摘要:

GENERAL DISCUSSIONDr. S. Levine (Manchester University) and Prof. G. M. Bell (Chelsea College,University of London) said: A comparison of our solutions of the modified P.B.equation with those obtained in our ref. (2)-(5) is of considerable interest. Althoughthese relevant papers on diffuse layer theory by the methods of statistical mechanicsare important, they have limitations which require discussion. In their work on theeffect of ion size, Stillinger and Kirkwood ( S . and K., ref .(2)) used a moment expansionin powers of the charging parameter assigned to an ion and retaining only linear terms,derived a differential-difference equation for the mean potential at an ion situated inthe diffuse layer of a single plate. They found that when 7coa 21-03 (corresponding toa 1-1 electrolyte concentration 2 0-6 M) this potential becomes oscillatory, indicatingthat the Gouy-Chapman theory has become inadequate. Since our present theorydoes not apply when ~ g a 2 0.5, we cannot compare our results with those of S.and K.at these higher concentrations. However, their results imply our C(c0) = +(rcoa)2 atsmall rcoa. Recently (our ref. (16)), by treating the ions as point charges situated atthe centres of equal size hard spheres which are embedded in a uniform (aqueous)dielectric medium, we have reproduced the differential-difference equation of S . andK. from statistical mechanics by another, simpler method. We find that the cavityeffect, corresponding to our 4cav is their only correction to the P.B. equation.Imageand excluded volume effects and the self-atmosphere of an ion outside its exclusionvolume are all neglected in their new equation ; this also implies the omission of our$Lay. At first sight this seems surprising, particularly in regard to the volume effect,since one would expect the latter to be an essential part of the ion-size factor. K. andS . do not introduce the image terms and since the Debye-Hiickel self-atmosphereenergy of an ion is proportional to the square of its charge, their linearization withrespect to the charging parameter implies the neglect of this term. The reasons forthe omission of the volume effect are that they have restricted themselves to equalexclusion volume and to large distances x from the plate, where the potential $(x) issmall.(Consequently, the integrals Il(x1) and Iz(x1) in their eqn. (24), which containthe volume effect, are absent in their final equation.)The justification for such approximations is made clearer by a study of fig. l a and2a, where we plot the various contributions to C(x) for a single plate at 0.1 M electro-lyte concentration, corresponding to fig. 1 and 2 respectively, of our paper. We seefrom curve V in fig. l a that, with equal exclusion volumes, the contribution from thevolume effect to C(x) is practically zero for x> 15 8, ( I C O X ~ 1.6). Indeed, only curvesI and VII, due to 4Mv and $Lay, are significant at such large x, where the potential $(x)is small (520 mV). In fig. la, $cav and 4LaV contribute to C(co), 0.087 and -0.017(at 0-1 M and exclusion volume radius a = 4A).Although &" has a simplephysical interpretation (it occurs because the effective charge on an ion, whichgoverns its self-atmosphere, differs from its true charge by the diffuse layer charge inits exclusion volume), and like +cav, it does not drop out with linearization in thecharging parameter, K. and S . have not included it. This can be understood from ourderivation of their equation, in which it was necessary to assume that the positions oftwo neighbouring ions in the diffuse layer are completely uncorrelated when theirseparation exceeds a and $&#O depends on such a correlation. K. and S. make thesame assumption in their eqn. (30) but it is not obvious from their analysis that&, = 0, because one of their ions is discharged.In omitting they have some-what overestimated C(m). (Higher terms in C(m) have been neglected here.) If the992 GENERAL DISCUSSIONdifference part of K. and S.'s equation is expanded in powers of a then an iterationprocess yields as the first two terms C(o0) = ~(~0a)2+3'~(~oa)4) K. and S . considersmall plate potentials, when C(x) increases with decreasing x to a much less extent thanis shown in fig. la. For example, we have calculated C(o0) = 0.070 and C(x0) =0.093 at 0-1 M and xo = 3-7Atp) if $(x,)=33 mV. At small potentials, C(x) can be0 60 50 40.30 20. I0-0.1-0.2X<&FIG. l(a).-Contributions to correction term C(x) corresponding to curves A, A1 in fig. 1 of our paper(0.1 M).I, cavity effect (dCav) ; II, polarization energy and concentration dependence of E ; IU,self-atmosphere-image ; IV, dielectric saturation ; V, electrostriction ; VI, volume effect ; VI, VII,cavity effect (&w).I2 0x(fQFIG. 2(u).--Similar plots to fig. la corresponding to curves A, A1 in fig. 2 of our paper.replaced by C(c0) for all x very nearly and hence a good approximation to diffuselayer theory at small rcoa( 50.5) is obtained simply by replacing I C , ~ by I C , ~ [1+ C(w)] inthe Gouy-Chapman equation. For unequal exclusion volumes, which is perhapsphysically more realistic, curve V in fig. 2a shows that the volume effect does not tendto zero at large x. For the same a (which is now the distance of nearest approacGENERAL DISCUSSION 93between oppositely charged ions) as in fig.la, the volume effect contributes a termequal to -0.014 to C(c0). Our results suggest that oscillations in the potentialoccur at rcoa greater than 1.03, the value obtained by S. and K.Krylov and Levich (our ref. (4), (5)) have critically examined the method of K. andS . and have concluded that the range of validity is restricted to very small potentials ;but qbLv is also absent from their results. In his use of the Bogolyubov equation instatistical mechanics for single particle distribution functions, Martynov (our ref. (6))makes the same assumption as S. and K. in regard to the absence of correlation betweenions in the diffuse layer. At small potentials he obtains a third-order differential-difference equation, and not s.and K’.s second-order equation. He finds thatoscillations in the potential set in at lcoa = 1.6, but this seem incorrect.In ref. (l), our treatment of the modified P.B. equation was formulated in terms oflocal thermodynamic balance (1.t.b.). However, the Loeb-Williams’ (L. and W., ourref. (21), (22)) equation for the image-self-atmosphere potential of point ions, whichwas originally based on the statistical mechanical treatment of strong electrolytes byKirkwood,l and which we have applied to ions of finite size, goes beyond 1.t.b. Inref. (16), we obtained this equation from statistical mechanics in a different way, whichclearly shows its physical meaning. An integral equation of the Kirkwood type (oralternatively of the Born-Green type) for the pair distribution function of two ions inthe diffuse layer was used, and elimination of the three-ion distribution function(closure) was attained by the following additivity approximation.Consider a pointin the diffuse layer outside the exclusion volumes of a specified pair of fixed ions. Weassume that the change (perturbation) in the mean local density of any ion species atthis point, due to the presence of the two fixed ions, is the sum of the changes due toeach ion in the absence of the other. If the ions are regarded as embedded in auniform dielectric and the polarization energy is ignored, then our treatment of thediffuse layer is practically equivalent to that of Buff and Stillinger (B. and S., our ref.(3(b)), who applied cluster theory.The density of an ion species at any point in thediffuse layer is obtained by equating our expression (2.30) (at A = 1) for the chemicalpotential of ion type i to its corresponding form in the bulk electrolyte. The resultingformula for the density should be compared with that of B. and S., which is given bytheir eqn. (46) when ion-size effects are included. Their potential qcav(zl) equalsour sum t,b+q&, apart from a small correction due to the variation in diffuse layercharge density within the volume of an exclusion sphere. Their term B(z1) describes amore general form of the volume effect and is contained in an earlier paper by us(ref. (12), eqa(2.22)-(2.24)), where we explained how to obtain the simplified expressionused in our present paper, if only linear terms in ion concentration are retained.Except for an unscreened image term (dl)(z~)), the last (integral) term in eqn.(46) ofB. and S. accounts for the image-self-atmosphere effect, including that due to +&.The function ~ ( 2 , l ) in their integrand may be identified with (&/e&bi where 44 is the L.and W. “ perturbation ” potential (our ref (1)) outside the exclusion sphere of ion i.The B. and S . eqn. (24) for ~ ( 2 , l ) is precisely the linearized L. and W. differentialequation for #t, for point ions. A final comment is required on the statement by B.and S. that ion-size effects invalidate the 1.t.b. method. In our opinion, the use of1.t.b. implies the neglect of local density gradients and image potentials, regardless ofwhether or not ions are assumed to have size.Finally, by using the Bogolyubov(Born-Green) equation for the pair-distribution function, Martynov has also obtainedthe linearized L. and W. equation for point ions (eqn. (29) in his part 11, p. 89, ourref. (6)). His approximations are equivalent to our additivity assumption, coupledwith linearization in the self-atmosphere potential.1 Kirkwood, J. Clzem. Physics, 1934, 2, 76794 GENERAL DISCUSSIONDr. M. J. Sparnaay (Philips Research Lab., Eindhoven) said: I am indebted toLevine and Bell for their addendum concerning the paper of Hiickel and Krafft 1 andmy paper.2 However, the paper by Levine and Bell, and their addendum, must leavethe reader with the question why the present author neglected the self-atmosphere andimage effects and the effect arising from the cavity potential.The answer is, that theseeffects were not neglected and where omitted, reasons were given.Results of a study by Williams3 of the self-atmosphere and image effect werecommunicated. Williams found, on the basis of Kirkwood’s analysis,4 that in generalthe potential was about 5 % closer to zero than calculated on the basis of the un-corrected Poisson-Boltzmann equation. The self-atmosphere (and image) effectswere considered as non-specific (i.e., not depending on the nature of the ions involved)as a result of a consideration of the cavity potential. The cavity potential wasintroduced by Huckel and Krafft in their theory of ionic solutions.This theory wasbased on Kirkwood’s analysis and represented an improvement over the originalDebye-Hiickel theory. The Hiickel-=aft correction terms could be derived by meansof the introduction of an effective ionic charge. In a 1-1 electrolyte the ionic chargesel( = e) and e2( = - e) were replaced by (see eqn. (7) of ref. (2)) :el(eff) = el + J pldVl; e2(eff) = e2 + J p2dV2,where p1 and p2 are formal charge densities inside the excluded volumes V1 and V2 oftwo hydrated ions. Now for p1 and p2, the same (small-potential) approach wasfollowed as in the Debye-Huckel theory. This means that the charge densities insidethe excluded volumes were considered proportional to the potential there. In doingso, the Huckel-Krafft correction terms to the Debye-Huckel theory were immediatelyobtained.For p1 the following expression was obtained :p1 = - q . ~ [ _ e r ~ _ l ]kT e ( l + ~ Z ) ~r ’and similarly for p2. Here E is the dielectric constant of the medium, ic-1 is the Debye-Miickel length, n is the number per cm3 of anions or cations, d is an average radius ofthe (spherical) excluded volume, and finally r is a distance, rG3.The bracketed part is the cavity potential. It consists of two parts with oppositesign. The Coulombic part points to the fact that, as in the D-H theory, an ion isconsidered as a centre of an ionic atmosphere and therefore, that it is essentially theactivity coefficient which is affected by the cavity potential. In the Gouy theory thecentral ion is replaced by a wall and essentially equipotential planes parallel to thewall are considered.The consideration of central ions in the Gouy theory leads tocorrections which are now known as self-atmosphere corrections. The reasoningin my 1958 paper 5 was that the Huckel-Krafft terms represented a correction to theself-atmosphere correction. Since in the 1958 paper corrections on corrections weresystematically neglected, the cavity potential was purposely left out. Thus, only anon-specific self-atmosphere and image correction remained. The largest spec@effect appeared to be the ion-size effect, this effect being considered as a “ van derWaals-b ” correction. It amounted to about 2 % for a 0.01 m solution and its effectwas opposite to the self-atmosphere effect which amounted to about 5 %.Therefore1 Hiickel and Krafft, 2. physik. Chem. 1955,3, 135.2 Sparnaay, 2. physik. Chem., 1957,10, 156.3 Williams, Proc. Roy. Soc., A , 1953, 66, 372.4 Kirkwood, J. Chem. Physics, 1934, 2, 767.5 Sparnaay, Rec. trav. chim., 1958,77, 872GENBRAL DISCUSSION 95the self-atmosphere effect often dominates and the net result of the corrections oftenis a slight compression of the Gouy layer.Thus, there is qualitative agreement with the results of Levine and Bell. However,it was mainly the specQicity of the ions that was of interest to the present author as isborne out by the subtitle of the 1962 1 paper : “ The specific influence of the ions ”.The major difference (the word “ major ” may seem somewhat misleading) betweenthe results of the present author in 1958 and 1962, and the results of Levine and Belltoday is, that the latter authors find much more specificity.This is already suggestedby a comparison of their fig. 1 with their fig. 2. Now the treatment of Levine and Bellrepresents a more advanced stage of the theory. Precisely this advanced stage callsfor a simple physical “ picture ” of the calculated specificity and I hope that this canbe given here by Levine and Bell.Finally the following point is worth mentioning. Attempts to improve the Gouytheory can be done in (at least) two ways : (a) a certain model of the ionic solution canbe selected and the statistical theory based on this model can be worked out in detail,(b) the model itself can be critically investigated.Levine and Bell have followed the first way.Their treatment is based on the“ hard-sphere ” model (the hydrated ions are considered as hard spheres in a structure-less medium). They found that even for 0.1 m solutions the original Gouy theory wasreliable. This was due to cancellation of the various corrections. However, thisresult is little more than an arithmetical statement, because the selected hard-spheremodel will certainly not represent the physical situation. Especially close to the wallthe average distance between the ions is very small in a 0.1 m solution, the distancebeing about 5-8 times the diameter of a water molecule (depending on the wallpotential). Under these circumstances the solvent cannot be considered as astructureless medium (see also other papers of this Discussion).Therefore the secondway seems to be more urgent both in the Gouy theory and in the study of concentratedionic aqueous solutions.Dr. S . Levine (Munchester University) and Prof. G. M. BeII (Chelsea College,University of London) (communicated) 2 In reply to Sparnaay, our results