOA thick but for measurements on thinner films an additional attenuator could be inserted in the circuit in order to reduce the voltage by a further factor of 100 (i.e. to 700 nV). The system was calibrated to give capacitance measurements of an accuracy of &2 %. f FIG. 4.-Modified bridge. RESULTS AND DISCUSSION This study of thick surface films has been in three main steps these are (1) an attempt to demonstrate the existence of a statically stable film of thickness greater than about 100 A (this approaches the practical limit of resolution of the technique) (2) to identify a group of liquid solvents which when pure would not give rise to a film in order that the effect of adding small concentrations of impurities might be studied ; (3) by observing under varying conditions the film forming properties of the solutions mentioned in (2) to attempt to elucidate the mechanism of formation of the films.UNPURIFIED HEXADECANB When hexadecane as supplied by the manufacturers was tested in the ball appara- tus the result was found to be dependent upon the time of immersion of the ball in the alkane prior to being dipped into the mercury. When immersion was for less than several minutes measurements of capacitance and conductance between the ball and mercury were indistinguishable from those obtained in the absence of lubricant indicating metallic contact between ball and mercury.Fig. 5 however is typical of the result obtained for longer periods of immersion; in this case the ball was in contact with the hexadecane for 30 min before being dipped into the mercury. The test was at room temperature (202 1OC). This graph shows rapid thinning of the film according to a constant exponent of time (t-2*5) until a thickness of 4,000 A was approached. Thereafter the plot departs increasingly from a straight line and eventually the film stabilized at a thickness of 1,800 A. The thickness curve was derived from the capacitance measurements according to the equation h = AEE,/C where A is the surface area of the ball 8 is the relative permittivity of the dielectric E is the rationalized permittivity of free space C is the capacitance of the system.225 The value used for E was that of bulk liquid hexadecane (2.29). The uncertain validity of this value in this situation is an unfortunate feature of this method of thickness measurement. The minimum possible value of E however is unity when the corres- ponding minimum film thickness would be 900 .$ a value far in excess of that expec- A . J . SMITH A N D A . CAMERON I 2 5 10 2 0 5 0 100 time (min) FIG. 5.-Thinning of unpurified cetane on steel ball. ted from the normal theory of adsorption referred to earlier. There is no correspond- ing maximum limiting value of E but there is little reason to expect that the value of E for the film would be far from that which has been assumed.PURIFIED HEXADECANE When the alkane was purified by adsorption on silica gel no film was obtained from hexadecane in either of the two forms of the apparatus. This was true even when the period of immersion of the steel ball or evaporated chromium plate was increased to 90 min. The criterion by which the absence of a film was judged was that the capacitance and conductance of the system measured in the presence of an oil were similar to those measured without an oil. A.R. benzene purified in a similar way gave a similar negative result as did untreated A.R. acetone. STEARIC ACID I N HEXADECANE SOLUTION A solution of 0.1 % w/w of stearic acid in purified hexadecane when tested in the ball apparatus yielded a film whose thinning is depicted in fig.6. The growth time of the film was 15 min at 20+ 1"C the experimental temperature. The calculated minimum value of film thickness was 2,100 A. This is based on a value of E = 2.2. (E for solid stearic acid is 2.21 E for hexadecane is 2.29.) Fig 5. SP1-H 226 n 3 W ' O E 6 THICK SURFACE FILMS time (min) Fro. 6.-Thinning of 0.1 % stearic acid in hexadecane solution on steel ball. time (min) FIG. 7.-Thinning of 0.1 % stearic acid in hexadecane solution at varying temperatures. represents the thinning of a film formed in the ball apparatus at a temperature of lOO"C the period of growth having been 15 min. The temperature was maintained at 100°C until the film had stabilized and was then raised in steps of 10°C. At a temperature of 100°C the film thinned more rapidly than a similar film formed and compressed at 20"C but stabilized at a greater thickness (19 x lo3 A).Slight further thinning occurred on heating to llO"C but over the temperature range 110-140°C A . J . SMITH AND A . CAMERON 227 the film thckness was constant. Between 150 and 170°C the film thickness again fell rapidly but stabilized when the temperature was fixed at 170"C the maximum attain- able by the oil bath used. Similar transition temperatures have been reported for greases.2 O The thinning of a stearic acid+ hexadecane film which was both grown and com- pressed on a steel ball at 170°C is depicted in fig. 8. The final thickness here was slightly greater (1.2 x lo4 A) than that of the film grown at 100°C and later heated to 170°C.Similar experiments were run with 0.1 "/o cetyl amine solution in hexadecane using the ball apparatus. Immersion time was for 1 h at temperatures of 20°C and 0 . 5 I 2 5 10 20 5 0 100 time (min) FIG. 8.-Thinning of 0.1 % stearic acid in hexadecane solution at 170°C. 95°C. In neither case was a film found. These results strongly suggest that the thick film might be caused by chemical means. A possible mechanism being the formation of soap fibrils which entrain the solvent molecules forming a two-phase soap thickened grease near the metal surface. NOBLE METALS A gold surface tested in the plate apparatus with a 0.1 % solution of stearic acid after a period of 15 min immersion failed to yield a film. The gold visibly amalga- mated during the test. It was expected that a 0.1 % solution of cetylamine in hexa- decane would protect a copper surface against amalgamation but this was not so even after 2 h of immersion in the solution prior to the test.There were apparently sufficient defects in the amine film to permit the mercury to penetrate it. The ball apparatus was modified by replacing the steel ball and wires by a rectangu- lar slip of platinum foil and platinum wires. A 0.1 % solution of stearic acid in hexadecane failec! to produce a film after immersion of the platinum in the lubricant for 30 min. A similar negative result was obtained with 0.1 % hexadecylamine solu- tion and platinum. 228 THICK SURFACE FILMS A valuable feature of these results is that they demonstrate that the mercury surface has no measurable effect upon the film under observation.This is pre- sumably due either to the absence of a thick film on the mercury surface or to the breaking of a fresh mercury surface as the solid surface moves under the mercury and to the movement of the mercury with the film as the film is squeezed during thinning ORGANIC ACIDS I N BENZENE SOLUTIONS Fig. 9 10 and 11 represent the thinning at room temperature of films formed from 0.1 % solutions of sebacic acid stearic acid and oleic acid respectively in benzene. time (min) FIG. 9.-Thinning of 0.1 % sebacic acid solution in benzene. The capacitance measurements have been interpreted as before in each case the value of 8 used has been that of the bulk solid acid. A striking feature of these results is that in all cases the capacitance values are at least three orders of magnitude greater than those obtained with stearic acid in hexadecane solution and indicate values of film thickness in the order of small numbers of molecular lengths.The sebacic acid result is interesting in this respect in that the calculated film thickness for this material (fig. 9) is of the same order as those of the other two acids despite the fact that an adsorbed monolayer of sebacic acid is approximately one tenth of the thickness of that of a monobasic fatty acid of comparable carbon chain length. Clearly the calculated values of film thickness plotted in fig. 9 10 and 1 1 must be regarded as approximate only. With films of this thickness the effect of the surface micro-relief of the steel ball would be to increase the effective surface area con- siderably over that calculated from the macroscopic topology of the ball.The uncertainty of the value of E used in the calculation of thickness discussed earlier is another possible source of error. The fact that stearic acid gives a far thinner film with benzene than it does with hexadecane is a striking demonstration of the effect noticed by previous workers s-7* of the solvent on the film forming properties of this type of solution. A . J . SMITH A N D A . CAMERON 229 time (min) FIG. 10-Thinning 0.1 % stearic acid solution in benzene. time (min) FIG. 11.-Thinning of 0.1 % oleic acid solution in benzene. CONCLUSION The most important conclusion of this preliminary study is that the thick surface film found by many previous workers does exist. The process of formation would appear to be due to chemical reaction rather than long-range surface physical forces.This is supported by their failure to form on noble metals or with hexadecyl amine on steel and by the similarity of the results obtained from stearic acid and sebacic acid solution in benzene. The thickness of the film is strongly dependent upon the nature 230 THICK SURFACE FILMS of the solvent being greater when solvent and additive are of similar molecular struc- t ~ r e . ~ - ~ * A possible mechanism of formation of the film is chemical attack of the surface by surfactants resulting in a soap film which is thickened by solvent molecules in the manner discussed in ref. (21). The authors are grateful to the Esso Petroleum Company for financial support for this work. W. Hardy Collected Works (Cambridge University Press Cambridge 1936.F. P. Bowden and D. Tabor Friction and Lubrication of Solids (Oxford University Press Oxford 1950). S. M. Bastow and F. P. Bowden Proc. Roy. SOC. A 1933 151 220. S. M. Bastow and F. P. Bowden Proc. Roy. SOC. A 1931 134,404. G. I. Fuks Research in Surface Forces I. (Consultants’ Bureau New York 1963) p. 79. G. I. Fuks Research in Surface Forces ZZ. G. I. Fuks and Bratova Doklady Acad. Nauk 1963 153 1106. S. J. Needs Trans. A.S.M.E. 1940 62 331. T. C. Askwith A. Cameron and R. F. Crouch Proc. Roy. SOC. A 1966 291 500. R. Bulkley US. Bur. Stand. J. Res. 1930 6 89. (Consultants’ Bureau New York 1966) p. 159. lo R. E. Wilson and D. P. Barnard J. h d . Eng. Chem. 1922 14,682. l2 B. V. Deryaguin Wear 1957 1,277. l3 B. V. Deryaguin N. N. Kharaeva A. M. Khomutov and S. V. Andreev Research in Sirrface l4 A. M. Taylor and A. King J. Opt. SOC. Amer. 1933 23 308. l 5 R. S. Bradley 2. Krist. 1936 96 499. l6 B. V. Deryaguin M. M. Kusakov and K. Krim Acta physiochim. 1945 20 35. l7 G. L. Clark B. H. Lincoln and R. R. Sterret Proc. Amer. Petrbl. Inst. ZZI 1935 16 68. Forces ZZ. (Consultants’ Bureau N.Y. 1966) p. 156. J. Tausz and P. Szekely Erdol Teer 1933 9 331. Wayne-Kerr Monograph no. 1. Malden Surrey). The Transformer Ratio-arm Bridge. (Wayne-Kerr Ltd. New 2o D. Evans I. F. Hutton J. B. Matthews J. Appl. Chem. 1952 2 252. ’’ W. J. S. Grew Ph.D. Thesis (University of London 1969) chap. 3 pp. 37-102.

ISSN:0370-9302    

DOI:10.1039/SD9700100221

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Thickness of Very Thin Films in Elastohydrodynamic Lubrication BY A. DYSON Shell Research Ltd. Thornton Research Centre P.O. Box 1 Chester CHI 3SH England Received 9th April 1970 Films of lubricant were formed between two loaded rolling steel discs and the thicknesses of the films were estimated from measurements of the electrical capacitance between the discs. Two mineral oils and one synthetic diester were used as lubricants. The minimum film thicknesses ranged down to approximately 10 nm (100 A) and agreed approximately with the predictions of an isothermal Newtonian theory in which the viscosity of the lubricant measured in bulk was used. Of the four sets of results examined only one that for the dieter gave measured film thicknesses greater than the theoretical ones. The discrepancy of 13 % is probably within the systematic errors of the experiment and of the theory.With this exception there was therefore no evidence that the viscosity of the lubricant was affected by the close proximity of the metal surfaces. The maximum shear rate in these experiments was of the order of lo7 s-l i.e. several orders of magnitude greater than those obtaining in other experiments reported in the literature in which the viscosity of a fluid was increased by the presence of the surfaces. There have been reports from many fields that the flow properties of a liquid may be modified by the close proximity of a solid surface. The modifications range from a small increase in viscosity to rigidity and the depth of the affected zone ranges from 1 nm to 10pm (10 to lo”). A review of work up to 1949 by Henniker gives references in support of the existence of such effects.In a later review Hayward and lsdale concluded that rheological abnormalities in pure liquids do not extend more than a few molecular diameters from a solid boundary although they admit that there is still some controversy about liquids containing surface-active materials. One remarkable result is that of Askwith Cameron and C r o u ~ h ~ who report that there is a “ plastic layer ” of a rigidity sufficient to arrest completely the descent of a flat metal plate at a depth of ca 1.9 pm (1 x Such zones of enhanced viscosity would have important effects in the lubrication of practical machinery. Ordinary mineral lubricating oils contain various polar compounds many of them surface-active in a non-polar hydrocarbon medium and thus belong to the class of liquids for which the evidence of the effect is strongest.Machine elements such as gear teeth cams and tappets and rolling contact bearings are lubricated by elastohydrodynamic lubrication and the minimum thickness of the film of lubricant in such devices is 0.1-1.0 pm i.e. within the reported range of action of the effect discussed. The rheological abnormalities could therefore have a con- siderable effect on the minimum film thickness which in turn strongly influences wear and fatigue failure in service. The work reported here was designed to investigate the practical importance of such effects. Fundamental work with real machinery is difficult but most of the essential elements are present in a disc machine which is considerably simpler both experimentally and theoretically.In such a machine the minimum thickness of the in.) even in pure cetane. 23 1 232 VERY THIN LUBRICANT FILMS lubricant film may be estimated from measurements of the electrical capacitance between the discs and may be compared with the predictions of a suitable theory based on the assumption that the viscosity is not influenced by the close proximity of the surfaces. If the viscosity were increased by the effect considered then the experimental estimates of film thickness would be greater than the theoretical ones. The fact that with one possible exception no such increase was detected implies that there were no significant increases in viscosity arising from the close proximity of the metal surfaces in the conditions of the experiments.Some estimates are given of the depth and degree of viscosity enhancement which it would have been possible to detect. EXPERIMENTAL The equipment used and the interpretation of the capacitance measurements in terms of minimum film thicknesses have been discussed previ~usly,~ and only an outline of the essen- tial details will be given here. Two discs each of case-hardened En34 steel 76.2 mm (3 in.) diam and 25.4 mm (1 in.) width were loaded together in rolling contact with a force of 2.45 kN (550 Ibf). The edges of one disc were rounded to give an effective contact width of 22.2 mrn (0.875 in.). The surfaces were ground to an eccentricity of less than 2.5 pm (O.OO0 1 in.) and to a surface roughness of 0.037 5-0.05 pm (1.5-2 microinch) c.1.a. Polishing with diamond paste im- proved the finish to approximately 0.02 pm (0.8 microinch) c.1.a.These surface-finish measurements were made by traversing a stylus instrument in an axial direction the datum line being generated by a skid. The bulk temperatures were measured by thermocouples embedded in the discs. Lubricant was supplied by ajet to the inlet to the contact and to the sides of the discs. The disc temperature was controlled by the temperature and rate of the oil supply. The capacitance between the discs was measured by an r.f. bridge at a frequency of 19 kHz. For the interpretation of the capacitance measurements in terms of film thicknesses it was assumed that the discs had the same shape as in the Hertzian case of dry contact with the addition of a constant separation h,. The dielectric constants of the lubricants were measured at various temperatures at atmospheric pressure and at a pressure of 345 MN m-2 (50,000 lbf in-2) approximately equal to the mean Hertzian pressure in the contact 357 MN m-2 (51,800 lbf in-2).The lubricants used were two mineral oils and one synthetic oil (a di-ester). Their properties are given in table 1 reproduced from an earlier paper.4 The TABLE 1 .-PROPERTIES OF LUBRICANTS code A B L description HVI MVI di(Zethylhexy1) kinematic viscosity /cS 37.8"C (100°F) 175.3 83.0 12.58 98.9"C (210°F) 15.36 8.8 3.31 absolute viscosity/P at 30°C 2.50 1.22 0.149 mineral oil mineral oil sebacate atmospheric pressure 60°C 0.505 0.263 0.062 1 100°C 0.126 0.073 0.028 2 absolute viscosity/P at 30°C 5.90 3.10 0.249 gauge pressure 34.5 MN 60°C 1.05 0.555 0.097 m-2 (5,000 Ibf i r 2 ) 100°C 0.232 0.135 0.042 di-ester was a commercial product but it was stipulated that it should contain no un- neutralized or partly neutralized sebacic acid.No special precautions were taken to avoid contamination. When the lubricant was changed the discs were cleaned by being rotated under no load while tissues soaked with a light paraffinic solvent were pressed against them. A . DYSON 233 RESULTS The results are given in fig. 1 and 2 in the form of graphs of the experimental film thickness he against the theoretical film thickness h, both quantities being expressed in nm. The theoretical estimates are based on the expression given by Dowson Higginson and Whitaker h = 1.6 (~q0)0.7a0.6(~~)0.03~0.43w-0.13 9 where U is the peripheral velocity of either disc; qo is the absolute viscosity of the 0 2 0 4 0 6 0 8 0 I1 theoretical film thickness ht/nm FIG.1 .-Comparison of experimental and theoretical film thickness for lubricants A and L 1 lubricant A ; 2 lubricant L. 2oi77-rtr V I I I t ) 0 20 4 0 6 0 8 0 I-u theoretical film thickness ht/nm FIG. 2.-Comparison of experimental and theoretical film thickness for lubricant B 1 constant temperature ; 2 constant speed. lubricant at atmospheric pressure at the mean temperature indicated by the embedded thermocouples; a is the pressure coefficient of viscosity; E' is the reduced elastic modulus given by E/( 1 - v2) where E is Young's modulus and v is Poisson's ratio for the material of the discs ; R is the radius of relative curvature and w is the load per unit width of contact. The theory is an isothermal Newtonian one and the viscosity used is that of the fluid in bulk.There is some evidence4 that thermal effects probably caused by viscous shear in the inlet zone are important for minimum film thicknesses of 1 pm or greater while some order-of-magnitude calculations suggested that such thermal effects would be negligible for film thicknesses of 0.1 pm or less. The results used in the analyses were therefore restricted to a minimum film thickness of 0.1 pm or less. They were analyzed statistically as a regression of he on h, all the error being assumed to arise in he. There was no evidence of non-linearity or that the regressions did not pass through the origin. The slopes of the lines were taken as the coefficients of the regression constrained to pass through the origin.The theory gives the minimum film thickness under the outlet bump (fig. 3) while the experiment gives the thickness of an assumed parallel film under the Hertzian contact zone which would give the observed electrical capacitance. The difference between these two estimates of film thickness is sketched in fig. 3 and in the example given by Dowson et aL5 the theoretical minimum film thickness h is approximately The numerical results are given in appendix 1. 234 VERY THIN LUBRICANT FILMS 85 % of the mean thickness in the Hertzian zone. It has been assumed that the same ratio applies to the present results. A further difficulty arises from the fact that the theory assumes a relation between viscosity and pressure of the form q = qo exp (ap). This is only an approximation to the properties of real fluids.The slope of the curve of the logarithm of viscosity against pressure usually decreases with increasing pressure. For lubricants A and B this curve was known only up to pressures of 103.5 MN m-2 (1 5,000 lbf i r 2 ) and for lubricant L up to approximately 0.828 GN (120,000 lbf in-”. In the theoretical estimates the value of a was taken to be the mean slope of the above curve between atmospheric pressure and a gauge pressure of 34.5 MN m-2 (5,000 lbf in-2). This 21 ROLLING VELOCITY .- OUTLET BUMP -CENTRE OF CONTACT FIG. 3.-Film shape near conjunction film shape according to exact theory; - - - - film shape assumed for purposes of estimation of film thickness from measure- ments of capacitance. (Sketch only not to scale.) (ii) h <2a FIG. 4.-Assumptions about enhancement of viscosity near surfaces.Film of lubricant withviscosity,q=-qo exp (up) equal to viscosity of the fluid in bulk m F.’ ilm of lubricant with enhanced viscosity kq = kqo exp (cip) Material of disc procedure over-estimates the effective value of a by approximately 8 % for lubricant A by 3 % for B and by 16 % for L. The estimate of the error is much more reliable for lubricant L than for lubricants A and B. The results of the statistical analyses are given in table 2. The crude regression coefficients have been corrected for the expected 15 % discrepancy between the theor- etical and experimental film thicknesses and for the error in the estimates of a. The reference to 95 % confidence limits refers to errors of a random nature only and does not take into account any possible systematic errors.DISCUSSION The observed results are now compared with those to be expected if there were any significant degree of enhancement of viscosity. Suppose that the viscosity is increased by a factor k compared with the bulk viscosity over a film of thickness a adjacent to each surface as illustrated in fig. 4. The film thickness h at the inlet edge of the Hertzian contact zone has been estimated by the method of Grubin,6 in which bound- ary conditions on the distribution of hydrodynamic pressure p = p(x) are imposed p-+co dp/dx = 0 at h = h,. A . DYSON 235 The shape of the gap between the two surfaces is assumed to be the Hertzian deformed shape with the addition of a constant separation h,. This procedure gives a good approximation to the results both of more refined theoretical solutions of the elasto- hydrodynamic problem and of experiment.An approximation due to Crook for the Hertzian deformed shape near the edge of the contact zone has been used. The resulting value of h, with viscosity enhancement assumed has been compared with the corresponding value h,(O) with no viscosity enhancement i.e. with k = 1 a = 0. The detailed analysis is given in appendix 2 and the results are shown in fig. 5 in the form of curves of h,/2a against hm(0)/2a. There are two linear asymptotes one through the origin with a slope of ks for /z,(0)/2a < (1 - k-l) (ka - l)-l and the other with unit slope and an intercept of (1 -k-l) on the positive axis of h,/2a for h,(O)/ 2a 9 (1 - k-l) (kP - l)-l. These asymptotes are independent of the assumptions made in the analysis except that the index of k may vary between 2/3 and 3/4.The results at intermediate values are so dependent. TABLE 2.