are not solelybased on a hard-sphere model embedded in a structureless solvent. Structure of themedium has been introduced, although perhaps crudely, through our use of the Flory-Huggins statistics and our correction for the compressibility of the water moleculesand ions. Alternatively, we have expressed the exclusion volume effect in terms of“ imperfect gas Mayer integrals ” which apply generally to short-range interactionsbetween the ions.In particular, we assume an oppositely-charged ion pair to have asmaller distance of nearest approach than a like-charged pair. Much work in strongelectrolyte theory has been based on the simplest finite ion model, a ‘‘ hard-sphereplus Coulomb potential”. This takes the sphere’s dielectric constant as equal tothat of the solvent and disregards the region of lower polarizability due to the “ bare ”ions’ interaction with neighbouring water molecules. However, Levine et aZ.2 (cf.Bellemans and Stecki 3 and Marcus 4) have developed a theory of the short-rangeforces between a pair of hydrated ions due their polarizability and have shown that, tofirst order, these are inversely proportional to the fifth power of the separation.The1 Sparnaay, Rec. tmu. chim., 1962, 81, 395.2 Levine and Wrigley, Disc. Faraday SOC., 1957, 24, 43, 73. Levine and Bell, Trieste S y i p . onElectrolytes, ed. Pesce (Pergamon Press, 1962), p. 77. Levine and Rosenthal, Chenzicnl Physicsoflonic Solutions, ed. Conway and Barradas, (Wiley and Sons, 1966), p. 409.3 Stecki, Adu. Chemical Physics, 1964, 6,413.4 Marcus, J. Cheni. Physics, 1965, 43, 5896 GENERAL DISCUSSIONproblem of the potential energy of a pair of hydrated ions (which includes the potentialof the mean force between the pair) is naturally worse near an interface, but in ref. (1)and in our paper we have endeavoured to allow approximately for the effect of ionicpolarizability on the self-atmosphere-image potential. Structure of water surroundingan ion is implied in this effect, which is also responsible for the polarization energy ofan ion and the dependence of dielectric constant on ion concentration.A complica-tion in the theory is that the hydration shell radius, inside which most modification ofpolarizability occurs, is usually smaller than the exclusion volume radius, whereaswe have assumed these radii to be equal. (This does not matter in the Debye-Hiickeltheory of electrolytes because of spherical symmetry of the ionic atmosphere.)Admittedly, the present results are in their early stages ; no account is taken of sucheffects as hydrogen bonding and flickering water clusters, and we still lack a satisfactorytheory of the dielectric and electrostrictive properties of the diffuse layer.Perhaps as equally significant as a proper ion model are the neglected higher termsin the expansions of the various corrections in powers of the ion concentration.Wehave made a preliminary investigation of such terms in the volume effect (our ref. (1 2))and have mentioned their role in the cavity effect in this Discussion. Another factorto be considered (ref. (12)) is that when the distance of an ion from the wall is less thanthree times the effective radius of its hydration sphere, then even the first-order terms(linear in ion concentration) in both volume and cavity corrections are altered becausea second ion is prevented from occupying positions between the first ion and wall.We were careful to state that our finding that the Gouy theory is a reasonableapproximation at 0.1 M of a 1-1 electrolyte, is not conclusive.An estimate of therange of validity of our calculations requires more information about the variousneglected factors discussed above. We suggest that if our value for C(x0)50-3 at thewall (corresponding to an increase in the Gouy wall-charge density by a factor5 J 1.3 and a wall potential ,< 70 mV for our example of unequal exclusion volumes)then the present calculations give reliable " first-order " corrections to the P.B.equation. Sparnaay's remarks about the cavity effect, which is determined by thedistance of nearest approach between ions and is therefore specific, are misleading.It is not a second-order correction, being indeed the only one considered by the authorsin our ref.(2), (4)-(6), when they predict that the potential-distance plot becomesoscillatory above 0.6 M of a 1-1 electrolyte. Also, in 0.01 M our self-atmosphere-image (s.a.i.) effects exceeds the volume effect for equal exclusion volumes, but forthe unequal exclusion volume case described by our curve C in fig. 2, these two termsare roughly equal in magnitude (-0-2) at the wall and the s.a.i. term dies off morerapidly with distance from the wall. We consider that in our ref. (I), in our paper andin the discussions, we have explained the physical nature of the cavity potential andits relation to the other corrections which, apart from the small compressibility effect,have already been investigated by previous authors.Prof. J.Lyklema (Agric. Univ., Wageningen) said: Levine has shown that in adiffuse double layer at fixed potential the charge can deviate by about 20 % from thecharge predicted by the simple Poisson-Boltzmann distribution. This conclusion isnot amenable to experimental testing. However, it would be possible to studyexperimentally at fixed potential the difference in charge for salts of different nature,e.g., LiN03 and RbN03. The AgI-system at elevated temperature could be used forthat purpose because specific adsorption is absent. Could Levine point out whether,according to his calculations, the difference between Li+ and Rb+ as the counter ionwould be big enough to be detected experimentally? The standard deviation in 00at 65" and above is about 7 %GENERAL DISCUSSION 97Dr.S. Levine (Manchester University) and Prof. G. M. Bell (Chelsea Coll. of Sci. andTech.) said: The difference between the electrolytes LiNOJ and RbNO3 in the modifiedP.B. equation for the diffuse layer would mainly occur in the values of the Debye-Hiickel diameter a and of the parameter 6, which measures the dependence of dielectricconstant on ion concentration. Because the Li+ ion is more strongly hydrated thanRb+, we must expect a and 16 I to be larger for LiN03 than for RbN03. Thisimplies that our correction function C(x) will be correspondingly larger and hence atfixed plate potential $0, the surface charge density I QO 1 will be greater for Li+. Atsmall plate potentials (-25 mV) only the change in a need be considered, because agood approximation to the modified P.B.equation is obtained by replacing C(x) byC(m) ; we merely obtain the Gouy-Chapman solution with rct replaced by rc; [1+C(co)]. If C~(GO) and Cz(co) correspond to the two diameters a1 and a2 describingthe two electrolytes, then the ratio of the two values of 00 becomes [(l + Cl(co))/(l+CZ(CO)}]*. If we put KOal = 0.5 and 1coa2 = 0.4 and assume equal exclusion volumesfor the different ion-pairs in each electrolyte, then the ratio of the charge densities is1.015. At larger potentials, C(x)> C(m) for small x and it is possible that a largerdifference will be found between the two electrolytes; however, no numerical com-parisons between different a and 6 have been made yet.It appears, therefore, that ifthe standard deviation in the experimental value of QO is 7 %, then the differencebetween the two electrolytes LiN03 and RbN03 will be difficult to detect in the absenceof specific adsorption, which is in agreement with Lyklema’s observations.Dr. S . Levine (Manchester University) and Dr. A. k. Smith (College of Technology,Liverpool) said : Although fig. 4 in Lyklema’s paper shows no maximum in chargeexcess of co-ions O-, the discreteness-of-charge theory, to which he refers, suggeststhat, provided K+ only is absorbed, a maximum may well occur at high potentials.This is supported by measurements by Smith and Fairhurst of the change in electro-phoretic mobility with PI of aqueous AgI in the presence of 0.1 M KN03.Amaximum in mobility in the region PI M 5 ($0 w - 300 mv) was found and similar resultshave been reported by others. A maximum in this region is also consistent with theflocculation results of Fairhurst, and Parfitt and Smith (F.P.S .) discussed below. Wewish to show how the constant Kin Lyklema’s eqn. (1) and consequently the potentialat the outer Helmholtz plane (O.H.P.) $a and the specific adsorption potential q5 of K+can be estimated from the position of such a maximum in Q-. From the electrostaticrelations for the Stern-Grahame inner region,Here Ka is the integral capacity and d the thickness of the inner region, y the distancebetween the inner Helmholtz plane (I.H.P.) and O.H.P., aP the charge density ofadsorbed ions on the I.H.P. (identical with Lyklema’s when K+ only is adsorbed) andYO the (outer) potential at the AgI surface.The potential $0 used by Lyklema isrelated to YO bywhere x, chosen so that $0 = 0 at the z.P.c., is chiefly due to the potential drop in theAgI solid phase and to adsorbed biphthalate ions at the Z.P.C. It follows that$0 = ~ o - x , (2)The decrease in Kz: with increase in I$o I is described by a variation coefficientt = (kT/e) . (d In KZ:ld$o) where t-0-02-0-04 in 1-1 electrolytes. The value of Kiat the maximum in c- is given approximately by 198 GENERAL DISCUSSIONwhere p = d-y, Ns is the density of Langmuir-type adsorption sites on the AgIsurface, K the Debye-Huckel parameter and g is a factor of the order of 1 in theself-atmosphere (discrete-ion) potential at an adsorbed (K+) ion, defined by 4D = - (pyop/d2 &)g, which characterizes the discreteness-of-charge effect.Solving for Kiand substituting (2) into (I), an alternative expression for x isThe biphthalate ion is assumed completely desorbed at the maximum in a-. Thevalue of Ki at the Z.P.C. in (3) is easily obtained for given t .The above steps presuppose that $a and up are known. These can be calculatedfrom Lyklema’s data by postulating a maximum in CT- at high I $0 I provided that Kis known, and our method is to choose K such that (3) and (4) give identical values forx. At the z.P.c., Lyklema’s eqn. (1) becomes a* = & K and since he assumes K = 0his values a+ and a- should be replaced by a$ and a!- respectively where a i = a+ & K.His equation (2) now yields $a from a’- making it possible to calculate ad.t, the excessof cations in the diffuse layer, fromad+ = J(~nkT/2n)[exp (- e$,/2kT) - 11.In this way as (= a!+ -ad+) can be determined. Because the experimental (ao,$o) plotis convex to the $0 axis in 0.1 M of KNO3, in addition to the identity of (3) and (5),another condition is C<Kt where C = (dao/d$o> and Kt = ao/t,bo.The discreteness-of-charge theory predicts a maximum in the potential at the I.H.P. (I $p I), where 1(brought to our attention by Prof. Lyklenia) ; K/ is given by (4) provided the last termin the square brackets is omitted. In the calculations we examined the possibility of amaximum in aL in the range - 250 to - 300 mV for $0, making use of Lyklema’s curvefor a+ in 0.1 M KNO3 in his fig.4. As an example, if a maximum is assumed ata’- = 0.9 pcoul./cm2 and $0 = - 250 mV, the conditions outlined above were satisfiedin a narrow range of parameters around y/d = 1/5, g = 5/4 and t = 0.04, leading toa common x = - 2 mV at Kw0.06 pcoul./cm2, where - 31 mV and Ki E 16pf/cm2. At the maximum in I $p 1, Cw 14 pf/cmZ and &Z 15 ,uf/cm2 ; also from theStern adsorption isotherm, modified to include the discreteness-of-charge potential4p, g5 x -0-5kT. The small values of ~ / d and 4 are consistent with Lyklema’s sugges-tion that the K+ ion is not very close to the surface.I at $ 0 ~ - 250 mV is also consistent withthe results of F.P.S. referred to above. These workers measured d In W/d In c (= m),where Wis the retardation factor for flocculation and c the concentration of flocculat-ing electrolyte (KN03) for negative AgI sols over a range of PI values.(dm/dpI),was found to be around zero in the range PI = 7 to PI = 5.5 ($0 = -200 to - 300mV), being positive at higher pl values and becoming negative for pI55. In therelation (dm/dpI), = (drn/dt,b&(d$d/dpI), the D.L.V.O. theory of colloid stabilityrequires that (dm/d$&>O. Thus the observed behaviour of (drn/dpI), is in semi-quantitative agreement with the discreteness-of-charge theory prediction of a maximumin I $6 I at $ox -250 mV ( ~ 1 ~ 6 . 3 ) . At lower PI values (d$d/dpI), will be negative,leading to negative values of (drn/dpI),. This support from flocculation experimentsfor the discreteness-of-charge theory is discussed in more detail elsewhere.1, 2Prof.J. Lyklema (Wageningen) said: In reply to Levine and Smith, the experi-mentally observed fact that a maximum occurs in the c(t,bo) relation is not necessarilyLevine, Smith, Mingins and Bell, 4th Znt. Congr. Surface-Active Substances (Brussels, 1964),B/II/3.x = C-$o+$a+(aolKi>+(1/ap/dKi)J,,, i n a - . ( 5 )C = (d/p)Ki + (e/kT)tao (6)A maximum in a- and therefore in I2 Levine, Mingins and Bell, J, Electroanal. Cheni., 1967, 13, 280GENERAL DISCUSSION 99in conflict with the constancy of o-, as calculated from the ionic components of chargetheory or measured directly.1 The reason is that [ is closely related to $6, hence amaximum in $6, as predicted by the discreteness of charge theory is reflected in amaximum of [.However, CT- is related to $& thus :For sufficiently high negative values of $6, the exponential term becomes negligible,making c- constant and independent of $8.Even if there would be some variation in$8, 6- would be too insensitive a parameter to detect the variation. Hence it is possiblethat a maximum Of @d is sensed by [ but not by 6-.In their calculation of 4 and K, Levine and Smith use eqn. (1) to calculate $6.From !+h thus obtained, ad+ is found. Because of the insensitivity of 6- to $6 and thefact that cr- cannot be determined accurately this procedure is less accurate than theindependent method to determine $a described in the paper. Nevertheless, thereported value for $6 in 10-1 M KNO3 agrees reasonably well with our calculation.There is a difference of 2kT between our overall specific adsorption potential q5 andLevine and Smith’s specific adsorption potential, modified to include the self-atmos -phere potential.The calculated low values for K seems of the right order of magni-tude, considering the experimental fact that simple electrolytes produce only aninsignificant shift in the zero point of charge.Dr. L. Romo (Univ. of Quito) said: Granting that there is an abrupt change at50°C in the structure of the double layer as shown by the stability measurements andthe phase transition in Lyklema’s paper, I wish to know whether the stability con-vergence shown in fig. 8 is not due primarily to the state of hydration of Li+ and Rbfions at temperatures 60°C.I would suggest that this problem can be resolved byperforming n.m.r. measurements of the hydrated phases adsorbed on to the AgImicelles.Prof. J. Lyklema (Wageningen) said: In reply to Romo, the free enthalpy AGh ofhydration for alkali-ions is about 100 kcal/g ion. It is unlikely that the relatively smallincrease in thermal energy, realized in the experiments (about 20 % in kT) is sufficientto dehydrate alkali-ions more or less abruptly and completely around 50°C. Thephase transition as detected by DTA experiments was found in the absence of salt,hence it is evidently a property of the AgI/solution interface and not of the bulksolution. Nevertheless, it would be worth while to check this by independentmeasurements.Besides n.m.r., double layer studies on other systems (with specialreference to the specificity as a function of 2”) could be considered.Mr. G. Frens (van’t HoflLab., Utrecht) said: Our recent work on AgI suggestssome comments on Lyklema’s paper, and especially about his fig. 2. We measuredthe horizontal distance between adsorption isotherms like those in fig. 2 by a directelectrical method. The point of intersection of the isotherms was determined, sincethis distance vanishes at a certain pAg. We were puzzled by the observation that thedistance between the isotherms was very small at the positive side of the potential ofzero charge, but Lyklema’s fig. 2 explains this, since it shows a shift in the potential ofzero charge with the concentration of inert electrolyte. My questions concern hisinterpretation of these observed facts.His fig. 2 shows that the intersection of the1 van den Hul and Lyklema, to be published100 GENERAL DISCUSSIONadsorption isotherms with the axis CT = 0 occurs at a differentll/o for different concentra-tions of inert electrolyte. The intersection of an isotherm with the axis CT = 0represents the point where there is as much Ag+ as I- adsorbed in the AgI-surface.The ll/o scale of fig. 2 is a pAg scale so that different points on it mean different con-centrations of Ag+ in the equilibrium solution. Thus, for different concentrations ofelectrolyte a different concentration of potential determining ions is needed to get thesame situation in the solid AgI.Since there is equilibrium at the AgI-solution interface and since the situationinside the solid AgI is the same in all cases, this is tantamount to the statement that,although the concentration of Ag+ in the solution changes, its electrochemicalpotential is not altered.In Lyklema's paper an explanation is suggested in terms of specific adsorption.I would ask him what he means by this term, and why such adsorption could explainthe observed facts.Furthermore, I would ask whether there is any direct experi-mental evidence that adsorption of anions on AgI at the potential of zero chargeexists.Prof. J. Lyklema (Agric. Univ., Wageningen) said: In my paper the concept of" specific adsorption " was treated in analogy with the double layer on mercury.Itmeans that ions adsorb by other than Coulombic forces. One characteristic is thatspecific adsorption of anions can occur on negatively charged AgI. On originallyuncharged AgI-particles (a0 = 0 or r A g + = rl-) a negative adsorbed layer of NOTions can develop. This adsorption offsets the I- and Agt--adsorption balance(rAg+#rI-). The original situation (rAg+ = rI-) can be restored by changing theconcentration of the potential-determining ions in the solution. For specific adsorp-tion of anions a higher bulk concentration of I- ions is required to re-attain the Z.P.C.Indifferent electrolytes can affect the Ag+ - I- adsorption balance (i.e., the electro-chemical potentials of potential determining ions in the solution and at the interface)in two ways, viz., by way of (i) the activity coefficient in solution or (ii) the electro-static term at the interface. As there is some experimental indication that the F-ion which does not adsorb specifically (see ref.(15)), has little or no influence on theZ.P.C. the second effect seems dominant. As far as the author knows, no directmeasurements of ion adsorption around the z.p.c have been made on Agll. Suchmeasurements would be useful, not only in connection with the problem underdiscussion but also in view of the ionic components of charge theory.Dr. K. A. Kini (Warren Spring Lab., Stevenage) said: Eyklema has expressed somedoubt about the application of the B.E.T. method (presumably using nitrogen at- 196°C) for the measurement of the surface area of silver iodide and has determinedvalues for the same by the negative adsorption and double layer capacitance methods.Using B.E.T.method for the adsorption of argon or nitrogen at liquid nitrogentemperature and the adsorption of xenon at 0°C (or CO2 at 25"C),1 we obtained theresults :surface area, m2/gmaterial N2 or Ar at - 196°C Xe at 0°C orC02 at 25°Ccoals and carbonized coals <1 50-300Linde Molecular Sieve 4A < 10 > 300The reason for these low areas by the nitrogen or argon method is that at the lowtemperature employed the adsorbent material may undergo contraction or the rateof diffusion of the gas molecules into the pores of the adsorbent may be slow.1 Kini, Fuel, 1964, 43, 173.2 Kini and Walker, J.SCY'. lristr., 1965, 42, 821GENERAL DISCUSSION 101Prof. J. Lyklema (Wageninen) said: Experiments performed in our laboratory byvan den Hul showed that the specific surface area of suspended AgI particles asmeasured by negative adsorption or double layer capacitance was about a factor 3higher than the B.E.T. surface area. The cumulative evidence shows that for doublelayer and adsorption from solution work the former area is the correct one, i.e., theB.E.T. method is open to doubt. The explanation for this discrepancy is not clear,but the suggestion of Kini that at low temperatures the sorbent could contract does notapply to AgI because the B.E.T.-area was measured with different gases at differenttemperatures up to room temperature.These results were self-consistent.Prof. E. Matijevic (Potsdam, New York) said: In the work by Lyklema one has toconsider the adsorption of the phthalate ion on the silver iodide sol. This ion ispresent in relatively high concentration (10-3 M) and will adsorb at z.P.c., andbelow and above z.P.c., although the adsorbed amount may vary. The adsorptionof counterions in the double layer will therefore be affected by the presence of thephthalate ion and the results cannot be fully understood until the adsorption ofphthalate is directly measured.Prof. J. Eykkma ( Wageningen) said: In reply to Matijevit, phthalate ions doadsorb around the Z.P.C. in 10-3 M solutions. However, at surface potentials morenegative than - 100 mV the phthalate is fully desorbed.This is illustrated by fig. 1in which ao(+o) curves at 25°C are plotted with and without phthalate. There is adefinite downward deflection around the Z.P.C. plus a shift in the Z.P.C. to the right, asI / -3 -21 /I I - 200 -300 mV 1 t+' $0FIG, l.--oo(+o)-curve for AgI at 25°C : (1) 10-3 M KNO3 ; (2) 10-3 M K biphthalate. The curvesmerge for sufficiently negative $0.observed in the paper. These effects are apparently due to specific adsorption ofphthalate ion. At sufficiently negative potential all phthalate is desorbed, the curvebecomes indistinguishable from the one in 10-3 M KNO3, pH = 6-7. This showsalso that there is no measurable effect of H+-ions at pH 2 4 . The 10-1 M curves areindistinguishable over the negative branch. This fact made us surmise that now theNO; ion is predominantly adsorbed, although it causes no drastic shift in the Z.P.C.The DTA experiments have been performed with AgI samples in absence of phthal-ate buffer.Hence the observed phase transition is a property of the AgI+watersystem and not a spurious desorption effect of phthalate ions.Prof. M. Mirnik (Zagreb, Yugoslavia) said: The basic electric parameter of theoriginal and extended versions of the Derjaguin-Landau-Venvey-Overbeek theory isthe surface potential $0, which is supposed to represent the total potential dro102 GENERAL DISCUSSION#solid - 4liquid across the double layer, and is also assumed in the two phase systems ofthe AgI type to be proportional to PI. In his paper, Lyklema uses the same surfacepotential as the basic electric parameter.However, in two-phase systems (i.e.,systems which are not, as are galvanic cells, composed of a minimum of three phases,two metallic electrodes and an electrolyte) this potential physically is not significantand it cannot serve as the basic electric parameter of the theories of colloid stabilityand coagulation. Colloid systems are composed of an electrolyte and a single solidphase which can be an insulator, an ionic solid or exceptionally a metal.The 5 potential is constant 1 with PI in contrast to the surface potential$*, measuredin a galvanic cell, which is proportional to PI. Hence instead of the surface potential,another electric parameter of double layer theories should be found. Otherwise thereis no explanation for the proportionality between $0 and PI.In order to adjust thetheory to this constancy with PI Overbeek 2 and Lyklema 3 introduced the hypotheticalviscoelectric effect which was not observed in the AgI system to cause the variation of[I+] in this way.To bring the potential of two electrodes of a reversible cell to the reversible valuethe formation of a positive and a negative charge which is transported across theelectrolyte by transference of ions is necessary.4 Since in the colloidal system thesecond electrode is not present, the charge cannot be formed by simultaneous oxida-tion and reduction and it cannot be transported across the electrolyte.The surface potential is 5 the electric tension U of a cell measured as the electro-motive force E defined thermodynamically byThe subscripts 0, R, ind, and ref designate the oxidized and reduced states of thepotential determining species of the indicator and the reference electrode respectively,n and n’ are the corresponding oxidation numbers.The remaining symbols have theirconventional significance.According to this equation U can only have a defined value and be physicallysignificant in a system when all species whose chemical potentials occur in the equationare present in the system. In the system, ionic solid-electrolyte solution, the referencemetallic electrode is absent and consequently the value of the right-hand side is notdefined. For the same reason the chemical potential of the metal @R)ind and of thenonmetal of the indicator electrode, i.e.the colloidal particle, are not definedeither.Fig. 1 serves as an example of several additional arguments. In all three schemes(a), (b) and (c) of fig. 1 the same electromotive force Erev = - Urev will be measuredon the smaller condenser irrespectively of which electrode is connected to thesmaller condenser (cases (a) and (b)) on which practically the total potential dropoccurs. The electromotive force will be the same irrespectively how high or of whichsign is the polarization potential in case c. Also, the variation of the activity of thepotential determining species of, e.g., the Cu electrode in scheme (a) will cause avariation of the potential on the Zn electrode condenser so that the potential drop ofthe Cu condenser is practically zero.With equal capacitances both electrodes wouldparticipate in the variation of the potential drop, whereas only in case (b) the totalpotential variation would be on the smaller Cu capacitance.1 Mirnik, Disc. Faraday SOC., 1954,18, 207.2 Overbeek, Disc. Faraday Soc., 1954, 18,210.3 Lyklema and Overbeek, J. Colloid Sci., 1961,16, 501.4 Mirnik, Kolloid Z. 2. Polymere, 1962,185,66.5 Mirnik, in preparationGENERAL DISCUSSION 103It is not, therefore, justified to assume that the potential drop &olid - 4liquid actingon a single electrode is a constant, characteristic of the metal and concentration of itsions. If such a potential drop existed it would be different : (i) for each differentsecond electrode, (ii) for different polarization and compensation potentials, and (iii)for different electrode capacitances (curvature, shape, size). Consequently forcolloidal particles the capacitance and hence the potential would depend on theirsize.In polydisperse sols each particle would have a different potential and there is noexplanation for its proportionality with PI. Still less could such a potential becharacteristic for solid-liquid system where the solid is an insulator, a semiconductoror an ionic solid and not a metal, yet virtually identical colloidal properties werera. Urn. C. UrrvFIG. 1 .-A reversible Zn/Zn2+ - CuZ+/Cu couple, liquid junction eliminated, when the capacitancesconnected to electrodes (a) CZ, < Ccu, (b) CZ, > Cc, and (c) when a polarization potential is imposedupon one of the capacitances. In all cases U = U,,, measured by an electrometer or the like, withthe electrometer connected to the smaller capacitanw in a and b or to the electrodes in c.observed with all these categories of solids.Hence, the existence of any electricpotential which is proportional to PI or pAg is impossible. The potential difference4solid-(Pliquid of an electrode in a cell, as well as in a simple solid-liquid two-phasesystem, is not measurable and cannot be defined; it therefore cannot be physicallysignificant. This was also Guggenheim's opinion which has been accepted byGrahame. 1Lyklema also assumes that during the addition of Ag+ and I- ions to sols andsuspensions when aged, i.e during the variation of his $0 potential, the specific surfaceremains constant. This assumption is wrong because (i) the adsorbed I- ions andAg+ ions are neutralized and transformed into fresh AgI to form a surface of differentproperties compared with those of the original aged surface.(ii) The I- and Ag+ions at PI 3-4 or pAg 3-4 are also neutralized causing the formation of new fresh AgI.Its total surface may be of the order of or even greater (particle size 0.6 to 10 mp) thanthe surface of the originally present aged AgI (particle size 100-200 mp or greater).