-RESULTS OF STATISTICAL ANALYSES OF COMPARISON OF EXPERIMENTAL AND THEOR- ETICAL FILM THICKNESSES lubricant A disc temperature /"C 75-1 04 rotational speed/rev min- l 11.5-182 number of points 10" about regression line 2.02 been detected at 5 % significance level estimate of residual variance/nm2 intercept/nm which could have 3.2 crude regression coefficient 1.111 corrected regression coefficient 0.99 95 % confidence limits of regression coefficient ik0.022 B 135 20- 108 4 0.17 1.4 1.222 1.06 10.035 B 84-135 103-108 10 6.15 5.9 1.035 0.90 f0.037 L 30-95 106- 107 25 1.19 1.1 1.219 1.13 fO.011 * Two points with film thicknesses slightly greater than 100 nm and not shown in fig.1 have been included in the statistical analysis. In fig. 1 and 2 he is regarded as an experimental estimate of h, and h as a theo- retical estimate of h,(O) subject to certain corrections discussed in the results section. The quantity a is unknown but since the relations between he and h are all linear over nearly a decade of variation then the results must lie in one of the two asymptotic regions. Since there is no evidence of an intercept greater than about 1 nm in the linear relation between he and ht then if there is an effect the results must lie on the asymptote passing through the origin. Of the four sets of results examined only one that for di(2-ethylhexyl) sebacate (lubricant L) shows evidence of a corrected slope significantly greater than unity while the mean of the four slopes is 1.02.If the slope of 1.13 for lubricant L is accepted the corresponding value of k would be approximately 1.18 while the depth a of the zone of enhanced viscosity would be large compared with 100nm. This does not seem to be a likely situation particularly as the mineral oils A and B corres- pond more closely than does the di-ester L to the class of lubricants for which the effect would be expected to be greatest i.e. a solution of polar materials in a non-polar solvent. With the possible exception of lubricant L then there is no evidence of any enhance ment of the viscosity near a surface. The cause of the differences in the corrected 236 VERY THIN LUBRICANT FILMS slopes given in table 2 and in particular the difference between the two sets of results for lubricant B is not known.Random errors in the slopes are comparatively small and the errors are mainly systematic. One possible source of error is a deviation of the real shape of the gap between the two discs from that assumed in the interpretation of the capacitance measurements in terms of film thicknesses but it is difficult to give a quantitative estimate. The difference between the points obtained for lubricant B at constant speed and at constant temperature may reflect an inadequacy in the theory. A similar discrepancy has been noted by Greenwo~d.~ It seems probable that the 13 % discrepancy in the slope for lubricant L is a result of systematic errors in the experiment and in the theory. 2-01 I / 1 1.5 2.0 M0)/2a FIG. 5.-Theoretical effect of viscosity enhancement on film thickness.If there were a zone of liquid of enhanced viscosity adhering to either surface and if the lubricant in this zone had a dielectric constant higher than that of the bulk liquid then the effect would escape detection. But it is unlikely that the compensa- tion would be exact over any wide range of film thicknesses since the calculation of the dx s h-2a[l-(k')-'] 9 where k' is the ratio of the dielectric capacitance depends on constant of the layer near the surface to that of the bulk fluid. This integral is different in form from that in eqn (2,10) appendix 2 which governs the hydrodynamic behaviour. Another possible source of error is that smooth surfaces are assumed both in the hydrodynamic theory and in the interpretation of the capacitance measurements in A .DYSON 237 terms of film thicknesses. In practice the surfaces are rough with a scale of rough- ness comparable with the minimum film thickness. But some evidence of non- linearity in the relations between experimental and theoretical film thicknesses would be expected if this roughness were an important source of error. In these experiments the lubricant is subject to a shear rate lo in the inlet zone which varies from zero up to a maximum of 3E/2hm i.e. of the order of lo7 s-I. The shear stress is more difficult to estimate but the maximum must be much larger than 100 kN m-2. These figures for the shear stress and the shear rate are several orders of magnitude higher than in those experiments reported in the literature in which posi- tive evidence of a much greater degree of enhancement of viscosity was reported.On the other hand Cameron and Gohar found evidence of a layer 50-100 nm thick with a viscosity 3 to 4 times the bulk value. Their experiments were similar to those reported here but they used a steel ball sliding against a glass plate and measured the film thickness by an optical method. If such a layer had been present in the work reported here it would almost certainly have been detected. The author thanks Mr. A. R. Wilson and Mr. W. J. Cairney for making the experi- mental measurements of film thickness and Mrs. A. C. Rowlands and Mr. A. Prothero for the computation of the numerical integrals in appendix 2. J. C. Henniker Rev. Mod. Phys. 1949 21 322. T. C. Askwith A. Cameron and R. F. Crouch Proc. Roy. SOC. A 1966,291,500.A. Dyson H. Naylor and A. R. Wilson Proc. Inst. Mech. Eng. 1965-6 180 (3B) 119. D. Dowson G. R. Higginson and A. V. Whitaker J. Mech. Eng. Sci. 1962 4 121. A. N. Grubin Fundamentals of the Hydrodynamic Theory of Heavily Loaded Cylindrical Surfaces in Symposium. Investigation into the Contact of Machine Components (Central Scientific Institute for Technology and Mechanical Engineering Moscow 1969 book no. 30) (D.S.I.R. Trans. no. 377). ’ D. Dowson and G. R. Higginson Elastohydrodynamic Lubrication (Pergamon Press Oxford London Edinburgh New York Toronto Paris Braunschweig 1st ed. 1966) chap. 6. A. W. Crook Phil. Trans. A 1961 254 223. J. A. Greenwood A Re-examination of Elastohydrodynamic Film Thickness Results (University of Salford Department of Mechanical Engineering Report November 1969).A. Dyson and A. R. Wilson Proc. Inst. Mech. Eng. 1965-6 180(3K) 97. A. Cameron and R. Gohar Proc. Roy. SOC. A 1966 291 520. ’ A. T. J. Hayward and J. D. Isdale Brit. J. Appl. Phys. (J. Phys. D) 1969 2 251. 238 VERY THIN LUBRICANT FILMS APPENDIX 1 .-EXPERIMENTAL RESULTS lubricant speed/rev teyp./ he/nm ht/nm lubricant speed/rev te2ip.l min-1 C min-1 C 41 76 82.8 74.9 30 75 66.6 61.9 20 75 51.4 47.0 15 74 42.8 39.0 11.5 75 34.6 32.4 A 11.5 85 25.2 24.2 50 71 112.4 100.6 45 73 98.0 86.0 155 104 106.7 95.2 182 104 118.0 106.7 108 135 37.1 30.6 40 135 19.4 15.4 24 135 13.1 10.9 20 135 11.8 9.5 103 103 104 106 106 106 106 106 107 108 84 90 100 107 114 118 122 128 129 135 81.5 77.4 62.8 66.3 57.1 54.0 50.0 47.8 45.7 41.0 40.0 39.0 38.1 37.1 35.6 34.2 34.2 33.3 34.5 30.8 107 95 91 87 83 82 81 79 77.5 75 L 73 70 64 58 56 54 48 44 41 36 30 he/nm 28.5 29.5 31.4 33.7 34.2 34.2 35.0 36.2 38.1 39.2 41.9 44.9 49.5 53.3 57.1 62.8 68.5 73.9 82.9 101 .o h,/nni 23.6 24.7 26.1 27.6 28.2 28.6 29.5 30.0 31.4 32.6 34.5 37.9 42.5 44.1 46.1 51.6 57.0 60.6 68.1 81 .o APPENDIX 2 EFFECT ON FILM THICKNESS I N PURE ROLLING OF ENHANCEMENT OF LUBRICANT VISCOSITY NEAR SURFACES MODIFICATION TO REYNOLDS’ EQUATION Let the local thickness of the whole film of lubricant be h and the bulk viscosity of the lubricant be y.Let the viscosity be increased to (kq) within two regions each of thickness a and each bounded on one side by a solid surface as in fig. 6. It is first assumed that h> 2a. Under the usual assumptions of the theory of hydrodynamic lubrication by thin films the hydrostatic pressurep in the film is a function only of the coordinate x in the direction of motion while the shear stress z is a function only of the coordinate y measured in a direction normal to that of the surfaces.These quantities are related by the momentum equation For a Newtonian lubricant 7 = v(du/dy) (2.2) where y is the viscosity and u is the velocity of a particle of fluid in the x direction. Substitution of eqn (2.2) in eqn (2.1) and integration twice with respect to y gives the velocity distribution u = (1 /2y)(dp/dx)y2 + Ay + B where A and B are constants of integration and may be expected to differ in the two regions O< I y I <(h/2-a); (hl2-4< I y I <(h/2). A . DYSON 239 By symmetry A = 0 and the velocity distributions in the two regions become O< I y I <(h/2-a) 11 = (2y)-'(dp/dx)y2+Bl (h/2 - a) < I y I d (h/2) u = (2kq)-'(dp/dx)y2 + B2.At y = i h / 2 the velocity U must equal the peripheral velocity ii of the surfaces bounding the film and this gives a value for BZ B2 = ii - (kq)-'(dp/dx)(h2/8). FIG. 6.-Co-ordinate system used in analysis. The velocity of the fluid at y = f(h/2-a) is given by ti1 = ( 2 k q ) - l ( d p / d ~ ) ( h / 2 - a ) ~ +B2 = ii-(2ky)-'(dp/dx)a(h-a). This must be equal to the velocity at the same point derived from the velocity distribution in the other region of the film 0 <I y I <(h/2-a). Therefore (2q)-'(dp/dx)(h/2 -a)2 +B1 = ii -(2kq)-'(dp/dx)a(h-a) or B1 = ii - (2kq)-'(dp/dx)a(h -a) - (2~)-l(dp/dx)(h/2 -a)2. P hl2 The flow rate in the x direction per unit transverse width is Q = 2J ' udy 0 h / 2 - a d x hJ2 - a d x = Ulz - ( 1 2 ~ ) - '(dpldx) { h3 - (1 - k ' ') [ h3 - ( h - 2 ~ ) ~ ] 1.(2.3) If the lubricant is incompressible and if there is no side leakage this flow rate must remain constant as h varies. In particular it must equal the value at that position at which dp/dx = 0. If the film thickness at this point is hm the flow rate is Uh, and if this is equated to the flow rate at the general position given by eqn (2.3) an expression for the pressure gradient is obtained 12qii(h - h,) dP & = h3 - (1 - k-I)[h3 - ( h - 2a)3]' ( 2 . 4 ~ ) 240 VERY THIN LUBRICANT FILMS This is valid for h >2a. and the conventional version of Reynolds’ equation gives If h <2a then the viscosity is kq over the entire thickness of the film (2.4b) For k = 1 or a = 0 eqn (2.4a) and (2.4b) reduce to the conventional integrated form of dpldx = 12kqu(h - h,)/h3 Reynolds’ equation dp/dx = 12qii(h - h,)/h3.APPLICATION OF THE MODIFIED REYNOLDS’ EQUATION TO THE ELASTOHYDRO- Consider two stationary elastic cylinders of radii R1 and R2 with parallel axes loaded together by a load w per unit length. The line of the unloaded contact is extended to a band of width 26 where DYNAMIC LUBRICATION OF ROLLING STEEL DISCS b = 2[2wR/nE’]* where E’ is the reduced modulus such that 2/E‘ = (1 - v f ) / E + (1 - V ; ) / E ~ and v l v2 and El E2 are the Poisson’s ratio and the Young’s modulus of the materials of the discs 1 and 2 ; R is the radius of relative curvature RIR2(R1+R2)-l. The shape of the gap between the cylinders outside this band of contact is given by hl = (b2/2R)(l (x/b)(x2/b2 - l)* I -In [ I x / b I + I x2/b2 - 1 I +I) where x is measuted from the line of centres.Over a small interval near the edge of the flat region of contact this expression may be expanded into the form where h2 = 24b2(3R)-1~4 + higher powers of E I x I = b(l+&) 0<&<1. In the elastohydrodynamic lubrication of rolling steel discs it is assumed that the shape of the gap between the discs on the inlet side is given by the Hertzian gap with the addition of a constant separation hm h = h2+h,. (2.6) ‘I = ‘70 exp ( 4 7 (2.7) The viscosity is assumed to vary exponentially with pressure ; and the boundary conditions for the pressure distribution are assumed to be p = O a t x = -03 p-+co dpldx = 0 at x = -b. (2.8) The above procedure gives a good approximation to the true film thicknesses both those obtained by a more refined theory including the simultaneous solution of the governing elastic and hydrodynamic equations and also to those determined experimentally.INTEGRATION OF REYNOLDS’ EQUATION Suppose first that hm>2a. Eqn (2.4a) and (2.7) give dP 12q&(h - h,) exp (- ap)- = dx h3 - (1 - k- ‘)[h3 - ( h - 2 ~ ) ~ ] ’ A . DYSON Integration of eqn (2.9) with the boundary conditions (2.8) gives 24 1 ( h - h,)dx - mh3 -(1- k-’)[h3 -(h - 2 ~ ) ~ ] a-l = 12qou In eqn (2.10) h is given by eqn (2.5) and (2.6). A convenient substitution is b2(2&)3(3Rhm)-l = tan2 8. h = h sec2 8. Eqn (2.6) and (2.5) then give Eqn (2.10) becomes ( 12qOiia)- = (2/3)”3(2Rh,/b’)2/3(b/h~)I ,(k,O) where n / 2 (sin $ ) 7 / 3 ( ~ ~ ~ 8)5/3d8 l-(l-k-’)[l-(l-~cos2 8)3]’ (2.11) (2.12) and p = 2a/h,.(2.13) Thus for constant p h is proportional to [ll(k,O)]~. If there is no increase in viscosity near the surface then k = 1 and Il(k,O) in eqn (2.11) becomes 12(7r/2) where 12(e) = (sin e ) 7 / 3 ( ~ ~ ~ 8)5/3de. s By a standard result in the theory of gamma functions By the use of other standard results qz+ 1) = zr(z) T(z)I‘(l-z) = z/sin (zz) Z2(n/2) may be evaluated as 85c/943. The ratio of the film thickness hm that takes into account the enhancement of the viscosity to the value hm(0) in the absence of such enhancement is therefore r = h,/h,(O) = [(9 ,/3/8.)11(k,0)]*. (2.14) Eqn (2.14) is valid only if h >2a. If h <% eqn (2.4~~) is valid only from h = m to h = 2a or by eqn (2.11) and (2.13) cos 6 = p-+ For films thinner than this eqn (2.46) must be used and l,(k,O) in eqn (2.14) must be replaced by W4) fZl(k,81) where 8 = cos-’ (p-3).Eqn (2.14) therefore becomes (2.15) 242 VERY THIN LUBRICANT FILMS ASYMPTOTIC FORMS The most convenient way to present the results is in the form of a plot of hm/2a = p-l against hm(0)/2a = (r,6)-'. The asymptotes ,630 p-+ 00 are of interest. If fl< 1 eqn (2.12) may be expanded as a power series in ,6 I,(k,O) = J (sin e)'i3(c0s e)5/3de+3(1 -k-')p/ (sin e ) 7 / 3 ( ~ ~ ~ 0)11/3d0+ . = ~ [ W 3 ) W 3 ) l l - w ] + 3(1 - k- '>B(3)[r(5/3)r(7/3)1r(4)1+ * n / 2 n / 2 0 = 8~/(9J3)[1+(4/3)(1- k-')P+ '1. Eqn (2.14) then gives Y = l+(l-k-1),6+*-. Neglect of terms of the second and higher orders in ,6 and division by r j 2 then gives a quad- ratic equation for p-' in terms of (r,6)-l (p-f)2 - p-yrp)-' - (1 - k-l)(rP)-l = 0. The solution is ,6-' = +((r,6)-' _+ [(rP)-' + 4( 1 - k-l)(r,6>-l]) *. If k tends to unity fl-' must tend to ($)-l so the positive sign of the square root must be taken. If (@)-l>l the solution becomes /3-' = (r,6)-'+ (1 - k-') or hm/2a = h,(0)/2a + (1 - k-I). If ,6>1 then Or in eqn (2.15) tends to n/2 L(k,O1) tends to zero and 12(01) tends to 8n/92/3. Thus or Y = h,/hm(0)+k* 12 /2a -+ k*[ hm(0)/2a]. These two asymptotes may be derived from first principles and are independent of any assump- tions about the nature of the lubrication or about the shape of the gap. Results at inter- mediate values of hm(0)/2a depend on such assumptions. The two asymptotes intersect at hm(0)/2a = (1 - k-l)(k% - 1)-l; hm/2a = ka( 1 - k-')(ka - 1)-l. Results for intermediate values of p were obtained by numerical integration of eqn (2.14) or (2.1 5).

ISSN:0370-9302    

DOI:10.1039/SD9700100231

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Mechanical Properties of Very Thin Surface Films BY A. D. ROBERTS AND D. TABOR Surface Physics Cavendish Laboratory Cambridge England. Receiced 2nd April 1970 This paper comprises two parts the first briefly describes some earlier work on the determination of the shear strength of a calcium stearate bimolecular layer deposited between two mica surfaces. The second part deals with some more recent studies on the mechanical properties of thin liquid films sandwiched between a rubber and glass surface. The rubber surface was soft and optically smooth so that certain liquids squeezed between it and polished glass formed a very thin film of uniform thickness. Squeeze film studies made with this system revealed the importance of repulsive forces exerted by electrically-charged double layers residing on the solid surfaces.These forces were able to support the externally applied pressure so that an equilibrium film some 200 A thick was obtained. When rubber and glass surfaces were sheared in the presence of an electrolyte solution double- layer repulsion helped to support the normal load and so protect the surfaces from serious abrasion. The presence of a soap in the solution led to further protection by an oriented monolayer on the rubber surface. Measurements suggested that these two factors operated to provide effective lubrication- the electrical repulsive forces to keep surfaces apart and the SDS monolayer for increased protection at points of intimate contact where the separating fluid had been locally penetrated. The mechanical properties of very thin surface films sandwiched between bearing surfaces are of great interest in lubrication practice.In order to study these proper- ties methods are required by which the true area of contact between the surfaces and the thickness of the films between them may be determined precisely. In addition the effect of surface asperities projecting through films must be eliminated. It is this last requirement that presents a great problem because however carefully surfaces are prepared asperities remain which are vast when considered on a molecular scale. In this paper two ways are described which overcome the problem. One of these is to sandwich thin films between molecularly smooth surfaces the other between highly deformable surfaces. Both allow films of unbroken uniform thickness to be maintained between them over relatively large areas.The surfaces are mica and rubber. Some time ago it was found that mica could be cleaved to yield areas up to 80 cm2 which were step free on both sides of the sheet and molecu- larly smooth. More recently optically smooth rubber surfaces have been prepared and when these are pressed against glass they conform well to the micro-irregularities upon it. Certain liquids sandwiched between the two form in effect very thin films of uniform thickness. Optical interference techniques have been applied to both types of surface to determine precisely the film contact area and thickness. An outline is given first of work carried out 15 years ago on the shearing of a calcium stearate bimolecular layer between mica surfaces followed by a description of recent work on the squeezing of very thin liquid films between rubber and glass.SHEAR STRENGTH OF A CALCIUM STEARATE BIMOLECULAR LAYER The surfaces produced by the cleavage of mica are so smoc th that the true area of Clean mica adheres contact between them is identical with the geometric area. 243 244 MECHANICAL PROPERTIES OF FILMS readily to itself and to many other surfaces its sliding friction is large and surface damage severe. These phenomena reflect the strong action of short-range surface forces. An investigation into the nature of the contact between molecularly smooth mica surfaces was carried out by Bailey and Courtney-Pratt and the determination of the shear strength of a calcium stearate bimolecular layer formed an integral part of their investigation . Two cleaved smooth surfaces of mica were bent into cylindrical sheets supported with their axes at right angles to one another (fig.la). The rear surface of each sheet was silvered. The cylinders could then be brought together and the contact region between them studied by multiple-beam interferometry. By viewing the contact in monochromatic light the fringe pattern obtained was a system of concentric rings surrounding an area of uniform tint. The boundary of the contact area was deter- mined as the position within the first ring where the first visible change in light intensity occurred (fig. lb). As a further aid to contact assessment the specimens were illuminated with white light and contact area fringes of equal chromatic order (FECO) examined through a spectroscope (fig. 1 c). With this system separations between mica surfaces along any selected line across the area of contact may be deter- mined to an accuracy of *4 A.In addition the length of the straight portion of the fringes may be measured and for a particular cross-section of the contact the boundary of the region of intimate contact thus known with certainty. A monolayer of calcium stearate was applied to each mica surface by the Lang- muir-Blodgett technique. After deposition the mica cylinders were brought together under a normal load and the contact region found to consist of a circular sandwich containing a bimolecular layer of stearate. Interference measurements indicated a total bi-film thickness of about 45 A. A tangential force was then applied by a spring device to one of the mica shells. At any fixed value of normal load when the tangent- ial force was increased a maximum value was reached at which the surfaces began to slide smoothly one over the other.At the point of slip if the area of contact cor- responding to this value be measured accurately from interference data then the ratio of the tangential force to contact area gives a value of shear strength. Measurements made yielded an average value for the shear strength of the bi-film of 250+ 10 g/mm2 compared with a value of 10,000 g/mm2 obtained for clean mica surfaces. A few experiments with three monolayers present on each mica surface indicated even lower values than 250 g/mm2. In these experiments the mean pressure over the bi-film during shearing was only about 10 atm which is some 500 times less than pressures encountered at the points of asperity contact between sliding metal surfaces.There is some evidence which suggests that the shear strength of mono- layers is roughly proportional to pressure but direct experimental confirmation is lacking. We now turn to the squeezing of very thin liquid films between rubber and glass. In most of these experiments the mean pressures developed over films during squeezing were only of the order of 1 atm. THIN LIQUID FILMS SQUEEZED BETWEEN RUBBER A N D GLASS Recently we have been able to prepare optically smooth rubber surfaces by hot curing rubber compounds against optically smooth formers of metal and glass.2 When spherical rubber surfaces are prepared they show excellent Newt on’s rings fringe-patterns against a glass plate. Scanning electron microscopy reveals that the ultimate smoothness of the rubber depends very much upon the surface finish of the metal or glass former.Under optimum conditions the rubber will form an exact replica of them. FIG. 1 .-(a) Crossed cylinders of mica in contact (top). ( b ) Interferogram of contact area between mica cylinders. The central light region is of uniform density and over all this region the specimens are in molecular contact. The area of contact extends as far as the first visible change in intensity ; magnification 40 x (middle). (c) The appearance of fringes of equal chromatic order along a selected line across the area of contact. The contact shown is between clean sheets of mica ; magnification 17 x (bottom). [To face page 244 \ \ \ \ &.- 0.Q 0 6 0 4 0 2 / 1=3100s 3 I = 2 9 3 3 O O U ( m ~ ) 0.2 0 4 0.6 0.8 FIG.2.-Interferograms produced between a spherical rubber surface and glass plate during the normal approach under a constant load ( 5 g) of one surface toward the other through a viscous oil. The oil was a dimethyl silicone (type MS 200 Midland Silicones Ltd.) of viscosity lo6 cSt. The elastic modulus of the rubber surface was 6 x lo6 dyn/cni-2 and it had a radius of curvature of 2 cm. The pictures together with deduced profiles show the formation and collapse with time t of a " bell " entrapment of liquid. A ;= 5461 A. FIG. 3.-Interferograms of the final stages of collapse of a liquid “ bell ” entrapment. The rubber surface appears to be attracted towards the glass at its nearest point of approach (initially at about “ 5 o’clock ” on the contact periphery) and adjacent parts wrinkle under local stress.With time (indicated in days by each frame) the seal spreads over the whole contact zone so trapping small “ islands ” of liquid. The contact load oil specification and rubber surface constants are the same as those in fig. 2. Magnification 1 4 ~ . 0-4 0.2 0 0.2 0.4 FIG. 4.-Interferograms and deduced profiles showing the marked difference in contact between distilled water (left) and water containing 0.3 % SDS (right). With distilled water trapped pockets of water are seen but for the SDS solution a thin equilibrium film of about 200 A uniform thickness is obtained. A = 5 461 A. A . D. ROBERTS AND D . TABOR 245 Although these surfaces are only as smooth as their optically polished formers which are covered with asperities hundreds of Angstroms high they can be brought into molecular contact with glass or any other surface provided it is fairly smooth.This is for two reasons. First the rubber is very soft so that asperities upon it may be easily deformed. Secondly the size of asperities is relatively small so their dis- placement has only to be slight to conform with opposing surface asperities. Strong adhesion to itself and other materials a high dry sliding friction and severe surface damage accompanying this all of which recalls the behaviour of mica reflect the extensive action of surface forces between optical rubber and other surfaces placed in contact. The optical interference between a spherical rubber surface and glass plate has been examined during the normal approach under a constant load of one surface toward the other through a viscous oil.This approach traps a " bell " of liquid in the centre of the contact zone. Interferograms of the entrapment using a very viscous dimethyl silicone oil are shown in fig. 2. They were produced without silvering the rubber or glass surfaces good fringe contrast following because both surfaces had the same refractive index (-1.51) and good fringe visibility because scattered background light had been eliminated. With time most of the liquid becomes extruded from the contact area and the " bell " begins to collapse. The final stages of collapse are shown in fig. 3. It appears that the rubber surface is attracted towards the glass at its nearest point of approach which is the peripheral " lip " and seals.So strong is this sealing that adjacent parts of the rubber surface wrinkle under local stress. Eventually the seal spreads over the whole contact zone cutting off and trapping small " islands " of liquid. These islands would seem to be permanently trapped for they do not shrink in size after many days. However if the experiment be repeated with less viscous silicones similarly trapped islands do disappear. This suggests that the rubber and glass are not in intimate contact but separated by a monolayer or two of silicone. With distilled water the bell formation and collapse is rapid and difficult to observe. Once again the rubber seals to the glass but islands of trapped liquid escape after a few hours. Presumably a few monolayers of water remain between the rubber and glass.But if the experiment be repeated with water containing only 0.3 % of sodium dodecyl sulphate (SDS) a drastically different result ensues. The surfaces no longer appear to seal and trap islands of liquid. The intense black hue of sealing is not seen but instead a grey-black tint extending over the entire contact area (fig. 4). The grey- black tint indicates the presence of a very thin liquid film. This film is far too thin to show interference fringes but its thickness can be measured photometrically. If dry intimate contact gives " perfectly " black interference of zero light intensity-there is a phase change of 180° at the air-rubber interface when the contact is viewed through the glass*-and a gap of optical thickness ;1/4 gives bright interference of intensity I, any gap of intermediate optical thickness nt will give an intensity I = I sin2 (27mt/;l) at normal incidence where n is the refractive index of the material in the gap and t its geometric thickness.A mean value of n = 1.33 is assumed for thin aqueous films and no corrections made for slight phase changes that might be incurred at intermed- iate structural interfaces in the liquid gap. Photometric measurements indicate film * It is not easy to obtain precise determinations of the phase change. Some ellipsometric measure- ments carried out by Dr. R. J. King at N. P. L. suggests that the phase change at the rubber surface may be less than 180 by 1 or 2". This would lead to an error of 12-24 A in the thickness. However our own measurements show that for dry contact the light reflected from the interface has within the sensitivity of our apparatus zero intensity.This implies that the effective phase change between the rubber and the glass is in fact 180". 