(iii) The addition of the Ag+ ions to stable sols causes coagulation when 6.2 > pAg > 4.5($0 > 0). The particles of fresh sols increase gradually from a size smaller than 10 mpat PI < 5 to an average size of about 150-200 mp when they finally coagulate.Theseparticle sizes were estimated from the measured intensities of scattered light attwo wave lengths.2 The existence of small particles was proved by Ottewill et aZ.31 Grahame, Chem. Rev., 1947,41,441, p. 449.2 Mirnik and TeZak, Trans. F'raday Soc., 1954, 50, 65.3 Horne, Matijevid, Ottewill and Weymouth, KoZZOZ~Z, 1958, 161, 50104 GENERAL DISCUSSION(iv) The addition of the I- ions to fresh and aged sols coagulated with Ag+ ions(6.2 > pAg > 4.5) during the variation of pAg causes repeptization immediately afterthe addition. The repeptization is stronger the higher the I- ion concentration(1-2<pI<9.8), the longer the contact and the greater the degree of agitation ; andcan be observed by the increase of the turbidity if the sol is shaken.All Lyklema’scalculations are therefore based on the erroneous assumption of a constant specificsurface of his AgI preparations.Since in the range of saturation 2 <PI < 7 we obtained a constant amount adsorbed,land under analogous conditions, van del Hula obtained a constant surface fromnegative adsorption, the adsorbed amount of I- ions-whether measured in g equiv.per mole AgI or multiplied by a constant value for the specific surface and expressedin, e.g., pcoulombs per cm2-is in this range of saturation proportional to the specificsurface. Any variation as given, e.g. in fig. 2 and 3, in the adsorbed amount is causedprimarily by the variation of the specific surface.A direct support to this conclusion follows from the comparison of differentmethods for specific area determination performed by van del Hul2; he states“ the surface area should be determined by a method related to the particular applica-tion at hand ”.In our opinion it should be determined independently for eachcomposition of the system (each PI value) when the equilibrium is established via thecondensation method of statu nascendi sols and via the peptization of aged prepara-tions. In systems with varying PI or concentration of counter ions CM, the specificsurface cannot be assumed constant.The conditions under which the specific surface was calculated to be constant withPI from negative adsorption, and the conditions for which we measured the constantadsorbed amount of I- ions, were obtained by independent equilibration of eachsystem with a given PI value overnight or longer.The conditions in Lyklema’s andour potentiometric titrations differ in that each measured point was obtained aftersubsequent addition of the Ag+ or I- ions to the same sol.Finally, from the analysis of fig. 2 given by fig. 4, in the double layer at, e.g.,$0 = 300 mV in 0.1 M electrolyte, the total charge is a&-) = 4-6 measured potentio-metrically as adsorbed I- ions. Due to the presence of cations the neutralizingcharge is o+ = 3-5. The negative adsorption of anions of 0- = 0.9 is assumed toleave an equivalent excess of cations in the layer. Then the total charge of cationsaO(K+) = 3.5+0.9 = 4.4, would be approximately equal to the ao(I-) = 4.6 charge(all in pcoul/cm2).This result is, however, impossible, since only anions not neutralized with cationswould be negatively adsorbed, i.e., excluded from the volume of the liquid adheringto the surface.The amount of the negatively adsorbed anions is derived from theincrease of the concentration of anions in the bulk liquid, determined analytically.There is no doubt that these anions are neutralized with cations in an equivalentamount. If the amount of neutral electrolyte brought in the layer liquid is s, thenafter the negative adsorption the remaining amount of anions and cations of theneutral salt is s+ = s- = s-a-, thereby not leaving any excess of cations in thelayer liquid, hence s- +ao(I-) = s+ +ao(K+) ; and ao(I-) = ao(K+) and not ao(I-) =Q++cT-, as erroneously supposed by Lyklema.In this case the calculated plots ofa+ and the value a+ = 3.5 in the given example would be impossible. For this reason,the counter ion adsorption could serve for the determination of specific surface areas,contrary to Lyklema’s opinion.1 Herak and Mirnik et al., Kolloid-Z., 1960,168,139; 1961,179.130; Croat. Chem. Acta., 1965,2 van del Hul, Pvoefschrift (Rijksuniversiteit, Utrecht, 1966).37, 79GENERAL DISCUSSION 105Prof. J. Lyklema (Agric. Univ., Wageningen) said: In reply to Mirnik, experi-mental evidence shows that for AgI f: is more or less independent of PI over severaldecades. This means that f: is too insensitive a parameter to be used for reflecting thebehaviour of $0. A different question is why is independent of PI.This problemwas treated in 1961.According to Nernst the proportionality between PI and $0 can be proved thermo-dynamically without recourse to redox processes. The question whether or notcoagulation/peptization equilibrium is attained during titration experiments isimmaterial for the point under discussion, viz., the constancy of the specific surfacearea Ao, because it was experimentally established that A0 did not depend on the stateof aggregation. We have never found a detectable effect of the way of preparation orthe AgI-suspensions (including their ageing) provided the adsorption is measured pefunit surface and not per unit weight. The total adsorbing area during titration isabout 50m2.If all iodide added to change the PI would be used to neutralizeAgN03 (this is an overestimation) about 0.25 x 10-4 g AgI would be formed. Evenif the specific surface area of this freshly prepared suspension was 20 m2/g (more thantenfold the specific area of the aged suspension) only 0.12 m2 extra surface would becreated which is negligible compared with the original 50 m2.Mirnik’s last remark is apparently based upon a misconception of the notions“ adsorption ”, and ‘‘ negative adsorption ’,. By negative adsorption is understoodthe state in which the Gibbs surface excess is negative. In order to obtain a KNO3exclusion, no MN03 needs first to be adsorbed (in Mirnik’s nvmenclature : s = 0).Further misconceptions arise with respect to Mirnik‘s using symbols like a-(K+),meaning the contribution to the double layer charge of the anion potassium. Thismight explain why Mirnik has difficulties in understanding the charge balance in theelectrical double layer.Mirnik has raised a number of questions of fundamental character.First heassails the applicability of a physically significant potential $0, related to p1 throughNernst’s law. Part of the argument stems from electrokinetic experiments ([($o)-relation) but as the relation between f: and $0 is complex this evidence is inconclusive.Secondly, he states that the potential drop across the AgI/solution interface is physi-cally insignificant because there i s no metallic reference electrode in the system.Heapparently overlooks that, due to adsorption of potential-determining ions, particlescan acquire a (say) negative potential. There need be no redox reactions at thesurface because the colloidal particle is not part of a galvanic cell giving a current.The potential $0 is essentially the Galvani potential difference at a given PI minus thecorresponding value at a given reference (e.g., the z.P.c.).Further he contends that the specific surface area A0 changes with PI, i.e., with $0.This is not the case. The adsorption measurements have been made on looselypacked suspensions of big particles. The greater part of their surface remains freefor adsorption after making contact with neighbouring particles. (Even a substantialjoint contact of adjacent particles does not exclude adsorption at the unexposed partof the surface.) Experimentally the constancy of A0 can be proved in the followingways : (i) the negative adsorption (N.A.), directly measured on a deflocculated sus-pension is equal to the N.A.calculated from the adsorption isotherms, obtained withflocculated suspensions (fig. 4). (ii) The ao($o) curves in 10-3 M salt (fig. 2, lowestcurve) do not show a discontinuity around the potential where the particles start topeptize, and resemble closely the corresponding curves obtained by integration ofdirectly measured double-layer capacitances on flat AgI electrodes. That AgI,created during the titration does not affect significantly the adsorbed amount ofpotential-determining ions can easily be shown by an elementary calculation but iI 06 GENERAL DISCUSSIONAgBr/Br'-SOL +f++--also corroborated by the good reproducibility of the curves in titrating from left toright, and in the reverse manner.Mirnik's last argument denies the existence of a contribution of co-ions to thecountercharge. Simple reasoning shows that a deficit of anions around a negativeAgI-particle can arise without disturbing the electroneutrality of the system.Con-sider an uncharged AgI-particle, brought into a solution of KI and KNO3. Let 10neutral molecules of KI adsorb and let 3 neutral molecules of KNO3 be expelled fromthe interface. Then both the solution and the double layer remain electroneutral as awhole. The charge distribution is : surface charge - 10 units, due to adsorption ofI- ; counter charge : + 3 units due to deficit of NO, ; plus +7 units due to excessof K+.Finally a denial of the existence of negative adsorption is tantamount tosaying that no Donnan equilibrium exists.Prof. E. MatijeviE (Potsdam, New York) said: The dependence of the coagulationconcentration of the counterions on their charge has frequently been employed as atest for the theories of colloid stability. Originally, the authors of the DLVO theorycounterion chargeFIG. 1.-Plot of log molar coagulation concentration for a negative silver bromide sol as a functionof counterion charge. Dashed line represents the DLVO z6-law using the monovalent ion as thereference point.have implied that their model should explain the coagulation behaviour of 1-1,2-2,and 3-3 electrolytes.1 Now, it appears that Derjaguin and Overbeek believe that thetheory, in its original form, is applicable to mono- and divalent ions only.In thelatter case it makes little sense to use the coagulation concentration dependence oncharge as a criterion to test the validity of the theory. The reason is best explainedby inspection of fig. 1. The circles give experimental results obtained with a negative1 see, e.g., Verwey and Overbeek, Theory of the Stability of Lyophobic Colloidr (Elsevier, NewYork, 1948), p. 119 ; or Overbeek in Kruyt, Colloid Science (Elsevier, Amsterdam, 1952), vol.I, p. 306GENERAL DISCUSSION 107silver bromide sol 1 and cations of charges 1-4. As observed on many other systems,a linear relationship is obtained if the log molar coagulation concentration isplotted against the counterion charge z.Dashed line represents the DLVO +lawusing the monovalent ion as the reference point. If the z6 law applies only to z = 1and z = 2 coagulation dependence on ionic charge cannot be used to distinguishbetween the DLVO and other formulations of the Schulze-Hardy rule.Prof. D. D. Eley (Nottingham) said: Have silver iodide particles an exposedsurface of iodine ions or silver ions? This will greatly affect the hydration picture,since iodine ions are not hydrated whereas silver ions are hydrated according tocalculations of Bernal and Fowler and later workers.Prof. J. LyMema (Wagerzingen) said: In reply to Eley, it is difficult to give agenerally valid picture of the exposed surface ions in AgI particles since this dependson several factors : (i) the crystal structure (cubic or hexagonal (wurtzite type) struc-ture, depending on mode of preparation), (ii) the crystal faces that are exposed, and(iii) adsorption of excess Ag+ or I- ions.In our case no particular care was taken to prepare a homogeneous system ofuniform structure, although the hexagonal form is perhaps pre-dominant in themixture.The experimentally observed fact that AgI is hydrophobic suggests that inthe majority of the exposed crystal faces, mainly iodide ions are at the surface.Dr. S . Levine (Manchester University) (partly communicated) : Teiak’s statementthat our theory will not predict experimental results suggests that he has not fullyappreciated the purpose of our present work.A comprehensive theory of colloidstability is complicated, requiring the simultaneous treatment of different aspects andwe have restricted ourselves to one aspect of this problem. Since the Gouy-Chapman(G.C.) theory of the diffuse layer (the P.B. equation) has been an essential part of theD.L.V.O. theory of stability, an examination of its range of validity and of the natureof corrections needed, is important. This task is of special significance because ofindications that the G.C. theory should be abandoned at higher electrolyte concentra-tions, approaching 1 M for the 1-1 valency case. One significant conclusion of ourinvestigations is that most papers concerned with corrections to the P.B.equation bythe local balance methods have failed to predict correctly the net change in the diffuselayer potential distribution, because one or more essential contributing factors, inparticular cavity effects, were omitted. Most hydrophobic sols coagulate in 1-1electrolyte concentrations of the order of 0.1 M or less, and our calculations suggestthat for such systems the G.C. theory of the diffuse layer is a reasonable first approxi-mation and can therefore be applied to stability theory.A problem relevant to the Schulze-Hardy rule is the applicability of the G.C.theory for polyvalent coagulating ions. Although we have not considered a 2-2electrolyte, some qualitative information can be gained from our calculations at 0.01M, 1-1 electrolyte.Plots in this case of the various contributions to the correctionterm C(x) in our relation (3.4) for a single plate show that although the cavity effectsagain dominate at large x, they are of the same order as the self-atmosphere-imageterms for values of x,< 30 A ( 1 ~ x 5 l), indicating that cavity potentials in 0.01 M are notas dominant as in 0.1 M. Also, at plate potentials < 80 mV in (1-1) 0.01 M, C(x)does not exceed about 0-15 (at x = xg) and diminishes to less than 0.01 at large x.The concentration 0.01 M of 1-1 electrolyte, corresponds to a 2-2 valency type athalf the plate potential and at 0.0025 M (giving the same value of the Debye-Huckelparameter), which is a typical flocculating divalent ion concentration for silver-halide1 MatijeviC, Broadhurst and Kerker, J. Physic. Chem., 1959, 63, 155210s GENERAL DISCUSSIONsols. Because of an increase in ionic hydration, the Debye-Hiickel radius a and there-fore the cavity effects, may be greater than for the 1-1 electrolyte. Also the polariza-tion energy and dependence of E on concentration will be more proriounced (by factor ofabout 2) and we can expect a fourfold increase in the image-self-atmosphere term,since this is proportional to the square of the ionic charge. From a study of ourresults at 0.01 M 1-1 electrolyte, it appears that in 0.0025 M of 2-2 electrolyte, themaximum value of C(x) is about 0.5 at plate potential < 40 mV, giving corrections tothe P.B. equation comparable to those calculated in our paper at 0.1 M, 1-1 electro-lyte. Alizade, Martynov and Melamed 1 (A.M. and M.) have considered the screenedimage correction to the P.B. equation at low potentials and calculate that for a metallicplate in a 2-2 electrolyte the plate charge density at a specified potential can be twicethat given by the G.C. theory. However, they have omitted the first term in ourrelation (2.15) which describes the self-atmosphere effect. For a dielectric plate, thiscontributes a positive term to C(x), whereas the image term is negative, and in ourcalculations with 1-1 electrolytes, the latter is the smaller of the two except at smallx. A.M. and M. have recognized the importance of the image term in 2-2 electro-lytes but our calculations indicate that the other corrections should not be ignored.We can expect the G.C. theory to become worse with increase in ion valency, althoughthis is partly offset by the decrease in coagulating concentration. It seems that withtrivalent and higher-valent ions, the image-self-atmosphere terms become important,and on this ground alone, present-day electric double layer theory is probably incapableof providing a quantitative explanation of their coagulating properties.Another important aspect of stability theory, which we have not discussed here, isthe role of the inner Stern region, involving the intricacies of inhomogeneous dielectricproperties, ion-size factors and discreteness-of-charge effects. Just as with thediffuse layer, we must expect the problems of the inner region to become moredifficult with increase in valency of adsorbed counter-ions. Because there is anumber of contributing factors in colloid stability, of which the G.C. theory is onlyone, a simple general Schulze-Hardy rule for coagulation, particularly at the highervalencies, is unlikely. The Teiak form of this rule (linear relation between thecoagulating concentration and the valency of the coagulating ion) seems empiricalsince theoretical attempts at its justification by TeZak and Mirnik are unsatisfactory,and it fails near the isoelectric point.* Alizade, Martynov and Melamed, Dokl. Akad. Nauk S.S.S.R., 1963, 151, 601