246 MECHANICAL PROPERTIES OF FILMS thicknesses of about 200A for these SDS films which do not collapse with time so long as the liquid does not evaporate. Aqueous solutions of SDS are surface-active so it seems reasonable to assume that the SDS molecules become adsorbed on to the rubber surface as an orientated close-packed monolayer with their negative polar end-groups in the water. It is possible that some SDS molecules may become similarly oriented on the glass surface but in any case the glass itself exposes a negative charge of OH- ions adsorbed on the surface from the aqueous solution. The negatively-charged groups may be expected to attract positive ions in the water so establishing a double-layer of charge at either surface.The concentration of positive ions in solution would be greatest at a surface and decrease with distance away from it. The double-layers of charge at either surface would mutually repel and tend to keep the surfaces apart. Thus if the charge density at either surface remains constant then an increase in pressure on the equilibrium film should thin it. Also if the number of ions in solution be increased the equilibrium film should thin again because the distribution of positive ion charges in solution becomes more constricted toward the surfaces owing to a type of screening effect and results in the electrical repulsion forces operating over a smaller distance. These two parameters pressure and ionic strength were chosen as variables for studying the properties of the electrical double-layers residing on the rubber and glass surfaces.300- - 250- 5 v) B .- 2oo- 5 150- E - rc IOO- 5 0 - 0 2 0.5 I 3 5 10 2 0 I i i b contact pressure (dyn cm-2 x lo5) FIG. 5.-Variation in the equilibrium thickness of SDS films sandwiched between a rubber sphere and glass plate with contact pressure and with solution ionic strength. Photometric measurements were made at the centre of the contact zone and the pressures quoted are the Hertzian pressures at this point they are 1.5 times the mean contact pressure as found from the load and area of contact. The results of measurements made upon thin sandwiched films of SDS are shown in fig. 5. Equilibrium films attained decreased in thickness both for an increase in pressure and ionic strength.have measured the thickness of " first black " soap films under compressive stresses up to about lo6 dyn cm-2 by applying air pressure to the film. First black soap films are of variable thickness determined by a balance of van der Waals attraction and contact pressure against double-layer Mysels and Jones A . D. ROBERTS AND D. TABOR 247 repulsion. Their results compare favourably with ours (fig. 6) which suggests that the charge distribution is similar in both cases. In making this comparison it is necessary to consider the magnitude of the van der Waals attractive forces between the rubber and glass. Calculations show that these forces are small compared with the applied pressure so that the true mean pressure on the film between rubber and glass is the applied load divided by the contact area.With our apparatus it is possible to set the glass surface in motion in a tangential direction and measure the frictional force experienced by the rubber hemisphere. When such experiments were carried out at different speeds of continuous sliding under a constant load it was found that at all speeds the frictional force was minute. The results (table 1) show this. At all speeds the effective liquid viscosity calculated from these data is less than 10 cP whether the film be 200 or 2,000 thick. - Results for suspended soap films ; 0 results for films between rubber and glass ; - - - theoretical results. 2 5 0 - h 5 f 24 .% 150- 6 E 100- I I I I 2 3 4 5 contact pressure (dyn cm-2 x lo5) FIG.6.-Comparison of SDS results for suspended soap films and for films sandwiched between rubber and glass. 8 = surface potential. TABLE 1 .-CONTJNUOUS SHEAR OF THIN LIQUID FILMS shear speed u film thickness h frictional force F (cm Is) (A) (I31 0.010 270 0.05 0.027 370 0.03 0.044 460 0.03 0.078 570 0.03 0.096 630 0.04 0.156 910 0.04 0.242 1 1 0 0 0.04 0.31 1 1,400 0.04 0.426 1,720 0.04 effective liquid viscosity 1 (CP) 8 3 2 1 1 1 1 1 1 Load 50 g rubber sphere constants radius of curvature 2 cm elastic modulus 6 x lo6 dyn Lubricant 0.3 % aqueous SDS. It is always difficult to decide what fraction of the frictional force is due to viscous shear alone and what is due to surface asperity interactions. In our particular case some insight into the matter is gained by examining the change in film thickness h 248 MECHANICAL PROPERTIES OF FILMS with velocity 2 of sliding.It was found (see fig. 7) that h was proportional to u 0 s 6 for film thicknesses between 400 and 2,000 A. Herreburgh has derived an equation h = 1.2 q 0 a 6 uo.6 R 0 s 6 w-Os2 E-Oe4 for predicting the minimum film thickness in an isoviscous elastohydrodynamic line contact for rubber-like materials where is the lubricant viscosity under normal pressures R the radius of curvature of the rubber- bearing surface of elastic modulus E and w the contact load. His result is in good agreement with the theoretical and experimental line contact results of Roberts and Swales for films about 20,000A thick. Therefore it is most interesting to see that our experimental results for a point contact in the film thickness range 400-2,OOO A also show the same variation in h with v as Herreburgh predicts.Thus it seems reasonable to calculate the liquid viscosity of our very thin films using his equation. It is only an approximate calculation and assumes a 40 % drop in h for a " point " contact as compared with a " line " contact due to increased side leakage of lubricant. According to Dowson and Higginson6 this is a fair assumption. The calculation yields a liquid viscosity of 2 cp for film thickness data in the range 400-2;OOO A. The " high " deviation in h below 400 A (fig. 7) may be due to asperity support electrical double-layer support or increased liquid viscosity. Electrical double-layer support at a distance of 200 8 would not be insignificant-the equilibrium distance for this load I I c -2 -1'5 -1.0 - 0 .s 0 0 logto v FIG. 7.-Variation in dynamic minimum film thickness h with sliding speed u for thin (200-2 000 A) aqueous films of SDS sandwiched between a rubber sphere and glass plate. The continuous line hasa slope of 0.60. is 110 &-so the " high " values in h may be mainly due to this factor. Thus it may be that the viscosity of water in films only 200-300 A thick sandwiched between solid surfaces is no greater than the bulk viscosity. The preceding sliding experiments were carried out at arbitrary values of the film thickness determined by the sliding velocity. A further series of experiments were carried out in which the surfaces were first allowed to reach their static equilibrium separation determined by the balance between normal load and repulsive double- layer forces.The surfaces were then set in relative motion. The frictional force increased rapidly as the glass surface was set in uniform motion reached a maximum and then fell as a relatively thick film formed between the surfaces. The peak value corresponds to the force required to shear the equilibrium film. These results are shown in table 2. For films thicker than 80A the friction is low and increases steadily as the film thickness diminishes. This is probably due to viscous shear of the A. D. ROBERTS AND D. TABOR 249 liquid film alone. For thinner films the friction is appreciably higher. Presumably shear of the SDS monolayer is involved or there may even be some rubber-glass contact. As regards the liquid viscosity of these extremely thin films the instantan- eous frictional force is complicated by viscoelastic relaxations of the rubber surface so no estimate may be made of viscosity.Only continuous shear experiments in which dynamic equilibrium has been established may be used for viscosity determina- tion. However the instantaneous results are useful for relative assessment. TABLE 2.-INSTANTANEOUS SHEAR OF EQUILIBRIUM FILMS molar conc. molar conc. equilibrium coefficient of SDS of NaCl film thickness A of friction peak value of 0.01 none 0.01 0.01 0.01 0.02 0.01 0.04 0.01 0.08 0.01 0.16 0.01 0.30 none 0.1 none 0.01 none 0.001 distilled water dry contact 150 120 95 80 70 60 55 40 40 90 40 " zero " 0.15 0.18 0.27 0.35 0.60 0.90 1.16 4 5 4 5 9 Load 5 g contact pressure 3 x lo5 dyn/cm-2 With distilled water alone the instantaneous friction is very high and only a little lower than for dry surfaces.A few experiments have been carried out with small amounts of strong electrolytes added to water with no SDS present. Unstable equilibrium films have been obtained with these solutions but they do not provide very effective lubrication. Films of strong electrolytes are very sensitive to contamin- ants and easily collapse leaving only a few monolayers of liquid between surfaces. Our thickness measurements at this stage are not very accurate but these films appear to be about 40 A thick. Collapsed films of distilled water also appear to be about the same thickness. Films of this thickness are termed " second black " and tend to be of constant thickness ; the nature of the forces determining their structure is uncertain.We conclude that in the presence of sodium dodecyl sulphate solution double- layer electrical repulsive forces occur between rubber and glass surfaces which are able to support pressures of the order of 104-106 dyn cm-2 (i.e. 10-2-1 atm). An oriented monolayer is formed on the rubber surface and perhaps on the glass too. Friction measurements in the presence of the SDS solution suggest that under pressures of about 1 atm two factors operate to provide effective lubrication-the electrical repulsive forces which tend to keep the surfaces apart and the SDS monolayer itself which protects the surfaces from intimate contact if the separating liquid is locally penetrated. The viscosity of the liquid (0.3 % SDS in water) sandwiched between the monolayers of SDS appears to be constant approximately (the bulk viscosity) whether it be 200 or 2,OOOA thick.USE OF MICA AND RUBBER FOR THIN FILM STUDIES In the two investigations that have been described two different types of surface In many ways mica is the ideal were used to trap the thin film under examination. 250 MECHANICAL PROPERTIES OF FILMS surface. It is molecularly smooth and simple to prepare. Films sandwiched between two of these surfaces may be resolved to + 3 .$ though the optics for doing this is cumbersome. Rubber cannot be made molecularly smooth the preparation of an optically smooth surface is technically difficult and films squeezed between it may only be resolved to about 40 A. Yet it has been found useful because when placed in contact with another material it provides a parallel contouring surface over relatively large areas. In practical manipulation optical rubber has advantages over mica. It is not fragile does not have to be silvered and is easy to mount into apparatus. Rubber has one other advantage. It is a pure van der Waals force solid whereas mica is ionic. Ion-ion interactions could in certain circumstances confuse the interpretation of results. We thank the Avon Rubber Company Ltd. for preparing the optically smooth rubber surfaces. A. I. Bailey and J. S. Courtney-Pratt Proc. Roy. SOC. A 1955 277 500. ’ A. D. Roberts Eng. Materials Design 1968 11 (4) 579. K. J. Mysels and M. N. Jones Disc. Furachy SOC. 1966 42 43. K. Herrebrugh A.S.M.E. J. Lubric. Tech. 1968 90,262. A. D. Roberts and P. D. Swales Brit. J. Appl. Phys. 1969 2 1317. D. Dowson and G. R. Higginson Elasfohydrodynamic Lubrication (Oxford Pergamon Press 1966).

ISSN:0370-9302    

DOI:10.1039/SD9700100243

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Smectic Model for Liquid Films on Solid Surfaces Part 1 .-Application to Monolayer Boundary Lubrication. BY E. DRAUGLIS A. A. LUCAS* AND C. M. ALLEN Battelle Memorial Institute Columbus Ohio U.S.A. Received 30th April 1970 A qualitative theory of films of hydrocarbon solutions of fatty acids described in a previous publication correlates many experimental data. According to this theory the fatty acid molecules and solvent molecules are arranged in layered structures similar to those in smectic mesophases. A simplified method for calculating the attractive energy between two such layers is described. The fatty acid molecules are assumed to behave as hard rods of constant electric polarizability and diamagnetic susceptibility. A potential describing the interaction between two rods is derived and used in a lattice sum calculation to compute the total energy of interaction of two layers as a function of average separation of molecules within a layer relative position of the two layers and the distance between layers.Calculation of frictional force and coefficient of friction by means of this model show good agreement with a similar calculation of Cameron if the proper assumption about the relative magnitude of energy as a function of relative position is made. 1. INTRODUCTION The question of whether liquid films possess properties significantly different from bulk properties when they are very close to solid surfaces has been of great interest. Griffiths and Bulkley are among the earliest investigators who studied the problem experimentally. Griffiths seems to have observed some surface film rigidity while Bulkley did not.Bastow and Bowden studied the problem by measur- ing the separation of glass plates immersed in the fluid of interest and obtained the the same separation for fluids as for air. Later these workers measured the flow of various liquids between two clean parallel glass plate^.^ No evidence was obtained that the fluid flow properties of ordinary liquids were different in thin films than in the bulk. Not unexpectedly solutions of ammonium oleate which can exist in a liquid crystalline phase did show anomalies. Positive evidence for the existence of anomalous films was obtained by Needs.s In his experiments he squeezed various fluids between two steel flats and observed stable residual films some 10,OOOA thick. performed similar but more carefully controlled experiments on well-characterized hydrocarbon solutions of fatty acids as well as some oils of practical interest.For the few systems where his results could be compared with those of Needs he obtained lower values for the film thickness than Needs although his values were almost always greater than 1,OOOA. Besides investigating the thickness of the films as a function of loading Fuks studied various other properties of the films as a function of fatty acid concentration surface Later Fuks 6 * * now Charge de Recherches au Fonds National Belge de la RecherchC Scientifique at Dept. of Physics University of Liege Liege Belgium. 25 1 252 SMECTIC MODEL FOR LIQUID FILMS energy of substrates on which the films were formed and temperature. Shear stress measurements were made on various systems for various normal stresses.Many of the measurements were made on various combinations of solvent and fatty acid and correlations of the data with hydrocarbon chain length were observed. These measurements constitute much experimental data in support of the existence of multimolecular films many 1,000 thick. One group of critics claims that the measurements themselves do not substantiate the presence of multimolecular films and the data can be explained as experimental artifacts caused by the presence of dirt and impurities in the fluids and asperities on the substrates. The work of Hayward and Isdale* provides an example of this viewpoint. A refutation of this argument is the general internal consistency of the data. A second body of criticism admits the validity of Fuks’ data but claims that other phenomena such as the deposition of particles of solid fatty acid or the formation of micelles due to the presence of water could account for Fuks’ results.The idea is that no continuous film having unique properties exists. Refutation of this type requires much experimental investigation of the structure and properties of the films formed in Fuks’ apparatus and a reasonable theoretical explanation of not only the phenomenological data of Fuks but also the results of any new experiments on the structure and ordering properties of the films. A reasonable approach to the development of a theory of the films is first to identify which of the various factors are the most dominant or to identify one or two physical features which must be present no matter what the complicating factors are.Then one can construct a comparatively simple model in which only these features are considered. Such a model has been described in a recent review on boundary l~brication.~ In this model the structure of the films is assumed to be similar to the structures found in smectic mesophases i.e. the molecules are arranged in layers with the long axis of each molecule perpendicular to the plane of the layers. Two assumptions are necessary to account for the formation of such structures. The first is that fatty acid molecules (and straight-chain hydrocarbon solvent mole- cules) within a few 1,000 of a solid surface behave as rod-like molecules and tend to align themselves parallel to each other by means of their mutual van der Waals’ forces to form ordered domains or clusters.The second assumption is that the electric dipole field caused by the formation of Fe-0 bonds arising from the chemisorption of a monolayer of fatty acid and the van der Waals wall forces are sufficiently strong to align these domains so that they are all perpendicular to the substrate. If these two assumptions are valid then one could expect a coherent continuous film having a structure similar to a smectic liquid crystal to form. As shown in ref. (9) this model explains qualitatively most of the data of Fuks. However there has been much criticism of this interpretation. 2. ATTRACTIVE INTERACTION ENERGY BETWEEN TWO ROD-LIKE MOLECULES Because of the complicated nature of the problem it is not feasible to create from first principles a general theory capable of correlating all of the data of Fuks.Instead we shall attempt to develop only a theory of the shear properties of the films based on our qualitative model. The first step is the development of methods for the calculation of the energy of interaction between two ideal smectic layers and the interaction energy of a molecule of a layer with respect to all other molecules in the layer. In principle such calculations are straightforward because only van der Waals’ forces need be considered. To obtain the interaction energy between two E . DRAUGLIS A . A . LUCAS A N D C . M . ALLEN 253 layers one merely sums pairwise the interaction energy between the given molecule and each molecule in the next layer and then multiplies by the number of molecules per unit area to obtain the energy per unit area.To obtain the energy of a molecule within a layer one sums the interaction of the molecule with every other molecule within the layer. In our calculation we shall simplify this process by deriving an analytical expression which approximates the attractive energy between two molecules. The basis for our calculation is the London-Kirkwood expression for the attractive part of the potential due to an atom or group of atoms l o F~ = -3mc2ax/r$ (2.1) where V l j is the attractive energy between CH2 group i of one molecule and CH group j of another molecule rij is the distance between two such groups rn is the electronic mass (9.109 x lo-,* g) a is the electric polarizability of a CH2 group (1.777 x cm3) and x is the diamagnetic susceptibility of a CH2 group (19.94 x The total attractive interaction energy of a molecule with respect to all the other molecules in an adjacent layer can be obtained by summing Vrj for all the CH2 groups within the molecule with all the CH2 groups in the molecules in the adjacent layer.(In the first approximation the end groups may be treated as CH groups.) To simplify the procedure we assume that each molecule can be replaced by a rod the electric polarizability and diamagnetic susceptibility of which is constant along the length of the rod. An expression for the attractive interaction energy of two parallel rods can be obtained by integration over the rods. The V, of eqn (2.1) are replaced by (2.2) cm3). dV = - 1.5 mc2aX(NN'/LL') dzdz'/r6 = - A ("'ILL') dzdz'/r6 where L and L' are the respective lengths of the molecules N and N' are the number of CH2 groups per molecule dz and dz' are elements of length and r is the distance between dz and dz'.Two distinct expressions are to be derived-the first of which gives the attractive energy for two rods in different layers as a function of interlayer separation and distance between rods within a layer and the second of which gives the attractive energy of two rods within the same layer as a function of distance between the rods. The interaction energy for two rods in different layers may be obtained as follows. Let a rod of length L be situated on the z axis with its lower end a distance E above the x-y plane and let a rod of length L' be situated parallel to the z axis below the x-y plane at a distance p = (x2+y2)* from the origin and have its upper end a distance below the x-y plane.The interaction energy is then found by evaluating the integral obtained from eqn. (2.2) (lengths of both rods) &+L' -E = -r$)AJz dzJ [ x ~ + ~ ~ + ( z - z ' ) ~ ] ' 3d~'. -L-E Assuming p not equal to zero and carrying out the integrations gives 254 SMECTIC MODEL FOR LIQUID FILMS where the function F(U,) is given by F( Ui) = - I/( 1 + U;) + 3 ui arctan u and u1 = (2&+L‘)/p; u2 = 2 & / p ; u = (2&+L+L‘)/p; u4 = (2&+L)/P. For the case p = 0 one obtains v(o,&) = - (NN’A/~oLL’)[(~& + ~ 7 - 4 - (24-4 - (2& +L + ~ ’ ) - 4 + (2& +~)-41. (2.5) The interaction energy for two rods within the same layer may be found in a similar manner. Evaluation of the integral V(p) = - L-2N2A dz dz’[p2 + ( z - z ‘ ) ~ ] - ~ 1 1 yields V(p) = (N2A/4p4L2)[3(L/p) arctan (Lip) + L2/(L2 + p 2 ) ] .3. MONOLAYER BOUNDARY LUBRICATION Before an attempt is made to use these potential functions to the multilayer films of Fuks it seems prudent to apply it to a simpler system for which either experimental data or theoretical calculations are already available. An example of such a system is that considered by Cameron in his theory of the role of van der Waals’ forces in monolayer boundary 1ubrication.l In this work the frictional force between two rubbing surfaces is assumed to arise from the van der Waals forces due to a single layer of stearic acid on each surface. The computation of the frictional force is performed by calculating first the attractive energy of interaction of a molecule in one layer with all the other molecules in the other layer by summing the contributions to the potential of all the CH2 groups within a radius of 8 A of the molecule.The rij are taken to be equal to the distances between X-ray scattering centres. The repulsive energy is taken as equal to the repulsive energy of the hydrogen atoms of the molecule that approach closest as the two layers are pressed together. Cameron’s calculation shows that this is negligible and we therefore omit it from our calculation. When one surface is moved parallel to the other from the stable equilibrium position to the next equilibrium position a point between these is encountered where the total interaction energy E2 is a minimum. If El is the energy at the equilibrium point and x is the distance between equilibrium points then the mean force F needed to move a chain from one equilibrium position to the next is give by El -E2 = Fx.(3.1) To simplify the calculation Cameron argues that E2 is negligibly small in comparison with El and therefore the frictional force is given by F = EJx. ( 3 4 This assumption has been questioned by Akhmatov and others.12 Our calculations seem to verify this criticism. To compute the attractive energy between two layers as a function of relative position and separation of the layers we carry out a simple summation procedure. We first consider a molecule of length L the major axis of which is parallel to the E . DRAUGLIS A . A . LUCAS A N D C. M. ALLEN 255 z axis located at the point (x,y) above an infinite layer of molecules of length L. These molecules are assumed to be arranged in a square mesh of lattice constant a.The origin of the x-y coordinate system will be taken on one of the mesh points. The horizontal distance plm between the isolated molecule and a molecule in the layer located at the point whose coordinates are ma and la is given by prm = [(x - + ( y - (3.3) The total attractive energy is obtained by substituting this value of p in eqn (2.3) and (2.5) and summing over m and 1. That is the total attractive energy is These summations must be done numerically for given values of lattice constant a and layer separation d = 2 ~ . Eqn (3.4) was summed numerically over a 31 x 31 square lattice by means of a CDC-6400 computer for several values of d x and y. Results of this computation are given in table 1 for values of the parameters used by Cameron i.e.a = 4.46 A TABLE ATTRACTIVE ENERGY AS A FUNCTION OF INTERLAYER DISPLACEMENT ( y = 0.0) position x energy of attraction (in units of a) (ergslcmz) 0.0 15.824 0.1 15.708 0.2 15.409 0.3 15.053 0.4 14.774 0.5 14.669 d = 3.09 i! ; a = 4.46 A. d = 3.09& N = 18 L = 24.69& and a surface concentration of 5 x 1014 cm-2. In this table the value of the energy for the point (0,O) is El and the value of the energy for the point (0.5a7 0) is E2. Calculations for many different values of d show that for no reasonable value of d can E2 be neglected. This is shown in table 2. In this table Emin is the energy at the point (OSa 0.5a). TABLE 2.-DEPENDENCE OF El Ez AND Emin ON INTERLAYER SEPARATION interlayer separation A 2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4 3.5 4.0 Emin El (ergslcrnz) (ergs/crnz) 20.298 26.802 19.020 24.195 17.842 21.975 16.756 20.067 15.572 18.414 14.828 16.972 13.973 15.704 13.181 14.583 12.448 13.586 11.769 12.694 11.139 1 1.892 8.590 8.867 a = 4.4GA.E2 (ergs /cmz) 22.148 20.454 19.115 17.812 15.754 15.521 14.575 13.681 12.864 12.114 1 1.425 8.704 Table 3 gives a comparison of our results with those of Cameron. The quantity F is calculated both from eqn (3.1) and (3.2) for various values of d which are to be 256 SMECTIC MODEL FOR LIQUID FILMS TABLE 3.-cOMPARISON OF HARD-ROD CALCULATION WITH CAMERON’S RESULTS P Fl2 F1 Fcameron 0.001 3.09 0.052 0.709 1.2 6.33 2.89 0.119 0.826 1.52 13.0 2.69 0.128 0.985 1.93 19.5 2.49 0.209 1.202 2.29 The first column gives the normal pressure the third column gives the frictional force calculated by eqn (3.1) the fourth column gives the frictional force calculated by eqn (3.2) and the last column is the frictional force as calculated by Cameron.ll (kgflcmz x 103) d (A) (kgf/cm* x 103) expected for the values of P shown if the linear compressibility is taken as 10 x dyn/cm2 as in ref.(1 1). Our results for Fagree within a factor of two with Cameron’s results if eqn (3.2) is used. However if eqn (3.1) is used our results differ by about an order of magnitude both from Cameron’s results and experiment. 4. DLSCUSSION The fact that our results for Fshow good agreement with the exact group-by-group calculation of Cameron if eqn (3.2) is used indicates that our rigid rod approximation is sufficiently accurate to encourage its further application to more complicated systems such as the multilayer model for the films of Fuks.A simple treatment of such a model will be used by the authors in a future calculation in which the assump- tion is made that the molecules are arranged in regularly spaced arrays within the layers as they are in the first chemisorbed layer. Shear forces for such a model will be calculated as a function of interlayer separation and distance between molecules within a layer. Comparisons will be made with Fuks’ data for reasonable values of the distance parameters. Good agreement with the data will probably not be obtained because one does not expect periodic spacing of the molecules to prevail except in the first layer. However such calculations will probably be useful when applied to known smectic liquid crystals in which periodicity does exist within layers.This work was in part supported by Naval Air Systems Command Contract No. N-00019-70-C-0139 Battelle Institute and the Columbus Laboratories of Battelle Memorial Institute. A. A. Griffiths Phil. Trans. A 1920,221,163. S. H. Bastow and F. P. Bowden Proc. Roy. SOC. A 1931 134,404. S. H. Bastow and F. P. Bowden Proc. Roy. SOC. A 1935,151,222. S. J. Needs Amer. SOC. Mech. Eng. Trans. 1940 62 331. G. I. Fuks Research in Surface Forces ed. B. V. Deryagin vol. 1 (Consultants Bureau New York 1964) p. 79. G. I. Fuks Research in Surface Forces 1966 2 159. C. M. Allen and E. Drauglis Wear 1969 14 363. lo J. G. Kirkwood Phys. Z. 1932 33 57. A. Cameron Amer. Sac. Lubr. Eng. 1960 2 195. l2 A. S. Akhmatov Molecular Physics OfBoUntdary Lubrication Fiziko-Matematicheskoi Literatury MOSCOW 1963 (translated and published in English by Israel Program for ScientificTranslations Ltd. Jerusalem 1966) p. 331 ff. 2R. Bulkley Bur. Stand. J. Res. 1931 6 89. * A. T. J. Hayward and J. D. Isdale Brit. J. Appl. Phys. 1969 2 2 251.