ISSN:0366-9033    

DOI:10.1039/DF9664200091

出版商:RSC    

年代:1966

数据来源: RSC

摘要:

Effect of Lyophile Surfaces on the Properties of BoundaryLiquid FilmsBY B. V. DERJAGUINDept.of Surface Phenomena, Inst. Physic. Chemistry,U.S.S.R. Academy of Sciences, MoscowReceived 27th June, 1966The characteristics of coagulation and stability of lyophobic colloids in diluted electrolyte solu-tions are now well explained in a quantitative way by the theory based on the forces of two kinds :those of van der Waals and those appearing as a result of overlapping of ionic double layers. Thequestion now arises whether this theory can be made to apply to all colloids for all electrolyte con-centrations by improving the theory but still basing it on only these two kinds of forces. Thenegative answer to this question can be established either by the analysis of coagulation of solsespecially in the presence of surfactants, changing the hydrophobic nature of the particle surfaceto a hydrophilic one, or by direct investigationof the properties of the layers of liquid near thelyophilic surface.In this paper an account is given of the results of the second approach to the problem.Resultsare given of the measurement of mechanical and thermodynamic properties of boundary layersof liquids. There are also discussed the direct experimental proofs of the ability of glass and quartzsurfaces to change the physical properties of many polar liquids to a great depth.Following Kallmann and Willstatter,;? the theory of stability of lyophobic colloids 1takes into account forces of two kinds : (i) effective attraction (negative componentof disjoining pressure) which is the net result of the van der Waals forces acting inthe zone of approach of two particles both between their molecules and the mole-cules of the dispersion medium ; (ii) effective repulsion (positive component ofdisjoining pressure) arising when the ionic atmospheres of the two particles overlap.The existing methods of evaluating these forces are based on the assumptionthat the properties of an electrolyte solution in the zone of approach differ fromthose in the bulk only in ion concentration.In other words, the dielectric andmechanical properties in the vicinity of and between the surfaces of the particlesare the same as in the bulk. These assumptions are justified by the following:(i) good agreement with the 26 law derived from theory for coagulation by uni-and bivalent counter ions 3 ; (ii) model-type experiments by means of which theforces of both kinds can be studied separately 4 or simultaneously.~There are, however, cases when the theory is either inapplicable (high values ofelectrolyte concentrations, unknown surface potential) or disagrees with experi-ment.Thus, measurements of the potential barrier on adhesion of crossed wiresexhibit maxima at high (1-3 N) concentrations of both univalent and bivalentelectrolytes, the height of which is no lower than at low concentrations. Thenature of this effect is still unclear. On the other hand, there exist sols which are stableat any electrolyte concentrations. Finally, Glazman 6 has shown that when certainsurface-active substances are adsorbed by lyophobic particles the regularities ofcoagulation cannot be described by theory.Various approaches may be used to account for the stability of such lyophilicor lyophilized sols. One is to assume the existence of repulsion forces when the10110 PROPERTIES OF BOUNDARY LIQUID FILMSparticle solvate layers overlap, resulting in destruction of their peripheral parts.Several papers show the stability of lyophilic sols to be in agreement with this pointof view.However, such proofs cannot be quantitative until a quantitative theoryof solvation is developed. Meanwhile, the existence of polymolecular solvatelayers with specific properties is sometimes questioned. Owing to this, directexperimental investigation of the special properties of liquids bounding on lyophilicsurfaces, properties due to the influence of these surfaces, are of prime importance.Owing to progress made in the solving of this problem,7 I need deal only witha few of the results.Our conception is that the properties of the boundary layersof liquids may change substantially due to changes in structure caused by contactwith a foreign phase. On the other hand, the effect of molecular and electro-static (dielectric saturation effect) fields is more limited. Thus, molecular fieldsdiminish rapidly with distance, while electric fields are large only at high electro-lyte concentrations within very thin double layers. In any case, there are manyinstances where a sharp change in the properties of boundary layers has been provedunambiguously, but cannot be traced to the action of these forces.At the sametime, these phenomena may be attributable to the interaction of the liquids withthe foreign phase, e.g., through a hydrogen bond, if it is assumed simultaneouslythat in liquids a change in structure of the surface monolayer may extent into thedepths of the liquid to a considerable extent. In this paper, we not only give proofthat the structure of boundary layers is different from that of the bulk, but also showthat the liquid-phase nucleus of specific structure, formed when unsaturated vapourscondense on a substrate may grow to a size of about 20 p, which far exceeds the range ofdirect surface action.The difficulty is to understand why boundary layers withspecific properties usually have a strictly limited thickness. It must be admittedthat the structure of these layers is different (being perhaps anisotropic) and cannotbe retained in the bulk phase.To confirin that boundary layers possess structural peculiarities (apart fromdirect methods of structure investigation) some of their structure-sensitive propertiescan be studied. As such, we consider first the mechanical properties relating toshear deformations or flow. In a number of papers the use of the blow-off methodby which the viscosity is determined as a function of the distance to the substrate,has made it possible to establish that the viscosity of the boundary layers of a numberof polar organic liquids differs from its bulk value.8 No such difference is observedwithin the limits of the high accuracy of the method for non-polar vaseline oil ofhighest purification.Hence, the specific boundary viscosity cannot be attributedto any secondary factors, such as microroughness of the substrate (glass, metal)which would have the same effect in all cases. Owing to technical diEculties theblow-off method has not been applied so far to such comparatively volatile liquidsas water. On the other hand, other methods give only a certain mean viscosity asa function of the layer thickness.9 Of more general applicability is a method bywhich the shear elasticity modulus of any liquids can be measured as a function ofthe distance to the substratelo (quartz).This method is based on the change inresonance frequency of piezoquartz (fig. 1) under the influence of the horizontalsurface of a quartz bar placea on it and separated from it by a plane-parallel layerof liquid of thickness H. An X-5 cut of the quartz is taken, and it is excited byapplying a voltage so that its horizontal faces vibrate in their own planes. Thequartz bar remains practically at rest, while the liquid interlayer undergoes sheardeformations. The frequency of self vibrations of the piezoquartz was about 75kiloherz and was determined from a resonance curve taken by varying the fre-quency of the input voltage. Tf the liquid displayed only its viscous properties wheB . V . DERJAGUIN 111the shear deformations are alternated, the resonance frequency should have de-creased.Without exception, experiments show the opposite, i.e., that the frequencyof the quartz increased. The positive frequency shift grew with decreasing amplitudeof vibration of the piezoquartz, tending to a definite limit. Proceeding from thislimit, found graphically, the effective modu-lus of shear elasticity could be determined.The influence of dissipative forces, whichare not taken into account in these cal-culations, could only reduce our shear modu-lus values compared to the “ true ” ones.Assuming for generality that the shear modu-lus of a liquid is a function of the distancez to the nearest quartz surface, G = G(z),the following formula could be derived forgraphical determination of the function G(z)from experimental data for various liquidinterlayer thicknesses H = 2h :G(h) = ‘so( 1/--d7;-).d(Af)-- (1) SFIG. 1 .-Piezoquartz with quartz plate.Fig. 2, 3 and 4 show the values of (An-1 plotted as a function of h for several liquids.For non-polar liquids the experimental points fit on straight lines passing throughthe origin. Thus, not only is the slope of the dependence, and therefore, accordingto formula (I), also the shear modulus, constant over the entire interval of distances2 0IS50 0’1 0 - 2HI 2FIG. 2.-Diagram for graphic determination of shear modulus of cc14 (1) and C6H6 (2).h studied, but there are no grounds to expect deviation from constancy at lowervalues either, the more so that at the limit, at h 4, Af+m and therefore, (An-1 -+O.For polar liquids the experimental points fit on straight lines which do not passthrough the origin.Ths means that: (i) in the thickness intervals studied, th112 PROPERTIES OF BOUNDARY LIQUID FILMSshear modulus is constant, i.e., has the bulk value independent of the influence ofthe quartz surfaces. At lower h values, the dependence of l/Afon h cannot inter-sect the abscissa, because this would signify a physically impossible transition ofAf from f a to -m, and therefore this dependence must deviate from rectilinear,1 3100 ccFIG.0 0. I 0'. 2Hi2alcohol (3).3.-Diagram for determination of shear modulus of acetone (1); ethyl alcohol (2); octylL I I I0 0.1 0.2 0.3 0 . 4 0 . 5HI2(3) distillate.FIG. 4.-Diagram for determination of shear modulus of water : (I) tridistillate ; (2) bidistillate B .V. DERJAGUIN 113its slope becoming more gradual. This means that the shear modulus at smallerthicknesses must be greater than in the bulk. The bend in the (l/AJ h) dependenceshould begin to the left of the extreme left experimental point, because the latteris still on the straight line, but to the right of its intersection with the abscissa axis.As these upper and lower limits of the thickness ho of the boundary layer with alteredmechanical properties are close, a good estimate of ho-values is obtained. Table 1lists the ho values and bulk values of G for a number of liquids. The data obtainedindicate unambiguously that there exist boundary layers of polar liquids about500-800 A thick near the quartz surface.substanceacetonewaterbenzenecc4acetic acidethyl alcoholbutyl alcoholhexyl alcoholoctyl alcohololeic acidvaselin oilcastor oilTABLE l.-SNEAR MODULUS OF LIQUIDSbulk value of theshear modulus purification grade T'Cdistill.25 0.4 x 104trildistill. 25 1.1 x 104absolut. 14 1.3 x 104distill. 16 2.3 x 104distill. 17 1.8 x 104absolut . 16 1x104$9 18 0.7~ 104Y > 26 0*9x 10425 1*2x 104filtdd 16 3 x 105Y9 14 3 . 2 ~ 105Y, 17 0.7~ 106thickness of the boundarylayer in p0-080.09000-060.060.070.080.08 -IA typical structure-sensitive property of water is its thermal expansion with acharacteristic minimum at 4°C. In a study of the expansion of water in glass capil-laries 180 A and more in radius, Fedyakin 11 found that for a radius of 200 A theexpansion of water is almost exactly rectilinear from -12°C to +50°C.Withincreasing the expansion graph becomes curved, and at z = l0OOA it acquiresthe usual shape. This gives an estimate of the thickness ho of the boundary layerwith specific properties, close to the previous one. Later, Karasev and I investigatedthe thermal expansion of water in the pores of a highly dispersed quartz powder(Aerosil), the hydraulic pore radius of which was about 150 A. We could detectno volume minimum above 0°C. When the water was removed from the pores ofthe powder (by heating) into a dilatometer with a wide volume, the normal shapeof the expansion curve with a volume minimum at 4°C was restored.This provesthat the distortion of the normal expansion curve in thin pores does not dependon the change in its composition due to the leaching of the quartz, but takes placeunder the direct influence of the quartz surface.Such investigations leave open the question of the structure of the boundarylayers of water or other liquids. However, Green-Kelly and myself12 have shownthat layers of water and several other strongly polar liquids about 30-200A thick,formed between the silicate layers of swollen montmorillonite possess a distinctbirefringence (about 0.002-0.003). This is proof of the oriented structure of theselayers-a sort of liquid-crystalline state.An obstacle to recognition of the specific properties in boundary polymolecularlayers is the traditional conception of liquid structure, i.e., of absence of long rangeorder in them.From this it is often, usually implicitly, concluded that no re-arrangement of liquid structure under the influence of an adjacent phase can spreadinto the depths of the liquid to a distance more than several molecular diameters114 PROPERTIES OF BOUNDARY LIQUID FILMSThis idea is closely related to another, according to which the structure (short rangeorder) and properties in the bulk of a liquid are single-valued functions uf the tem-perature and pressure. Of special interest, therefore, both for the question discussedin this paper and for the liquid state, are experiments in preparing liquid modifica-tions with altered physical properties. The first observation was made by Fedyakin 13who observed the behaviour of liquid columns, e.g., of water, methyl alcohol andacetic acid, introduced into sealed glass capillaries 1-2 ,Y in radius.He discoveredthat “ daughter” columns appear and grow evenly at both ends of these columns,while the initial column grows shorter. This indicated that the daughter columnspossessed a lower vapour pressure than the initial one, the vapour pressure of whichdid not differ (owing to the small curvature of its menisci) from the tabulated value.Later, Fedyakin and I 14 demonstrated that with water, e.g., the viscosity of theanomalous columns is about 12 or 15 times its normal value.FIG. 5.4hamber for production of anomalous liquid columns from unsaturated vapours.As the tests were performed in narrow glass capillaries the results could havebeen influenced by passage into the columns of the leaching products of the capillarywalls; but this was shown to explain the effects observed only in part.15 Never-theless, experiments were made afterwards in open quartz capillaries from whichthe air was preliminarily evacuated to accelerate growth of the columns and for moreexact control of the vapour pressure around the capillaries.16 The capillary radiiwere varied over a wide range from 1 to 30 p.Fig.5 is a schematic representation of a unit for preparing anomalous columnsdirectly from the vapour, without resorting to primary columns. The air wasevacuated from chamber 5 surrounded by water jacket 2, 6, through cock 1.Thechamber communicated with the side branch 9 containing the liquid under studywhose vapours filled the chamber. By means of water jacket 10 the temperaturein the branch could be maintained at a definite lower level than in chamber 5 andthus any desired relative pressure p/pI less than unity could be set up in the latter.To determine p/ps it was sufficient to measure the temperature difference betweenthe liquid surface and the chamber near capillary 7 fixed on holder 8 by meansof thermocouple 3’. By measuring, additionally, the temperature near the capillarB. V. DERJAGUIN0.04 '0.03.P z 2 0.02-d0 0 9 .1150*04.0.03 -$J 0.02- - 2a 0.01-0 -PipsFIG. 6.-Measurement of equilibrium vapour pressure of water columns.0900.91 0.92 0.93 0 .9 4 0.96 0.97 0 . 9 116 PROPERTIES OF BOUNDARY LIQUID FILMSwith thermometer 3, the p/pS value in the capillary could be determined. Theinside of the capillary was observed through window 6 by means of a measuringmicroscope with dark-field illumination.At first the capillary was empty and its channel scattered light. The columnappearing subsequently was darker owing to lower light scattering, making it easyto measure its length. For preparation of anomalous columns with certainty itwas sometimes necessary to establish communication between the chamber and apump for several seconds, after which the preset value of p/ps was restored by closingthe cock. It is noted that after their appearance the columns (or drops on theouter surface of the capillary, which also appear as a rule) grow at a constant rateu = AhJA in an unsaturated atmosphere for tens of hours, reaching a size ofseveral cm.By measuring the growth rate for different values of p/ps we obtained the depen-dences shown in fig.6,7,8. The intersection of the corresponding lines with the abscissasubstancewaterY99979Y39 ),Y$ 99 9acetoneY Y9 9methyl alcohol9 9 Y ?,Y 7 9 acetic acid1 9 9373 ,7TABLE 2capillary materialglass N46,N23quartz glasscapillaryradius535121822121212123815682061523Pips0.330.930.930-930.930.930.970.930.910.900.950.950.950.940.940.940.950-950.95roc2020202020204119.554202020202020202020axis allowed determination of the relative vapour pressures pa/ps.Table 2 givesthe results for a number of liquids in capillaries of different radii r. For water,measurements for different temperatures are given. The absence of any perceptibledependence of pa/ps on r indicates that in these experiments there were bulk modi-fications of the corresponding liquids formed, while the part played by the wallswas essentially to ensure the appearance of a nucleus of the new phase. Possibly,an adsorption layer of definite structure and sufficient thickness may serve as sucha viable nucleus.Subsequent measurements 19 revealed that the viscosity of water columns in quartzcapillaries does not exhibit any dependence on the capillary radius either within thelimits of experimental error.The viscosity values were approximately 15 timeshigher than the tabulated value. Later, Churayev, Yershova, Zheleznyi and I 17made measurements of the thermal expansion of water columns. For this, afterobtaining the columns (4-20mm long), the quartz capillaries were sealed at bothends and placed in a cryostat. The column lengths were measured with a comparateB . V. DERJAGUIN 1178 0 -40 -0-to the nearest 2 p. The results are given in fig. 9. Curve 2 represents the expansionof normal water zccording to tabulated data. The experimental points obtainedwith a water column introduced into the capillary in liquid form fit well on it.Curve 1 is the expansion of a column obtained from vapours, for whichp,/p, = 0.93(at 20°C).The length increment in both cases is taken as arbitrarily from 0". Thesharp difference between the two curves is evident. Besides the leftward shift ofthe minimum volume temperature, in the 20-40°C range the temperature expansionof anomalous water is approximately 13 times greater than that of normal water.In glass capillaries, 1-2 p in radius, a similar result had been found.13120 1 PI -- 2 0 0 2 0 3 0 4 0I I I It"Cwater.FIG. 9.Thermal expansion of columns; lower curve, the normal water; upper curve, specificAn essential question is that of the stability of both phase states of water incapillaries. After heating to 150", and maintaining at that temperature for 1 hand subsequent cooling, the expansion curves reproduce exactly. Moving thecolumn into a wider (up to 0.1 mm) part of the capillary and returing it again hasno effect either.Great differences are observed in the behaviour of columns of the two kinds oncooling (see fig.10 in which the scale of the ordinate axis is condensed comparedto fig. 9). A normal water column can usually be supercooled from -20" to-3O", after which it crystallizes, as is evident from the sharp expansion of thecolumn accompanied by spreading of the crystallization front. When heated,the column at 0°C becomes rapidly shorter, indicating melting. '' Specific "columns usually display a steep increase in length at a temperature of from -40to -6O"C, after which a normal course of hardly perceptible shortening beginson further cooling.When the column is heated, a sharp decrease in length occursin the interval from -30 to -15"C, this being an indication of quasi-melting.However, the change in length of the specific column at quasi-freezing and quasi-melting is only about 5 %, proving that specific water does not change into ordinar118 PROPERTIES OF BOUNDARY LIQUID FILMSice. The same is indicated by the microscopic observations made by Zheleznyiwho found spherical particles to form at - 60" and to disappear at - 15°C. Itappeared also that the descending part of the hysteresis loop can be traversed inthe reverse direction if the spheroids of the "second phase" do not disappearbefore the " turn ", i.e., a reversible phase transition is possible.Thus, with" specific " water, the " phase transformation '' is spread over the region from - 30to - 15°C. After passing the hysteresis loop with disappearance of the sphericalparticles, i.e., at temperatures above - 1O"C, the initial thermal expansion curveis reproduced without change.We summarize: the usual state of water and certain other liquids is thermo-dynamically metastable. Therefore, in an investigation such as that described-40-20 2 0 4 0FIG. lO.-Cooling and heating diagram for columns of: (1) normal water ; (2) specific water;circles, first cycle ; triangles second cycle ; (2') heating.here, it would be convenient to call '' usual water " metawater, and the anomalouscolumns-orthowater.The practical stability of metawater must be due to the highvalue of the work of formation of a critical nucleus of orthowater. Possibly also,the growth of orthowater in contact with metawater is a slow process having a highactivation energy.For the present discussion it is more essential to emphasize that the experimentsdescribed prove the variability of structure of a number of liquids and thereforemake it easier to understand their ability to change it under the influence of aninterface with a different phase.Finally, despite the paradox of the phenomena observed, their theoretical treat-ment, for water, e.g., seems possible on the basis of present-day conceptions ofits structure. Moreover, Gurikov,l8 who proposed a theory of the thermodynamicproperties of water at the triple point on this basis, concluded that there may exista modification of water with a lower vapour pressure.Compared to " meta-water " this modification features a higher filling of the cavities in an ice-like skeletoncemented by hydrogen bonds.1 Derjaguin, Zzvest. Akad. Nauk U.S.S.R., Ser. Chim., 1937, no. 5, p. 153 ; Acta physicochim.,1939,10,333 ; Trans. Faraday SOC., 1940,36,203 ; 1940,36,730. Derjaguin and Landau, ActaPhysic Chim., 1941, 14, 633 ; J.E.T.P. (Russ.), 1945, 15, 663. Verwey, Philips Res. Reports,1945, 1, 33. Verwey and Overbeek, Theory of the Stability of Lyophobic Colloids (Elsevier,Amsterdam, 1948) ; Trans. Farnday Soc., 1947, 43, 517.2 Kallmann and Willstiitter, Naturwiss., 1932, 20, 952.3 Neiman, Koll.J. (Rum.), 1966, 28, 110B . V. DERJAGUIN 1194 Derjaguin and Abrikossova, Disc. Faraday SOC., 1954, 18, 24 ; Quart. Reu., 1956, 10, 295 ;J. Physics. Chem. Solids, 1958, 5, 1 ; Uspechi Physisch. Nauk, 1958, 64, 493. Kitchener andProsser, Proc. Roy. SOC. A, 1957, 242, 403. Black, Longh, Overbeek and Sparnaay, Trans.Faraday SOC., 1960, 56, 1597. Derjaguin and Titijevskaya, Proc. 2nd Znt. Congr. SurfaceActioity, 1957, 1,211. Scheludko, Platikanov and Manev, Disc. Faraday Soc., 1965, 40,253.5 Derjaguin and Voropayeva, J. Colloid Sci., 1964, 19, 11 3.6 Glazman and Sapon, Research in Surface Forces, II (Consult. Bureau Press, New York, 1966).7 Research in Surface Forces, II, ed. Derjaguin (Consult. Bureau Press, New York, 1966).SDerjaguin and Karasev, Proc. 2nd Znt. Congr. Surface Actiuity, 1957, 3, 531. Derjaguin,Karasev, Zakhavaeva and Lasarev, Wear, 1958,1,277. Derjaguin, Zachavaeva and Chomutov,Research in Surface Forces, II.9 Derjaguin, Zakhavaeva and Lopatina, Research in Surface Forces, ZZ.10 Bazaron, Derjaguin and Bulgadaev, Compt. Rend. U.S.S.R., 1966,166,639 ; J.E.T.P., 1966,51,11 Fedyakin, Compt. Rend., U.S.S.R., 1961, 138, 1389. Derjaguin and Karasev, Koll. J. (Russ.),12 Derjaguin and Greene-Kelly, Trans. Faraday SOC., 1964, 60, 449.13 Fedyakin, Kolloid. J. (Russ.), 1962, 24, 4.14 Derjaguin and Fedyakin, Compt. rend. U.S.S.R., 1962, 147, 402.15 Fedyakin, Derjaguin and Talaev, Compt. rend., U.S.S.R., 1965,165, 878.16 Derjaguin, Talaev and Fedyakin, Compt. rend., U.S.S.R., 1965, 165, 597.17 Derjaguin, Churaev, Erschova and Zhelemyi, Compt. rend., U.S.S.R., in press.18 Gurikov, J. Strukt. Chem., 1965, 6,817; 1966, 7, 8.19 Derjaguin, Fedyakin and Talaev, Compt. rend. U.S.S.R., 1966, 167, 376.1969.1962,24,471

ISSN:0366-9033    

DOI:10.1039/DF9664200109

出版商:RSC    

年代:1966

数据来源: RSC

摘要:

Role of Water Structure in the Interpretation of ColloidStabilityBY G. A. JOHNSON, S. M. A. LECCHINI, E. G. SMITH,J. CLIFFORD AND B. A. PETHICAUnilever Research Laboratory, Port Sunlight, CheshireReceived 16th May, 1966Some studies of the stability of model colloidal systems have shown marked deviations from theclassical Derjaguin-Landau-Veny-Overbeek (DLVO) theory. These deviations have been inter-preted in terms of structuring of the water next to the particles. To test this postulate it was decidedto establish whether the fastest flocculation rates of a model system in the form of monodispersepolyvinylacetate sols were solely diffusion controlled as predicted by classical Smoluchowski theory.The flocculation rates of these sols in the presence of excess electrolyte were measured as a functionof temperature using light-scattering methods and a Codter counter.The experimental rates werelower than those predicted by the theory tending to confirm the initial postulate. Further evidence ofthis phenomenon has been gained from a study of relaxation times of the water in the system, as afunction of temperature and particle concentration, using spin echo n.m.r. techniques.Studies 1 of the flocculation of arachidic acid and octadecanol sols at differentelectrolyte concentrations have shown marked deviations from the Derjaguin-Landau-Venvey-Overbeek (DLVO) theory. For example, arachidic acid sols werestable at zero zeta potential. To explain the observed experimental results obtainedon these systems, it was postulated that structuring of water next to the particlesmight provide an additional repulsive potential particularly noticeable when theelectrostatic potential became very small.Some support for this theory was ob-tained by Fawcett et aZ.2 who measured the relaxation times of water protons aroundsilver iodide sols by spin echo n.m.r.3 They suggested that at room temperaturethe correlation time for molecular motion4 of water molecules around a particlewas longer than for bulk water, and that the correlation time increased wjth de-creasing surface charge.To verify the existence of structured water around colloid, particles polyvinylacetate (PVA) sols were chosen as model systems, because this type of particle couldbe prepared in monodisperse spherical form and with varying sizes less than 1 p.To observe the flocculation behaviour of these particles under conditions in whichthe electrostatic repulsion potential is insignificant, flocculation rates were measuredas a function of temperature by light scattering in the presence of excess electrolyte.Flocculation rates lower than those predicted by the Smoluchowski theory 5 wouldthen be strong evidence for an additional energy or entropy barrier.The light scattering results for the PVA sols could not be interpreted in termsof Rayleigh theory.6 The variation of the relative flocculation rates with temperature,however, showed marked deviations from Smoluchowski theory, with activationenergies larger than expected for a single diffusion-controlled process in water.Itwas therefore felt desirable to obtain absolute flocculation rates on a closely-relatedsystem by means of a Coulter counter.7 For this purpose, a polyvinyl toluene(PVT) latex of 1.9 p diam. was chosen. The fastest absolute flocculation rates for12JOHNSON, LECCHINI, SMITH, CLIFFORD AND PETHICA 121PVT sols showed similar variations with temperature to those found for the PVAsols. In addition, the absolute rates at low temperatures were below those pre-dicted by the Smoluchowski theory.To obtain inore direct evidence for changes in water structure, techniques arerequired which can measure the mobility of water molecules, or conversely theirdegree of hydrogen bonding. The most suitable are spectroscopic techniques suchas infra-red, nuclear magnetic resonance and dielectric relaxation.In this paperwe present evidence obtained from n.m.r.Continuous wave n.m.r. techniques have been used to detect water binding inbiological colloidal systems. The line widths and intensities for the protons ofwater in these systems have been compared with those observed for pure water.Many of the results of such measurements have been incorrectly interpreted becauseof errors introduced by instrumental artefacts8 Moreover such factors as fieldhomogeneity prevent absolute measurements of molecular mobility. Direct measure-ments of the relaxation time of water protons are of particular value because thespin-lattice (longitudinal) and spin-spin (transverse) relaxation times, TI and T2,can be directly related by the theory of Bloembergen, Purcell and Pound (BPP)9to the correlation time characterizing molecular motion, provided that it is assumedthat nuclear relaxation is caused by the interaction of magnetic dipoles under theinfluence of molecular Brownian motion.This assumption is justified for the systemsand temperatures under consideration. 5l.m.r. relaxation times can be convenientlymeasured by pulse techniques (spin-echo n.m.r.>.3 Such methods have been usedto study water in ovalburnin solutions where evidence was found for longer correla-tion times than for pure water.10While it is probable that most biological membranes and colloids in aqueousphases are associated with at least one molecular layer of " bound " water we presentevidence to show that from the viewpoint of colloid stability the important changesin water structure occur when the colloid particles are separated by distances atwhich the London-van der Waals dispersion forces are significant.EXPERIMENTALPREPARATION OF PARTICLESThe reagents used for the preparation of the two types of PVA sols (0-13 p and 0.8 p diam)were as follows.Vinyl acetate, Hopkin and Williams Ltd., was redistilled twice before useto purify and remove hydroquinone present at a concentration of 20 p.p.rn. Sodium octa-decyl sulphate was prepared from pure octadecanol by sulphation with sulphur trioxide. Allother reagents were of A.R. grade. The nitrogen was oxygen free (BOC white spot). Distilledwater was passed through a mixed-bed ion exchange column and redistilled from alkalinepermanganate.Glassware was cleaned with chromic acid and thoroughly steamed beforeuse.300 ml lO-5 M sodium octadecyl sulphate(the stabilizer) was placed in a reaction vessel, 1 g potassium persulphate (the initiator) wasthen added and the reaction vessel placed in a thermostatted water-bath at 60 f 1 "C. Whenthe solution had reached the temperature of the bath, nitrogen was passed through for 5 rninto remove oxygen, which has an inhibiting actior, on the polymerization. Vinyl acetate(6 g) was then added and the solution vigorously stirred for 3 h. The sol was then filteredthrough Whatman paper 541 and dialyzed for 2 weeks against pure water, using cellophanedialysis bags.This removed electrolyte, sodium octadecyl sulphate and any unreacted vinylacetate remaining in the suspension. For the 0.8 p particles the preparation was identicalexcept that 5 g potassium persulphate were added, nitrogen was then passed through thesolution for 15 min, 50 g vinyl acetate added and stirring continued at 60°C for 30 min.The 0.13 p particles were prepared as follows122 WATER STRUCTURE AND COLLOID STABILITYElectron micrographs of the two sols are shown in plate 1 and the size distributions obtainedfrom these are shown in fig. 1 and 2. For the n.m.r. measurements a concentrated suspensionwas made which could be diluted to the required particle concentration,diameter (microns)FIG. 1.-The size distribution histogram for the 013p PVA particles.diameter (microns)FIG.2.-The size distribution histogram for the 08p PVA particla1 micronPLATE la.-Electron micrograph of the 0 . 1 3 ~ PVA.10 micronPLATE 1b.-Electron micrograph of the 0.8~ PVA.[To face page 122JOHNSON, LECCHINI, SMITH, CLIFFORD AND PETHICA 123The octadecanol dispersion 11 was made by ultrasonic emulsification of molten octade-canol in pure water at 70°C. The resulting emulsion was diluted by pouring it into coldwater at 10°C. The mean diameter of the particles was 0.3 p. All suspensions were storedin a refrigerator at 2-3°C. The polyvinyl toluene was a monodisperse latex (1-9 p diam.)supplied by Dow Chemical Co.FLOCCULATION EXPERIMENTSOPTICAL DENSITY MEASUREMENTSMost of the flocculations were followed by measuring the optical density of the sols as afunction of time, using a Unicam SP600 spectrophotometer fitted with a thermostatted test-tube holder.1 The relative flocculation rates were calculated from the initial slopes of theoptical density-time plots.The sols were flocculated by the addition of O-Srnl 0.1 MLa(NO3)3 to 10 ml of the sol. The same results could be obtained by using other electro-lytes at concentrations above the flocculation values. Some of the flocculations were followedby measuring the light scattered at 20" as a function of time, using a Brice Phoenix light-scattering photometer.12COULTER COUNTERThe flocculation of polyvinyl toluene lattices was followed by the Coulter counter,7 50 plportions being fed through a 30 p orifice tube.Measurements were taken every 24 h on alatex having a concentration of the order of 105 particles per ml and suspended in a mixedelectrolyte solution of 0.3 M sodium chloride and 0.1 M lanthanum nitrate. At 30°C theflocculation time,l3 the time taken for the total particle count to decrease to one half of itsinitial value, was about 8 days. The suspension was placed in a 50ml sample tube andslowly turned end-over-end in a thermostatted water tank to prevent sedimentation. Thespeed of the rotation was 20rev/h, well below the value where the effect of shear on theflocculation rate can be experimentally measured. The flocculation rate was calculated froma plot of the reciprocal of the particle concentration against time.NUCLEAR MAGNETIC RESONANCETo measure the relaxation times TI and T2 a Bruker-Physik pulse spectrometer 302 wasused at an operating frequency of 60 Mc/sec.Short 7'1 were measured by repeated 180"-90"pulses, longer TI by 180"-90"-180" pulses.14 T2 were measured by the Gill and Meiboom 15modification of the Carr-Purcell14 method, in which a 90" pulse is followed by a train of180" pulses. The pulse separation was small enough to eliminate the effect of diffusion onthe echo decay. Other TI measurements were made by using conditions of adiabatic rapidpassage.16 The temperature was controlled to fO.1"C by means of a gas-flow thermostat.17It is necessary to guard against the presence of free radicals and paramagnetic substances,since these will markedly affect the experimentally determined relaxation times.Prolongeddialysis should remove any unreacted initiator and therefore no free radical source would beexpected to remain in the system. For the PVA sols, this was checked by electron spinresonance and found to be the case. Atmospheric oxygen will also affect the results but theconventional degassing technique involving repeated freeze-pump-thaw cycles, could not beused because the sols flocculated on freezing. At a given temperature, for all sampleconcentrations in equilibrium with the atmosphere, a constant amount of oxygen in thesuspension would give rise to a small systematic error in the relaxation measurements. Theexperimental results show a more complicated behaviour (see later).RESULTS AND DISCUSSIONFLOCCULATION MEASUREMENTSFig.3 shows the dependence of log I< on 1/T, where K is the fastest flocculationrate (in particle ml-1 sec-I), for the PVT latex. Experimental rates were measuredwith the Coulter counter, and hence are absolute rates. The interesting featureis the apparent change in activation energy governing the flocculation process whic124 WATER STRUCTURE AND COLLOID STABILITYoccurs at a temperature around 35°C. At the lower temperature the activationenergy has the value expected for a diffusion-controlled process, whereas the absoluterate is lower than that predicted by the Smoluchowski theory. Probably at lowtemperatures the additional barrier, whose presence we postulate to explain theexperimental results, is controlled by the same physical process which determinesthe rate of translational diffusion, viz., the making and breaking of hydrogen bonds.If, e.g., two particles, after having diffused through bulk water, find that at a givenseparation the water decreases its configurational entropy by becoming morestructured (i.e., “ colder ”) then the resulting decrease in their mean diffusion velocitywould reveal itself in a slowing down of the flocculation process.In a conventionalpotential-energy against distance diagram, the sum of the additional barrier provided\6l / T x 103 in OKFIG. 3.-The dependence of the absolute flocculation rate K on the absolute temperature Tfor thePVT particles.by structured water and the London-van der Waals’ potential,l3 cannot lead to anet potential energy barrier of much more than kT, otherwise the system would bestable.Moreover, it must act over distances comparable to those dictated by thedispersion forces, otherwise a secondary energy minimum would be formed. Ifthis was deep enough for flocculation to occur then the activation energy barrierwould still have the Smoluchowski value.When the temperature is increased beyond 35°C for PVT sols the layer of morestructured water begins to break down under the influence of thermal motion andthe energy barrier postulated above slowly disappears. The flocculation rate willtherefore increase with temperature more rapidly than predicted for a diffusionprocess in bulk water, although at a high enough temperature the theoreticalSmoluchowski slope should be restored. It is not surprising that the experimentalrate in fig.3 eventually crosses the theoretical line, since the simple Smoluchowskitheory does not take into account possible increases in rate due to attraction of theparticles at close range.1JOHNSON, LECCHINI, SMITH, CLIFFORD AND PETHICA 125Fastest relative flocculation rates for PVA sols of 0.13 and 0-8 p diam., and foroctadecanol of mean diameter 0-3 p are shown in fig. 4. Although the rates arenot absolute, the indication in fig. 4 that the rate of flocculation for the 0-13 particlesis lower than for the 0-8 p particles is probably correct. This may be concludedfrom the fact that during the flocculation process, the ratio of the times taken forthe turbidity of the two sols to approach close to the limiting value is much largerthan that required by the differences in particle concentrations.The behaviour of the 0.13 p PVA particles in comparison to that of the 0.8 pparticles shows a further contrast.The log rate against 1/T plot, in the temperaturerange examined, always has a slope greater than that given by the SmoluchowskiI I L3.2 3.4 3.61/T lo3 in OKFIG. 4 . 7 7 . e dependence of the relative flocculation rate R on the absolute temperature T for the0.13 and 0 . 8 ~ PVA sols and for octadecanol sols.theory. On the arguments given previously, the slope below a certain temperatureshould revert to its theoretical value, although this may occur at a temperaturewhich is too low for flocculation rates to be measured experimentally.This suggeststhat the additional energy barrier associated with water structuring starts to breakdown at a lower temperature for the smaller than for the larger particles, showingthat the 0.13 p particles order water to a lesser extent, in spite of having lowerflocculation rates. Since the repulsion potential due to water structuring mustoppose the London-van der Waals attractive potential, which is directly propor-tional to the radius of the particles, then for the same flocculation rates of the twoparticle sizes the water structuring must be considerably greater for the 0-8 pparticles. A lower flocculation rate for the 0.13 p particles means that the radialdependence of the water structuring potential obeys a lower power law than theLondon-van der Waals potential, giving rise to a higher net energy barrier for th126 WATER STRUCTURE AND COLLOID STABILITYI 1-3 3 .2 3,4 3-61/Tx 103 in OKFIG. % T h e dependence of the relative flocculation rate R for the 0.1 3p PVA particles on the absolutetemperature T in glycerol+ water solutions. Continuous lines are at constant composition ; brokenlines are at constant viscosity.1.4I .;I-ch 0.e\ gm E:M 0.6 -0.40.2V0.6 0.8 1.0 1'2log -fIFIG. 