ISSN:0370-9302    

DOI:10.1039/SD9700100251

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

GENERAL DISCUSSION Prof. J. LykIema (Wageningen Netherlands) said With reference to the paper by Churayev et al. when liquids are forced through porous media or narrow capillaries the resistance against flow is not only of a hydrodynamic but also of an electrokinetic character. The principle is that between the ends of the capillary a streaming potential is generated which in turn gives rise to a counter electro-osmotic flow. EspeciaIly in narrow capillaries tlus counter-osmotic flow can attain values that are comparable with the induced Poiseuille flow and due account of this complicating effect should be given before something definite can be said with respect to either the viscosity of the liquid or the effective radius of the capillary. As a first approximation the order of magnitude of the electrokinetic contribution can be assessed as follows.The streaming potential V generated by the pressure P amounts to v = &cP/4nqrc (1) where the symbols have their usual meaning. The velocity of the counter-flow obeys If up is the average liquid velocity due to the Poiseuille ff ow this quantity follows from Q = nr2tl = nr2P/8ql (3) v,/v = ~ ~ c ~ / 2 q n ~ 1 ~ r ~ (4) Hence indicating that the ratio between the primary Poiseuille flow and the generated counter- flow can become appreciable in capillaries of small radius r. For aqueous solutions at room temperature uE/vp = 3.8 x 10-19 52/Kr2 (5) if [ is in mV K in Q-l cm-l and r in cm. As Chuayev et al. do not give the quality of their water nor its pH it is not certain what values for 5 and IC should be substituted.If IC- i2-l cm-l and 5-50mV VE/V)P N 1 0 - y r 2 = 10-*/r2 % (6) indicating a reduction of 1 % in capillaries of 1 p and of 100 % in capillaries of 0.1 p. Comparing these conclusions with the experimental results plotted as an apparent increase in viscosity in fig. 4 of Churayev et al. it appears that the reduction in flow could well be due completely to counter-osmotic flow so that there is no indication of thick viscous layers or enhanced viscosity. The obvious experiment to ensure absence of electrokinetic counter-effects is repetition of the author’s experiments in electrolyte solution instead of pure water because addition of salts reduces t and increases K (see eqn (5)). The effect of electrolyte on q is of secondary order. Dr. N. V. Churayev Dr. V. D. Sobolev and Dr. Z. M. Zorin (Acad.Sci. Moscow) (communicated) In reply to Lyklema to evaluate the eIectroviscous effect Lyklema uses the Helmholtz-Smoluchovsky eqn (I) which cannot be used for thin capillaries where overlap of the diffuse ionic atmospheres takes place. Our calculations based 257 258 GENERAL DISCUSSION on formulae taking into account the overlap 1* show that the electroviscosity effect is considerably smaller 2 % for [ = - 15 mV and ic = 2 x Q-l cm-l. We do not understand why Lyklema has taken in his calculations too high a value of 5 = - 50 mV which is well known not to be characteristic of the quartz-water interface. Moreover the absence of considerable electroviscosity effect was proved by direct experiments with Though diffuse ionic layers were depressed we obtained the same viscosity change q/qo with decreasing radii of capillaries as for clean water.Thus our experimental results cannot be explained by electroviscosity effect (or counter-osmotic flow) which was naturally taken to account and specially verified. Dr. G. Peschel (Universitdt Wiirzburg) said In the viscosity measurements by Churayev et al. of water with very thin capillaries they obtained the surface zone viscosity of water structurally changed by adjacent solid walls. We also carried out experiments for determination of the surface zone viscosity of pure water. In the method developed by us we brought together two polished fused silica plates-one plate is planar the other spherically formed with a curvature radius of 100 cm- within a very small distance under the action of a constant force.Both plates were immersed in the liquid in question. The movement of the spherical plate was registered by a special device. The theory which we developed also takes into account the surface roughness of the plates. For water we found an enhanced surface zone viscosity which is detectable up to plate distances of about 1,500A and is sensibly dependent on the outer force and on the degree of coverage of the silica surfaces with hydroxyl groups. The main point is that the surface zone viscosity showed pronounced maxima at about 15 32 45 and 61°C ; these temperatures are known to be the temperatures of higher-order transition of the water structure. The most pronounced maximum was at about 32°C. Churayev et al. too find a maximum at about this temperature. May be they could also detect the other maxima if their data were more closely spaced.The anomalous effect we found vanishes at about 70°C. The effect found by them trails off at the same temperature. In this respect our work agrees well with their evidence. Are their measurements disturbed by dust particles ? Dr. B. A. Pethica (Unilever Res. Port Sunlight) said In an earlier paper Deryaguin and his coworkers measured the shear moduli of thin liquid layers on quartz and estimated the altered region to be ten times larger than that reported by Churayev Sobolev and Zorin in their present paper the latter estimate being based on viscosity measurements. Do these two experimental methods measure different aspects of the boundary layer properties or do they refer to different surface characteristics in the quartz used? Dr.B. Stuke (Phys. Chern. Inst. Miinchen) said With regard to the paper by Churayev et al. some time ago we have done exactly what Lyklema has suggested and came to the same conclusion. The time was measured for a given amount of liquid to pass through filters of sintered quartz (bacteria filter with guaranteed maxi- mum pore size Schott u. Gen.) by pumping under manostatic conditions. Three N KCl solution in the same quartz capillaries. N. V. Churayev and B. V. Deryaguin Dokl. Acad. Nauk. 1966,169,396 ; KoIloidZhur. 1966 28 751. N.- V. Churayev and B. V. Deryaguin Research in Surface Forces (Nauka Moscow 1967) p. 295. Lars Lidstrom Actn Polytechn. Scand. 1968,75,100. Diplomarbeit L. Kolodziej (Munchen 1962). GENERAL DISCUSSION 259 different filters were used with maximum pore sizes 1.2,1.7 and 3 p.For two different electrolyte solutions with concentrations ~ 0 . 0 1 N the ratio of times z was equal to the ratio of the corresponding viscosities q, i.e. q / z k = v]i/qk. For solutions with concentrations < 0.01 N (index 1) compared with solutions with concentrations > 0.01 N (index 2) the result was zl/z2>q1/q2 depending (monotonically) on the concentration of solution 1 and pore size. The value of the ratio passing times for pure water compared with an electrolyte solution of sufficient concentration (maxi- mum value) under conditions where qHzO/q2 - 1 for the filter with 1.2 p was zH20/~2 + 1.05. Prof. N. V. Churayev (Moscow) said In reply to J. Lyklema and B. Stuke they were of the opinion that the increase of viscosity of water in thin quartz capillaries observed by us could be connected with the electroviscosity effect.To evaluate this effect Lyklema uses the Helmholtz-Smoluchovski equation the applicability of which is limited by the condition of xr% 1 (where I/K is the Debye radius of the diffusion-ion atmospheres Y is the radius of the capillary). For water (IC = lo5 0 cm-l) and capillaries with r = 10-4-4 x cm in our experiments the values of ICY are in range 10-0.4. Thus for the thinnest capillaries where the greatest change of viscosity is found the application of the Helmholtz-Smoluchovski equation becomes impossible. For w z l one should take into account the overlap of the diffusion ion atmospheres that are considered theoretically in ref (1) and (2). Basing on the equations derived there we may evaluate the influence of the electroviscosity effect in the case of interest.For the quartz/water interface at pH 6.5-7 [ = - 15 mV ; the specific conductivity A of distilled water was 2 x 10F 0-l cm-l . The calculations show that the maximum variations of viscosity q not to exceed 2-3 %. Moreover the highest viscosity changes would have been observed at 1 ~ r z 2 ~ 7 cm and this was not found. All this leads to the conclusion that the results obtained could not be ex- plained by a simple influence of electroviscosity. However we did not confine ourselves to theoretical estimations but undertook additional experiments with the electrolyte solutions. In fig. 1 are the results of comparative measurements of viscosity of water and a N KCl solution in the same quartz capillaries (Y = 6.9 x cm and r = 8.9 x cm).As the IC values with electrolyte are higher by an order of magnitude than those with distilled water one would have expected an abrupt drop of viscosity of the solution in the microcapillaries if the cause of changes in q were due to the electroviscosity effect. As the plot however shows the experimental points for water and the solution lie on the same line corresponding to the increased viscosity values q/q0 = 1.27 and 1.34. N KCl solution could not have been caused by the electroviscosity effect. Calculations using eqn (5) employed by Lyklema (the equation can be applied in this case for K r z 5) show that the differences in viscosities would be as small as ca. 2 %. Thus the experimental data are also in contradiction to his hypothesis of the effect of electroviscosity.The explanation for the increased viscosity of water in fine-porous bodies is usually given from the standpoint of the electroviscosity effect as has been done in N. V. Churayev and B. V. Deryaguin Dokl. An S.S.S.R. 1966,169,396 ; Kolloidn. zhurn. 1966 28 751. Sb. Issledovaniya u oblastipouerkhnostnykh sil (Research in surface forces) “ Nauka ” (Moscow 1967) p. 295. C. L. Rice and R. Whitehead J. Phys. Clietn. 1965 69 4017. L. Lidstrom Acta Polytechn. Scand. 1968 75 100. J. 7’. Davies and E. R. Rideal Interfacial Phenonzeiza (Academic Press 1961) p. 126. i.e. for the capillary radii r z 2 x An increase of 30 % viscosity of 260 2 0 I0 CI) 1 > 0 I0 2 8 20 10 m \ E * o > 10 20 GENERAL DISCUSSION P atm FIG. 1.-Velocity v of motion of liquid in quartz microcapillaries against pressure P of gas in the chamber (a) r = 690 A ; t = 20.5"C ; I = 4.21 cm; 1 water ; 2,3 N KCl; ? / T O = 1.34.(b) r = 840A; t = 20°C ; I = 3.56 cm ; 1 water ; 2 N KCl ; q/qo = 1.27. P atm GENERAL DISCUSSION 26 1 the paper of B. Stuke. It probably does occur in a number of systems. Neverthe- less for fine-porous bodies (generally at K r - 1) if one does not consider the overlap of diffusion ion atmospheres l’ the influence of electroviscosity is over-estimated. Her~niker,~ while measuring the flow potential by short-circuiting the electrodes did not detect changes in flow rate though the viscosity of the liquid was higher by more than 20 %. Deryaguin Zakhavayeva and Lopatina4 have observed increases of several fold of the viscosity in thin pores of clays at a concentration of electrolyte of ca.0.1 N when the diffuse layers were totally suppressed. It should be noted that the highest viscosity differences are obtained in very thin layers and pores 5 * when (at ~ r - + 0 ) the influence of the electroviscosity effect does not grow rapidly (a con- sequence of the erroneous application of the Helmholtz-Smoluchowski equation) but in contrast falls abruptly. 9 All this leads to the conclusion that the real cause of the viscosity changes is the modification of the structure of liquids under the influence of surface forces. The nature of the changes depends not only on the radii of the capillaries but also on the temperature of the liquid and/or the concentration of electrolytes that influence the structure of the boundary layers. found an increase in shear modulus for the boundary layers of polar liquids with thicknesses of the order of cm from measurements by the resonance method on piezo-electric quartz at a frequency 75 kHz.This does not mean however that anomalous layers of the same thickness should be also observed during the viscous flow of a liquid under a constant shearing force. The boundary layers of polar liquids seem to have complex str~cture.~ The layers of water in close proximity to the surface are modified to a greater extent and have in particular a considerable yield value 6 amounting to ca. 100 dyn according to Nerpin and Bondarenko.lO This result was obtained by observing changes in the filtration permeability with increase of the pressure gradient. It follows from the estimation of 6 that these layers could not take part in flow in our experiments and are caused only by the “ narrowing ” of capillaries.The thickness of the boundary layer was determined l 2 to be 80 A i.e. close to our value. At the same time a broader boundary region with less modified properties appears to exist. Its limits are observed e.g. in measurements of the shear modulus and the dis- joining pressure. l 3 In reply to Pethica Bazaron Deryaguin and Bulgadayev ’ 9 Mr. D. W. J. Osmond (Z.C.I. Ltd. Slough) said It has long seemed to us in industry that the thick layers of modified solvent postulated to form at interfaces are N. V. Churayev and B. V. Deryaguin Dokl. An S.S.S.R. 1966,169,396; Kolloidn. zhurn. 1966 28 751. Sb. Zssledovaniya v oblasti poverkhnostnykh sil (Research in surface forces) “Nauka” (Moscow 1963 p.295. C. L. Rice and R. Whitehead J. Phys. Chem. 1965 69,4017. J. C. Henniker J. Colloid. Sci. 1952 7 443. B. V. Deryaguin N. N. Zalchavayeva and A. M. Lopatina Sb. Issledovaniya v oblasti pover- khnostnykh sil (Research in surface forces) (“ Nauka ” Moscow 1961) p. 175. Bull. Rilem. 1965 27 27. G. Peschel Muter. constr. 1968,1,529 ; G. Peschel and K. H. Adhger Naturwiss. 1969,56,558. Z. M. Tovbina Sb. Issledovaniya v oblasti poverkhnostnykh sil (Research in surface forces) (“ Nauka ” Moscow 1967) p. 24. ’ U. B. Bazaron B. V. Derjaguin and A. B. Bulgadayev Dokl. AN S.S.S.R. 1965 160 799; 1966 166 639 ; Zhurn. exper. teor. jiz. 1966 51 969. * B. V. Deryaguin Disc. Faraday Soc. 1966,42 109. W. Drost-Hansen Znd. Eng. Chem. 1969,61 10. l o S. V. Nerpin and N. F.Bondarenko Sb. Trudovpo agrojizike 1969 19 27 l 1 G. Peschel and K. H. Adlfinger Naturwissen. 1967 54 614, 262 GENERAL DISCUSSION implausible except as in Cameron’s paper as chemical or other artifacts. Our reasons for doubt are as follows. It is now a commonplace to prepare concentrated dispersions of both polar and non-polar particles in organic liquids often in the presence of fatty acids. In such dispersions the particles radius is often about 1,000 8 and the phase volume about 50 %. If layers of 1,000 A thickness surround such particles seven times as much solvent as is present would be required to form the layers. We may safely assume therefore that all of the solvent present would be heavily modified if such layers existed. Dispersions in a continuum of “ modified” solvent of this type would be expected to show marked rheological anomalies especially at low rates of shear.In practice such dispersions are commonly mobile liquids whose chief anomaly is intense shear thickening or dilatancy at very high rates of shear. It therefore seems that either such layers do not exist or they are so weak as to have little practical effect. On the other hand it is well known in the Paint Industry that fluid dispersions in inert non-aqueous media of particles of reactive materials such as calcium carbonate zinc oxide or basic lead carbonate are converted on the addition of fatty acids to gelatinous dispersions. It has long been accepted in agreement with Cameron’s results that this gelling is due to thicken- ing of the continuous phase by metal soaps formed by chemical attack on the particle surfaces.Dr. A. Cameron (Mech. Eng. Dept. Imperial College) said Osmond’s results from paint technology confirm that when the liquid or the components in the liquid do not react chemically with the particles there is no change in the viscosity. When they do react as for whiting and a fatty oil the result is a plastic putty. These are completely consonant with our findings. Dr. H. E. Ries (Chicago) said In spite of the ingenious studies on thick surface films described by Smith and Cameron and others I believe that there is something unique about films one molecule thick-partly perhaps because I worked with Harkins and partly because of our research on monolayers extending over the last 35 years. I believe everyone agrees that the monolayer is the last line of defence in many areas-friction wear and rust (also for emulsion and foam stability evaporation retardation etc.).Certainly the monolayer is held by forces stronger than those that hold any succeeding layer and on solid surfaces it is the only layer that can be chemisorbed. More specifically OUT radiotracer adsorption studies performed under carefully controlled conditions clearly demonstrate that adsorption levels off at one monolayer. Perhaps the most pertinent experiments were those in which radiostearic acid was adsorbed from n-hexadecane solutions on iron films vapour-deposited in high vacuum on the window of a Geiger tube. Although initial adsorption is relatively rapid the equilibrium adsorption plateau at the monolayer level extends over at least 70 h.l Our mixed-film studies at the water/air interface as well as at the metal/water interface in rust-prevention experiments also support the monolayer concept.I should thus like to ask Smith and Cameron whether some of their work suggests that the monolayer provides protection of some special significance ? Dr. A. Cameron (Mech. Eng. Dept. Imperial College) said In reply to Ries we have done nothing to displace the monolayer. Our studies are solely directed to finding under what circumstances a thick viscous layer could be formed. With D. C. Walker and H. E. Ries Jr. J. Colloid Sci. 1962,17,789. GENERAL DISCUSSION 263 stearic acid and cetane a monolayer occurs both on platinum and on steel but it is only with steel that the 1,OOOA film OCCLWS. Dr. D. Tabor (University of Cambridge) said Cameron’s paper shows that when a surface immersed in a liquid appears to produce long-range effects this is due to chemical reaction and the formation of a relatively thick film of reaction products.These results are in close agreement with measurements made by Bowden and Moore in Cambridge nearly 20 years ago using radioactive tracer techniques. Although this seems fully confirnied for fatty acids I find it hard to believe that dimethyl silicones will produce reaction products at room temperature. Although as Willis observed a solid surface can greatly accelerate the chemical degration of silicones the effects were negligible below about 90°C. If reactions do occur presumably they must be due to impurities in the silicone or to contaminants in the system as a whole. I would welcome Cameron’s comments on this.Finally may I ask a question as a tribologist rather than a surface physicist. Tribologists are interested in two questions first do surfaces produce long-range effects in liquids ; secondly is this long-range effect stable or is it easily disrupted by shear i.e. is the modified film rheologically strong or not? Cameron has given an answer to the first question-could he give one to the second? Dr. A. Cameron (Mech. Eng. Dept. Imperial College) said Tabor may rest assured that we have taken full acount of the work of Bowden and Tingle and Bowden and Moore on the reactivity of stearic acid with metals. In the work he quotes it is interesting to note that after 20 h 4 molecular layers i.e. a lOOA thick layer of stearic acid reacted with zinc. We find an immobilized layer 1,000 thick which we believe is caused by solvent meshed by soap fibrils.From normal grease technology it is reasonable to think that a soap content of about 10 % would be enough to achieve such thickening. This corresponds to 100 A of “ solid ” iron stearate the same 4 layers as Bowden and Moore found. The only difference is that they indicated t h s took 20 h to form at 25°C while we find our films are grown in h. We also wonder if the correction they applied for radioactive material adsorbed on the walls of their vessels was accurate. The aim of their work was quite different from ours which is simply to study the liquid layer at the surface of metals. We certainly are not competent to discuss the interesting low viscosity layer which silicones seem to form at metal surfaces though our result confirms Deryaguin’s.We have not attempted to elucidate the mechanism. In reply to Tabor’s second question work now being carried out by flowing mercury past the films show that they are very fragile. A shear stress of about 5 dyn/cm2 seems to break them down. On the other hand stearic acid seems to promote fluid lubrication (see our ref. (9) fig. 6). Mr. A. J. Groszek (B.P. Co. Ltd. Sunbury) said In view of the resultspublished previously by Cameron I expected that adsorption of stearic acid from n-hexadecane on iron oxides and metals such as iron would give a higher heat of adsorption than the adsorption from a non-matching solvent such as n-heptane. We determined therefore heats of preferential adsorption of stearic acid for the two solvents in the flow-microcalorimeter.The heats of preferential adsorption obtained for three different iron oxides and for ground iron powder are shown in table 1. Stearic acid saturates the oxides at a concentration of 1 g/l. i.e. the concentration at which F. P. Bowden and A. C. Moore Trans. Furuduy Sac. 1950,47 900. D. Tabor and R. F. Willis Wear 1969 13,413. 