6.-The dependence of the ratio of the relative flocculation rate R to the absolute temperature Ton the viscosity q for the 0 . 1 3 ~ PVA particles in water and glycerol+water solutions.x, water0, glycerol- -, theoreticaJOHNSON, LECCHINI, SMITH, CLIFFORD AND PETHICA 127smaller particles.However, both the magnitude and the distance over which therepulsive potential due to water structure extends are greater for the larger particles.That the whole phenomenon of water structuring is one of general importancein understanding colloid stability and is not merely reflected in the flocculationbehaviour of polymer lattices, is shown by the octadecanol curve in fig. 4 whichdisplays the same characteristic anomalous slope as the smaller PVA particles.A further test of this theory lies in investigating whether the fastest flocculationrate at a particular viscosity can be decreased even more than for that in water byusing other structured liquids. For this purpose, fastest flocculation rates for 0.13 11PVA particles were measured in a number of glycerol+water mixtures at varioustemperatures.The results are plotted in fig. 5 in two ways. First, the fastest ratesare given as a function of temperature for a range of constant glycerol concentra-tions. The rates at these concentrations show the same apparent activation energyfor flocculation as for the sol in water. This behaviour is to be expected sincethe activation energy for diffusion, as judged by the temperature dependence of theviscosity,25 is almost identical in this range of solutions to that for water. Secondly,flocculation rates at constant viscosity (i.e., the glycerol concentration is changing)are shown as a function of temperature. Above a certain glycerol concentrationthe flocculation rates become independent of temperature.In fig. 6 these tem-perature independent rates are replotted as a function of viscosity. Also shownare the fastest flocculation rates in water and the theoretical slope expected fromSmoluchowski theory. The fastest floccuIation rates in glycerol +water mixturesshow even larger deviations from simple theory than for water alone, suggestinggreater structure promotion and consequent stabilization.N . M . R . RELAXATION MEASUREMENTSTo obtain experimentally significant changes in the relaxation times of waterprotons for a PVA sol it is necessary to work at high particle concentrations, i.e., - 1010 particles per ml for the 0.8 p PVA, and N 1012 particles per ml for the 0.13PVA. All the initial measurements of relaxation rates were carried out at theseparticle concentrations on dialyzed stable dispersions.The structured water can be regarded as composed of a number of regions oflower molecular mobilities and correspondingly different nuclear magnetic relaxa-tion times to bulk water.When the lifetime in any of these regions is short com-pared to the relaxation time (i.e., fast averaging prevails) the overall relaxationprocess is characterized by one mean relaxation time, which is the inverse of theweighted average relaxation rate.19 Experimentally the transverse and longitudinalmagnetizations always changed exponentially, showing that fast averaging pre-vailed at all times.The T2 results obtained for the 0.8 p PVA particles are shown in fig.7 as afunction of temperature. At any given temperature, T2 values are lower than forbulk water and a similar break appears in the relaxation time plot at 30-35"C, aswas observed with the flocculation data. The steeper part of the PVA curve may beexplained by assuming that at temperatures below 30°C the extra ordering of watercaused by the sol is increasing with decrease in temperature. At a still lower tem-perature of 13°C the slope reverts back to the value it had above 35"C, indicatingthat the increase in order has reached a limiting value.This extra structuring below 30°C as revealed by Tz measurements cannot bean effective barrier to diffusion because the flocculation data for the 0.8 p PVAparticles in this temperature range shows a temperature dependence close to tha128 WATER STRUCTURE AND COLLOID STABILITYexpected for a diffusion-controlled process in bulk water.Because n.m.r. theorywhen interpreting a T2 value cannot distinguish between small amounts of highlyordered water and large volumes of less ordered water (and all intermediate cases),it is possible that the structured water below 30°C will belong to the former category,and could be associated with charged surface groups of the polymer particle.Evidence for structure breakdown above 35°C (which the flocculation data suggest)appear to be forthcoming when the flatter part of the T2 plot for the sol at the higher6.0 IL-'2 9 3.0 3.1 3.2 3.3 3'4 3.5 3.6l l T x 103 in OKFIG. 7.-The dependence of the spin-spin relaxation time TZ on the absolute temperature T for the0 .8 ~ PVA particles and for " non-degassed " water.temperatures is compared with that of " non-degassed water " (i.e., water in equi-librium with atmospheric oxygen). The two curves over this temperature rangeare converging, suggesting a gradual breakdown in water structure with increase intemperature. However, the slope of the T2 plot for non-degassed water is low,when compared to that of TI for degassed water? On degassing the water, T2showed the same dependence on temperature as T1.21 Moreover, the high-tem-perature part of the T2 curve for the sol gives an apparent activation energy equalto that for degassed bulk water. This observation led us to investigate the possi-bility of oxygen being removed from the water as the concentration of the sol wasincreased.By testing for oxygen colorirnetrically and polarographically we foundthis to be the case, oxygen in solution being barely detectable at high particle con-centrations. The low slope for non-degassed water can be explained by an addi-tional T2 relaxation mechanism involving a hyperfine interaction with the unpairedelectron in molecular oxygen. This relaxation process must have a dif€erenttemperature dependence to that of nuclear dipole relaxation. A second contributioJOHNSON, LECCHINI, SMITH, CLIFFORD AND PETHICA 129to the effect of temperature on T2 for non-degassed water will arise from the changein oxygen solubility with temperature. This contribution would tend to give asmall increase in the apparent activation energy.The n.m.r.results presented so far, although providing strong evidence forstructured water, do not give a good fit with the flocculation data. Indeed, sincethe T2 values for the 0.13 p particles show the same breaks as the 0.8 p particles,merely shifted to slightly lower temperatures (- 5°C lower), these discontinuitiesare not particularly relevant in explaining the flocculation data. This unsatis-factory correlation probably arises because in the n.m.r. experiments the processof flocculation is not simulated, the particles being simply considered in isolation,I5.0 0 0 10.000 20,000separation between particles in 8,FIG. 8.-The dependence of the spin-spin relaxation rate 1/Tz on the interparticle separation for the043~ PVA particles at 2, 12, 23, 33, 45°C.x, 2 "C +, 12°C 0,23"C V, 33°C D,45"Cwhereas during the flocculation process, the particles approach each other to smalldistances. It was impossible to follow the flocculation process by n.m.r., becausethe particle concentration has to be large enough to allow detection of changes inrelaxation times due to structured water, so precluding the possibility of a measurablyslow flocculation.However, this difficulty could be overcome in part by measuringthe relaxation rate as a function of particle concentration in a stable system, andincreasing this concentration until the mean separation between particle surfacesbecame less than lOOOA. In fact, when the transverse relaxation rate 1/T2 wasplotted against Concentration for 0.8 p PVA particles at a series of different tem-peratures, the relaxation rate increased rapidly at high particle concentrations.This effect is shown in fig.8 where the relaxation rate is plotted against the meanparticle separation. Another important characteristic of these plots is the markedtemperature dependence of the relaxation rate at high particle concentrations.Further, the results also show that the solvation energy barrier is constant at lowtemperatures, breaking down and presumably disappearing at high enough tem-peratures. This is exactly the model we postulated to explain the flocculation data130 WATER STRUCTURE AND COLLOID STABILITYfor the 0.8 p PVA particles. The same effect can be seen more clearly when the datashown in fig.8 is plotted as a function of temperature at different particle concentra-tions (fig. 9). The slope of the log T2 against 1/T plot at the higher temperatures,for concentrations above 1012 particles per ml is greater than for degassed water.To check that TI gave the same pattern as the T2, the above experiments wererepeated, with TI being measured as a function of concentration at a series of differenttemperatures (fig. 10). Unfortunately, we could not measure TI for exactly thesame temperatures at which we had measured T2, since we were using adiabatic0 . 0 6 L L L -2 9 3.0 3.1 3-2 3.3 3.4 3.5 3 . 61/Tx lo3 in O KFIG. 9.-The dependence of the spin-spin relaxation time T2 on temperature Tas a function of particleconcentration for PVA 0-13p particles.(a) 5 x 1010 part/ml ; (b) 1 x 10 11 partlml ; (c) 4 x 10 11 partlml ; (d) 1 x 10 12 part/ml;(e)l.5 x 10 12 part/mlrapid passage techniques on a spectrometer which is not capable of the same degreeof flexibility in thermostatting as the spin-echo instrument.The TI relaxation timesas a function of particle concentration show the same pattern as the T2 results,except that values of TI and T2 at any given particle concentration can be markedlydifferent, even though the BPP theory does not predict this behaviour under con-ditions of extreme line narrowing and in the absence of any additional relaxationmechanism. (In fact, there are minor differences even for bulk water between valuesof l/Tl and 1/T2 at fixed temperatures because of the scalar coupling between thespins of the proton and that of 017, present in small amounts in water, modulatedby chemical exchange.22) The differences we observe between TI and T2 for the solare probably due to more specific surface effects, whose nature is not understood,but which have been reported in the literature.239 2JOHNSON, LECCHINI, SMITH, CLIFFORD AND PETHICAL I I 10 5,000 IQOOO lS,OOO 2c )OO131separation between particles in 8,FiG.10.-The dependence of the spin-lattice relaxation rate l/T1 on the interparticle separation forthe 0.81~. PVA particles at 13, 21, 30, 39°C.x, 13°C +21"C 0, 30°C V, 39°CI I I I 1 1 I I II1000 moo 3000 4000 5000 6000 7000 eooo 9000 11separation between particles in 8,I 000FIG.11.-The dependence of the spin-spin relaxation rate 1/T2 on the inter-particle separation forthe 0 . 1 3 ~ PVA particles at 2, 13, 24, 34, 46°C.x, 2°C I-, 13°C 0, 24°C A, 34°C V, 46°3 32 WATER STRUCTURE AND COLLOID STABILITYIt was suggested earlier that in this type of n.m.r. experiment the approach oftwo particles to distances where they are known to interact strongly is being observed.Water structuring at these distances must result either from overlap of electricaldouble layers or from the interactions due to London-van der Waals dispersionforces. The importance of the electrical double-layer interaction can be tested bythe addition of electrolyte, which results in a decrease of the double-layer thickness.T2 was measured at high particle concentrations for the 0-8 p PVA particles in0.001 M sodium chloride.This concentration of sodium chloride was chosen asit would have a large effect on the electrical double layer but would not produce anexperimentally noticeable change in the T2 value of bulk water. T2 for the soldecreased slightly, giving an effect directly opposite to what would be expected ifdouble-layer overlap was the cause of water structuring. This result suggests thatthe electrical interactions disrupt structure formation.Nr2X 1018FIG. 12.-The dependence of the spin-spin relaxation rate 1/T2 on the particle area per ml of suspensionfor the 0.13 and 08p particles at 2"C, where N is the number of particles per ml of suspension and ris the particle radius.If the effect is associated with dispersion forces it should increase with particleradius.This can be seen when the T2 data for the 0.13 p PVA particles (fig. 11)are compared with that for the 0.8 p particles. The magnitude of the relaxationrate, for a given particle separation, is greater for the larger particles. The re-laxation rate also starts to increase rapidly for the 0.8 p particles at an interparticledistance which is almost twice that for the 0.13 p particles. An analysis of theT2 data for the two particle sizes is in complete agreement with the model put forwardto explain the flocculation data. Thus, (i) the total amount of structured waterper unit surface area is greater for the larger particles, as shown in fig.12. Therelaxation rate compared to the particle volume in unit volume of suspension showsthe same trend. (ii) The dependence of water structuring on particle radius asmeasured by the ratio of the relaxation rates for the two PVA particles at a fixedseparation and temperature obeys a lower power law to that given by the London-van der Waals potential between particles. (iii) The relaxation rate for the smallerparticles shows a greater temperature dependence at the lower temperatures. ThJOHNSON, LECCHINI, SMITH, CLIFFORD AND PETHICA 133evidence presented here generally supports the views of Derjaguin on the long-rangestructuring of boundary layers of water.26We thank our colleagues, Dr. C . Smart, Dr. K. E. Lewis and Dr. E. Willis forhelpful discussions.1 Johnson, Goldfarb and Pethica, Trans. Faraday SOC., 1965,61,2321.2 Fawcett, Parfitt and Smith, Nature, 1964, 204, 775.3 Hahn, Physic. Rev., 1950, 80, 580.4 Pople, Schneider and Bernstein, High Resolution Nuclear Magnetic Resonance, (McGraw Hill,5 Von Smoluchowski, 2. physik Chem., 1917,92, 129.6 Oster, J. Colloid. Sci., 1947, 2, 291.7 Wachtel and La Mer, J. Colloid. Sci., 1962, 17, 531.8 Williams, Ann. N.Y. Acad. Sci., 1958, 70, 763.9 Bloembergen, Purcell and Pound, Physic. Reu., 1948, 73, 679.10 Daszkiewicz, Hennel, Lubas and Szczepkowski, Nature, 1963,200, 1006.11 Hollingshead, Johnson and Pethica, Trans. Faraday SOC., 1965,61, 577.12 Brice, Halwer and Speiser, J. Opt. SOC. Amer., 1950, 40, 768.13 Overbeek in Kruyt, Colloid Sci. (Elsevier, 1952), 1, 278.14 Carr and Purcell, Physic. Rev., 1954, 94, 630.15 Gill and Meiboom, Rev. Sci. Instr., 1958, 29,688.15 Clifford and Pethica, Truns. Farday SOC., 1965,61, 182.17 Ermakov, Zagarets and Grunan, Russ. J. Physic. Chem., 1964, 38, 566.18 Fuchs, Mechanics of Aerosols (Pergamon, London, 1964).19 Zimmerman and Brittin, J. Physic. Chern., 1957, 61, 1328.20 Simpson and Carr, Physic. Rev., 1958, 111, 1201.21 Lecchini and Smith, unpublished results.22 Meiboom, Bull. Amer. Physic. SOC., 1960, 5, 176.23 Woessner and Zimmerman, J. Physic. Chern., 1963, 67, 1590.24 Resing, J. Chem. Physics, 1965, 43, 669.25 Handbook of Chemistry and Physics (Chemical Rubber Publishing Co., Ohio, U.S.A., 1962-3).26Derjaguin, SOC. Expt. Biot. Symp., 1964, 19, 55.New York, 1959)