264 GENERAL DISCUSSION the integral heats of adsorption shown in table 1 were obtained. Some of the heat evolved is due to chemisorption but a large part is caused by reversible physical ads0rption.l TABLE IN INTEGRAL HEATS OF ADSORPTION OF OCTADECANOIC ACID ON TO IRON OXIDES AND GROUND IRON FROM WHEPTANE AND Il-HEXADECANE SOLUTIONS heat of adsorption J m-2 * adsorbent from n-heptane from n-hexadecane y-Fe203 0.042 0.222 y-Fe203-Fe304 0.158 0.126 a-Fe203 0.113 0.158 ground iron 0.276 .0.251 * adsorption from solution containing 1 g/l. octadecanoic acid. Perhaps the most striking result was obtained for y-Fe,O, for which the heat of adsorption from n-C 6 was considerably greater than that taking place on adsorp- tion from n-heptane. The result can be explained by a substantial orientation of n-hexadecane taking place near the surface saturated with stearic acid or perhaps several layers of mixed film forming near the surface. It is striking that the high heat of adsorption is obtained only for one of the three iron oxides examined and that there is no difference in the heats for ground iron which emphasizes the import- ance of the nature of the surface in these adsorption experiments. The effect of the chain length of fatty acids on the heat of adsorption was also obtained for the adsorption of small amounts of fatty alcohols and acids from hexa- decane onto a-Fe203 at low surface coverage as shown in table 2.The C16 acid produces a much greater heat effect than C6 acid although an almost identical number of moles were adsorbed in both cases. These results indicate that the matching of the chain-lengths of the solvent and solute can certainly increase the heat of adsorp- tion and therefore stability of the adsorbed films which agrees with the results reported by Cameron et al. on the relatively high effectiveness of the matched solvent/solutes in certain lubrication experiments. TABLE 2.-PULSE ADSORPTION OF NORMAL ALCOHOLS AND CARBOXYLIC ACIDS ON TO a Fe2O3 FROM II-HEPTANE AND n-HEXADECANE heat of adsorption kJ mol-1 adsorbate from n-heptane from n-hexadecane acetic acid < 0.8 < 0.8 hexoic acid 18.2 9.6 hexadecanoic acid 50.2 77.0 octadecanoic acid 59.2 86.0 butyl alcohol 12.1 4.2 hexyl alcohol 21.0 11.3 hexadecyl alcohol 42.7 60.7 octadecyl alcohol 67.0 75.0 What remains undecided however is whether the high heat of adsorption attesting to vertical orientation of hexadecane and stearic acid is confined to a close-packed monolayer or several more loosely-packed layers extending into the bulk of the liquid.The fact that the amount of stearic acid adsorbed from n-C and n-C, on a-Fe203 is the same in spite of the difference in the heats of adsorption suggests that the orientation is not confined to the monolayer. A. J. Groszek Ann. 1970 ASLE Con$ (Chicago).GENERAL DISCUSSION 265 Dr. A. Cameron and Dr. A. J. Smith (Mech. Eng. Dept. Imperial College) said With regard to the paper by Dyson the gel-like layer formed on the surfaces of steel does not contribute to the EHL film when tested at the very high shear rates of the disc machine. There has been further work extending our results which shows that the gel is extremely fragile under shear and apparently breaks down very easily. The puzzle however still remains that 0.1 % of stearic acid in cetane has a profound influence on the lubrication characteristics of a 4-ball machine with 1 in. balls running at 200rev/min. The scuffig load is considerably extended and also the electrical resistance is likewise much increased. The two seem to go hand in hand. One is therefore left with the dilemma that the film thickness measurement described here shows that there is no effect while the scuffing tests and the electrical resistance tests show a marked effect.It would be interesting if Dyson could give any electrical resistance values for the types of system discussed by Askwith Cameron and Crouch. Dr. A. Dyson (Thornton Res. Centre Chester) said Cameron and Smith raise a point which is of great interest and importance but which has not yet been satisfactorily explained to my knowledge. There seems to be no necessary connection between film thickness and scuffing load. Film thicknesses calculated for the conditions obtaining just before scuffing vary over nearly an order of magnitude even if the metal surfaces and the lubricant are similar. Furthermore additives of the “ extreme pressure ” type do not seem to have any effect on film thickness under conditions in which it can be measured; yet they increase the scuffing load.Recent work on the effect of very small concentrations of oxygen and of water vapour indicates that these must be regarded as “ extreme pressure ” agents since scuffing loads are often very much less if oxygen and water vapour are carefully excluded from the rubbing system. We do not understand in detail the mechanism by which these materials act and I suggest that the effect of stearic acid may be essentially similar in nature. The apparent fragility of the films formed under the conditions reported by them is interesting but as I understand it the observation indicates only that there is at least one point of intimate electrical contact over the rather large area of interface between the mercury and the solid metal.A coherent film could still persist over the remainder of the area. Another possibility is that the films formed by reaction with a metal surface which has been freshly exposed by abrasion are stronger than those formed on a surface protected by physically adsorbed or chemisorbed layers such as are presumably present under the conditions of their work. I am not able to offer useful comments on electrical resistance under the conditions used by Askwith Cameron and Crouch. Dr. D. Tabor (University of Cambridge) said Dyson’s paper illustrates the gap between tiibologists even of the most sophisticated kind and surface chemists. The problem is not only one of concepts but of communication and I sympathize with the difficulties he must have experienced in coping with some of the preceding papers.There is one point in his paper that may have escaped the attention of the surface chemists. In his experiments the mean contact pressure between the discs is ca. 3,600 atm and this is exerted on the oil film trapped between the surfaces. As a result the viscosity of the trapped film is about 1,000 times the atmospheric viscosity. This suggests that other rheological properties of the film may be vastly different from those obtained in a normal laboratory. It is well known e.g. that small amounts of a fatty acid can greatly improve the lubrication properties of a mineral oil. I would welcome any comments from surface chemists about the possible 266 GENERAL DISCUSSION effects of very high pressures on the rheological properties of surface -active materials adsorbed on a solid surface from a non-polar solvent.Dr. K. J. MyseIs (R. J. Reynolds Winston Salem) said Dyson's experiments were not directed to the very thin monomolecular surface layers which are bound to play an important role in lubrication. Nevertheless the very high pressures prevailing under these lubricating conditions may have a considerable stabilizing effect upon these very thin layers. This is indicated by the experiments reported by Ash and Findenegg who found that adsorption was accompanied by contraction which could be interpreted as corresponding to the solidification of a monomolecular layer of liquid near the surface. It follows that such a layer will be stabilized by higher pressures just as the freezing point of hydrocarbons is raised by it.Dr. H. E. Ries (Chicago) said I have two questions and a comment related to the elegant studies of Roberts and Tabor (a) Have the authors considered controlled deposition (Langmuir-Blodgett) of the film-forming compounds on the rubber and glass surfaces? (b) In the earlier work with films of calcium stearate on mica at what surface pressures were the films transferred? Perhaps in both these cases the monolayers should be transferred at several different surface pressures because marked differences in film structure have been demonstrated through a relatively small pressure range. For example in our electron-microscope studies on transferred films islands irregular in size and shape appear at low pressures a continuous film with discontinuous open spaces at greater pressures (with the ratio of covered to uncovered area increasing with pressure) a homogeneous continuous film at still greater pressures then ridges and folds and finally long narrow flat platelets two molecules thick following collapse '* Dr.A. D. Roberts (University of Cambridge) said In reply to Ries an attempt was made to measure the shear strength of monolayers of sodium dodecyl sulphate and stearic acid sandwiched between rubber and glass surfaces. Monolayers were deposited on surfaces from bulk solution and excess solution " squeezed " out between the surfaces either under the influenced of a high load or by allowing the solvent around the contact region between surfaces to evaporate so that the excess solution in the sandwich would be " sucked out " by concentration difference.The separa- tion between the surfaces was measured interferometrically. In some cases no film appeared to be trapped between the surfaces in others the film was about 40 A thick. This was assumed to be bimolecular film and its shear strength determined by measuring both the force required to shear the film and the area of the contact zone. By using both soft and hard rubber surfaces it was possible to change the contact pressure over the range +-6 atm. Results of preliminary measurements suggest that the shear strength of monolayers of sodium dodecyl sulphate and of stearic acid increase with pressure over this range. Unfortunately the interfero- metric method does not indicate whether the trapped film is continuous or not.For this reason we have not quoted any specific values. However at the higher pressure range ow shear strengths were comparable with those observed by Bailey and Courtney-Pratt in their mica experiments. The next step of this work would be to deposit monolayers by controlled deposition using the Langmuir-Blodgett technique . H. E. Ries Jr. and W. A. Kimball Proc. 2nd Int. Congv. Surface Activity 1957 1,75 ; Nature 1958 181,901. H. E. Ries Jr. and D. C. Walker J . Colloid Sci. 1961,16 361. GENERAL DISCUSSION 267 Films of calcium stearate were transferred at a surface pressure of 16 dyn/cm castor oil being used as the piston oil. Experiments were also attempted using oleic acid as the piston oil which exerts a pressure of about 30 dynlcm but in this case the higher pressure tended to crumple the film.Dr. Th. F. Tadros (Plant Protection Ltd. Bracknell Berks) said With regard to the paper by Roberts and Tabor in the measurement of film thickness of sodium dodecyl sulphate between rubber and glass was the pH controlled? The electrical double-layer charge at the glass-solution interface is not only dependent on ionic strength but also on the pH of the solution. By addition of salt the pH would vary so that the two curves of their fig. 5 can only be compared if the pH was main- tained constant. Another point is with regard to the presence of specifically adsorbed Ca2+ ions (in the Stern plane) on glass which would cause adsorption of dodecyl sulphate ions with the head groups towards the surface and the hydrocarbon chain in solution. Mysels has commented that this seems unlikely since Ca2+ ions would be screened by negative silicate sites.However from the work on quartz flotation by sodium dodecyl sulphate,l in the presence of bivalent ions e.g. Ba2+ dodecyl sulphate ions are probably held next to barium activated quartz by association with specifically adsorbed ions in the Stern layer. This seems to occur at high Ba2+ concentration in solution. Dr. A. D. Roberts (University of Cambridge) said In reply to Tadros the salt added to films of sodium dodecyl sulphate (SDS) sandwiched between rubber and glass was sodium chloride. This is a strong electrolyte and therefore not expected to alter the solution pH significantly. Measurements made on solutions of 0.01 M SDS containing different quantities of NaCl ranging in concentration from 0.01 to 0.30 M revealed that the pH was constant at 6.0+0.1.The pH of 0.01 M SDS without added NaCl was 6.14k0.02. Dr. A. Cameron (Mech. Eng. Dept. Imperial College) said It is indeed gratifying to note that some work which I did in Prof. Sir Eric Rideal's laboratory in 1943 has been resurrected by Drauglis et al. The fact that in those days I did all the tedious summations using a slide rule naturally limited the accuracy of the results. There does seem to be some confusion in their paper in that the authors have not recognized that the reason for putting the energy E2 much smaller than El is that when the chains are in between the two equilibrium positions the repulsion energy must equal the attractive energy or be of the same magnitude. I therefore do not understand their approval of Akhmatov's criticism as Akhmatov also did not appreciate this.Could Drauglis make some comment on this point? It is not clear why they chose a square unit cell. Also have they any intention of extending their model to include a terminal CH3 group? They may be interested that in the Lubrication Laboratory at Imperial College this problem is being studied as completely as possible by myself and M. J. Sutcliffe. What is gratifying is that the first results show that the setting angle is 45" which agrees to considerable accuracy with the latest X-ray determinations. Dr. D. Tabor (Univwsity of Cambridge) said Drauglis has described a detailed analysis of the forces between long chain molecules. He has then deduced the frictional force between surfaces lubricated by fatty acid monolayers using the model proposed by Cameron some years ago.Such a calculation implies that Drauglis A. M. Gaudin and D. W. Fuerstenau A.I.M.E. Trans. 1955,202,66. 268 GENERAL DISCUSSION has in effect calculated the bulk shear strength of a bimolecular layer of fatty acid. I find this a little disconcerting. These computations give us a value of the theoretica2 strength of the bimolecular layer ; the real strength may be very different indeed if shear involves the movement of dislocations. Would Drauglis discuss this ? Dr. E. Drauglis (Battelle Mem. Znst. Ohio) said Although Akhmatov was apparently unaware of Cameron’s reason for neglecting E, (ref. (12) of our paper) we agree with his criticism because the assertion that when the chains are in between the two equilibrium positions the repulsive energy must equal the attractive energy or at least be of the same order of magnitude has not been proven.Undoubtedly adopting this assertion makes the calculations much easier but this is not sufficient reason for doing so. We find it difficult to believe that the terminal CH3 groups of each chain adjust their positions as the layers are squeezed together so as to make the assumption true for all values of the interlayer separation. The more elaborate calculations being carried out at his laboratory by Mr. M. J. Sutcliffe should do much to clarify this point. In our analysis we chose a square unit cell in order to simplify the calculations. This results in a negligible error. In the future we may make our model more elaborate by adding a terminal CH3 group to our polarizable rods but the rod model is sufficient for our present purposes.In reply to Tabor our calculations and those of Cameron give the contribution of the van der Waals forces to the friction between two ideal layers of stearic acid. For two real surfaces covered with a monolayer of stearic acid one would expect that the arrangement of the molecules would not be nearly so orderly as in our models. Many dislocations and other irregularities could be expected to be present. These would lower the strength of the interaction between the layers. Since our values for the coefficient of friction are already far below the experimental values the con- clusion is that the van der Waals forces are insufficient to explain monolayer lubrica- tion; other phenomena must be invoked. There is no question but that asperity interaction is very important in such a system.However it is not our purpose to develop new theories of monolayer lubrication. Our only purpose in introducing Cameron’s theory of monolayer lubrication was to serve as a check on our calculations based on the rigid rod model. Our primary purposes are to develop and verify our hypothesis that layered structures must exist in certain types of boundary surface films. A theory capable of quantitative correlation of the data of Fuks and develop- ment of a better understanding of the rheology of smectic liquid crystal-like structures are needed. Dr. G. Frens (PhiZips Res. Lab. Eindhoven) said Would not the study of the Brownian motion of a single particle in a very dilute suspension be a better criterion for the existence of thick structured layers of anomalous viscosity near a surface than those hydrodynamic experiments which have been under discussion. If such layers did indeed exist they would cause considerable deviations from Einstein’s law in that they would make q (or a) appear to be higher than the values which can be determined independently. One would have to do a careful experiment so that no electrical forces or interactions between particles would interfere. But this can be done and to my knowledge no proof for thick structured layers has been found in such experiments. D = kTJ6nqa = a2/2t

ISSN:0370-9302    

DOI:10.1039/SD9700100257

出版商:RSC    

年代:1970

数据来源: RSC