ISSN:0366-9033    

DOI:10.1039/DF9664200120

出版商:RSC    

年代:1966

数据来源: RSC

摘要:

GENERAL DISCUSSIONMiss C. A. Gilchrist, Mr. J. Rogers, Dr. E. G. Vaal and Dr. P. A. Winsor (“Shell”Research Centre, Chester) (communicated) : In connection with the “ structuring ’’of water next to hydrophilic surfaces we record some observations made on highresolution n.m.r. spectra of the N,N,N-trimethylaminododecanoimide (CJ 1H23 . CO .N- . N+(CH3)3)+ water system, studied by X-ray methods by Clunie, Corkill andx SPECTRC SHOWING HYDROCARBON GROUP PECKS i 1 0 SPECTRA NOT SYOWING HYCROCARRON GROUP P E A GI B POINTS SHOWING TRCNSITION OF SPECTRAL TYPESx 95 1601 50 -II 40,-50,-II ).. @--&L-. &----€i10 20 30 40 50 60 70 80 SO 1C.CIMIDE IN WATER, %W0X0OPTICAL IDENS17X-RAY(b)S___._I-_ 1 - - 1 - 1 L _I_- 1 L LIU 20 30 40 50 60 70 80 90 100imide in water, % wFIG.1.-Phase diagrams for imide/water systems, (a) from NMR spectra (b) after Clunie, Corkilland Goodman.1Goodman.1 The phase diagram for this system obtained by these authors and thatdetermined from high resolution n.m.r. spectra are shown in fig. 1. The structureof the ‘‘ neat ” or lamellar phase G may be regarded as constituted from an infinitenumber of indefinitely extended planar aqueous foam films piled one upon the otherwith elimination of the vapour phase.19 2 The detailed structures of the viscous iso-tropic and middle phases are less certain.19 3 Although we made our n.m.r. measure-ments primarily from interest in this aspect, certain evidence concerning the relativemobility of water molecules in relation to their distance from the hydrophilic surfaces1 Clunie, Corkill and Goodman, Proc.Roy. Soc.,A, 1965, 285.2 Luzzati and Husson, J. Cell. B i d , 1962, 12, 207.3 Luzzati and Reiss-Husson, Nature, 1966, 210, 1350,13GENERAL DISCUSSION 135Our results are summarized inHigh resolution n.m.r. spectra were observed corresponding to the protons of theimide molecules in the mobile isotropic phase S, in the viscous isotropic, " cubic ",liquid crystalline phase V1 and in the molten imide but not in the middle MI, neat G,or solid imide phases. This distinction enables a phase diagram to be drawn on thebasis of n.m.r. spectra as shown in fig. 1. Since we made these observations similarn.m.r. results have been published by Lawson and Flautt 1 and also by Zlochower andSchulman 2 for other amphiphile + water systems.An interesting point, which does not appear to have been noted in this work,concerns the high resolution n.m.r.spectra of the water protons in the imide + watersystem. High resolution spectra due to the water protons appear clearly for the Sand V1 phases but are absent for the G phase at high concentrations ( > 75-80 % imide).of the bimolecular leaflets of imide was obtained.fig. 2.VAIDE, E O TI.- % w WATER 0WATER, % looII65 15 60 40 50 59FIG. 2.-Block diagram showing typical high resolution n.m.r. spectra in the various phase regionsof the N,N,N-trimethylaminododecanoimide + water system.They appear at lower concentrations with the G phase, and with the M1 phase, butwith diminished intensity in comparison with those shown by the S and V1 phases.The absence of the high resolution spectra corresponding to the imide protons withthe MI, G and solid imide phases and their presence with the S and V1 phases appearsto indicate a much greater limitation of rotary Brownian movement of the imidemolecules in the MI, G and solid phases than in the S and V1 phases.1Similarly? the absence of high resolution spectra corresponding to the waterprotons at high imide concentrations with the G phase appears to indicate that thewater molecules in close proximity to the immobile bimolecular imide leaflets areimmobilized by the '' structuring " effect of the hydrophilic surface.With furtheraddition of water and expansion of the water layers between the bimolecular leaflets?the water molecules further from the hydrophilic surface are less under its influenceand are able to undergo rotary Brownian motion and consequently to give a highresolution spectrum.1 Lawson and Flautt, Molecular Crystals, 1966, 1, 241.2 Zlochower and Schulman, Division of Colloid and Surface Chemistry, Amer.Chem. SOC. Abstr,152nd Meeting, 1966136 GENERAL DISCUSSIONIn the S and V1 phases the free rotation of the imide molecules (probably, however,still aggregated as micelles of limited size) will be accompanied by free rotation of anyassociated “ structured ” water and high resolution spectra for water protons as wellas for imide protons are obtained. In the MI phase the water molecules close to thestable, infinitely extended cylindrical micelles 2 would be expected to be immobilizedwhile those more remote from the micellar surfaces might be expected to undergo freerotary Brownian motion.These expectations accord with a comparison of the n.m.r.spectra for the S , MI and V1 phases. When our observations are taken in conjunctionwith the micellar dimensions of the G phase calculated by Clunie, Corkhill and Good-man,1 the thickness of the immobilized water layer at 80 % concentration (mole/ratioH2O/imide = 2.9/1) is about 3A at each imide/water junction, i.e. 6L$ within the‘‘ sandwich ’’ as a whole.Dr. K. J. Mysels (R. J. Reynolds, Winston-Salem) (communicated) : A feature ofthe phase diagram for the imide +water system of Clunie et al.was the absence of anytwo-phase region separating the one-phase regions of this two-components system.Apparently, the authors were unable to detect these two-phase regions normallyrequired by the phase rule, despite considerable effort. The phase diagram based onn.m.r. spectra for the same system now shown by Gilchrist et al. in fig. 1 includes“points showing transition of spectral types” which are identified as correspondingto two-phase systems in fig. 2. It would seem therefore that the phase boundariesshould be drawn on both sides of these points rather than through them. Theseresults also confirm that the two-phase regions are narrow and therefore not easy todetect by methods other than n.m.r.Prof. D. D. Eley (Nottinghdm) (partly conzmunicated) : When two colloidal particlessurrounded by ice-like layers approach each other it is possible that the dipole-dipoleforces between the water molecules in the two opposing ice surfaces are repulsive andtend to cancel out the attractive van der Waals forces.This is one way in which ice-like layers might lead to stabilization of colloid particles. For oriented rigid layers oforganic molecules, would not the anisotropy of polarizability effect, mentioned earlierby Fowkes, lead to the possibility of repulsive effects between two approaching colloidparticles ?Prof. J. T. G. Overbeek (van’t HoflLabs., Utrecht) said: It is true that two layersof dipoles with opposite orientation would repel one another. However, the range ofsuch a repulsion is not larger than the lateral distance between the dipoles in one layerand therefore in general would not be expected to be strong enough to cancel out thevan der Waals attraction between colloid particles, unless the particles are very small.The anisotropy of the polarizability might indeed lead to a repulsion, but only in asituation that is not symmetric with respect to the two particles, e.g., if the layers havea different orientation on each of the particles.Dr.P. C. Scholten (Philips Res. Lab., Eindhoven) said: The vibrations in piezo-electric quartz crystal will in general not be restricted to one dimension only. InBazaron’s experiments a vibration in vertical direction (thickness), however, wouldlead to a compression-expansion of the water layer between the two quartz crystals.This would also cause an increase of the resonance frequency.Is Derjaguin sure thatsuch vibrations were small enough to be ineffective?Dr. P. C. Scholten (Philips Res. Lab., Eindhoven) said: When a vertical soap filmis pulled from a solution, its thickness depends on the velocity of pull-out. Mysels1 Luzzati and Husson, J. Cell. Biol., 1962, 12, 207GENERAL DISCUSSION 137et al. found that the thicknesses obtained agree quantitatively to within 20A with ahydrodynamic calculation by Frankel. Derjaguin remarked that in these experimentsa layer of increased viscosity along the surface would not necessarily affect the thicknessof the film pulled out, as this thickness is not determined by slow draining, but by afast film formation process in the transition zone between bulk and film.The forma-tion of the viscous layer could be a slow process, lagging behind the film formation.Now in Frankel’s calculation, the surface of the transition zone between film and bulksolution was assumed to be inextensible and the surface needed for the new filmthought to be produced by expansion of the bulk surface. In the experiments, thebulk surface was often aged for hours and sometimes days. If this surface had ahighly viscous layer, this would certainly be dragged along through the transition zoneinto the film and make it thicker. If, on the other hand, the bulk surface remainedstationary and the new surface were formed in the transition zone, the film thicknesswould be smaller than calculated.As no large deviations, other than those due tovan der Waals and double layer interaction forces, were observed, viscous surfacelayers are unlikely to be present in these systems. However, the nature of a soapsolution/air interface is different from that of the silica/or glass/solution interface nearwhich the viscous layers described by Derjaguin were observed.Prof. B. V. Derjaguin (Acud. Sci., Moscow) said: In reply to Scholten, in ourexperiments we used a quartz block cut from a monocrystal in such a manner that thenormal (vertical) component of vibrational displacements was very small. Theresonance frequency shift resulting from this normal component is independent of thethickness h and the nature of the liquid film introduced between the surface of the blockand the quartz bar.In addition, this frequency shift can be determined from theexperimental resonance shifts corresponding to high values of 62 and used for correc-tions. The “ corrected ” shift values, depending only on the shear elasticity of theliquid films, tend to zero at sufficient high h-values. For h-values of about 1 p thisshift correction is small2I am familiar with Frankel’s calculations as I published almost the same calcula-tions 3 some years previously. Thus, hydrodynamic analysis shows that the film isformed by the liquid being dragged not only from surface layers but also from thebulk. Therefore the resulting film thickness depends on a certain mean viscosity andthe input of the surface viscosity cannot be estimated on the basis of Scholten’s purelyqualitative arguments.Dr.G. D. Parfitt (University of Nottingham) said: The question of the existence ofpolymolecular solvation layers around colloidal particles and the role they play inthe stabilization of dispersions arouses much debate. Derjaguin’s experiments offerconvincing evidence that thick boundary layers of polar liquids exist near glass andquartz surfaces, but up to now there are insufficient stability and other relevant dataavailable for the effect of these layers on coagulation behaviour to be put on a soundtheoretical basis. Given the right conditions it is probable that such layers can existin non-polar systems. To explain the stability of dispersions of Graphon in alkyl-benzene + heptane solutions we have postulated the existence of an oriented layer (ofthe liquid crystalline type) of alkylbenzene molecules lying parallel to the surface andextending some distance from it.4 It seems reasonable to suppose that such condi-tions might prevail in other systems in which non-polar surface active agents act asstabilizers for powders in hydrocarbon media.1 Lyklema, Scholten and Mysels, J.Physic. Chem, 1965, 69, 116.2 see also JEPT, 1966, N10.3 DokIady Akad. Nauk. S.S.S.R., 1943, 39, 11.4 Parfitt and Willis, J. Colloid Interface Sci., 1966, 22, 100138 GENERAL DISCUSSIONProf. B. V. Derjaguin (Acad. Sci., Moscow) said: In reply to Parfitt, non-polarliquids (e.g., benzene) can also form boundary layers with specific properties under theinfluence of adsorbed monolayers of some polar compounds of similar structure (e.g.,nitro benzene).Dr.K. J. Mysels (R. J. Reynolds, Winston-Salem) said: A puzzling feature ofDerjaguin’s description of the behaviour of the anomalous “ orthowater ” is itsfreezing behaviour which seems to be reversible yet spread over the - 30 to - 15°Crange. As this is not in accordance with the phase rule for an equilibrium one-component system I would like to ask him how this should be interpreted.]Prof. B. V. Derjaguin (Acad. Sci., Moscow) said: In reply to Mysels the anomalouswater columns diverge in many respects from the behaviour of one-componentsystem (not only below zero temperature, but also above).Therefore there is noquestion of standard application of the phase-rule to anomalous water. I would likefor the present to avoid the formulation of some hypothesis which would explain thetwo-component behaviour of anomalous water. Certainly, trivial explanations arenot possible.Prof. D. D. Eley (Nottingham) (partly communicated): I would ask Derjaguinwhether it is possible for him to express the elastic energy of shear in his rigid liquidlayers in molecular terms, e.g., in terms of ergs per deg. of angle per intermolecularbond, corresponding to the molecular picture of shear in terms of the bending ofintermolecular bonds. With these figures one might be able to estimate whether theenergy terms can arise from dipolar or hydrogen bond forces, each of which mightlead to orientation over several molecular layers, e.g., some 20L$ or so.Again, afirst layer of non-polar molecules may acquire polarity by charge-transfer adsorptionto a solid surface and induce dipoles over two or three layers of molecules.Derjaguin and Clifford inform me, however, that orientation effects maybe occurover 1,000 L$ between two opposing surfaces, and that one surface alone is not sufficientto secure these long-range orientation effects. If this is so, we are dealing with aninteresting new phenomenon.Prof. B. V. Derjaguin (Acad. Sci., Moscow) said: In reply to Eley, our results havea double significance : (i) They show the existence in the bulk of all liquids of a modulusof shear elasticity. The corresponding bulk value of the modulus is independent of thespecial properties of boundary layers and must be interpreted in some way similar tothat used in the theory of elasticity of solid bodies (especially amorphous).Suchinterpretation should first of all take into account the Born repulsion forces. Theattractive, e.g. van der Waals, forces are of prime importance in the theory of shear-strength. (ii) Our measurements show that the modulus of shear-elasticity of polarliquids (water, alcohols, etc.) increases in the vicinity of quartz surface (starting atsome distance of about 700-8OOA). This inference has been made on the basis ofexperiments with films which in no case allowed an overlapping of two boundarylayers. The film thicknesses always surpass 2000 A.Prof.B. TeBak (Zagreb, Yugoslavia) said: With regard to “ usual water ”, meta-water and the anomalous “ orthowater ” for the methoric structures of coagulatingsystems in electrolytic media, the characteristics of the boundary region of the solidphase with the interacting constituents of the solvent and the solute show transitionalstructure which is different from either the bulk solution or the pure solvent. In thissense, the experimental results of Johnson et al., are far from conclusive ; thus, theextrapolation of the relaxation rate as a function of the particle concentration in astable system to that of a flocculating system requires justificationGENERAL DISCUSSION 139Mr. J. Clifford (Unilever Res. Lab., Port Sunlight) said: At present we are not ableto measure nuclear magnetic resonance relaxation times in flocculating systems.Wewould agree that the fact that anomalous water molecule mobility in stable systems andanomalous flocculation rates are both observed for the same sol is not in itself adequatejustification for assuming that the two phenomena are related, as an assumption thatthe properties of water are essentially similar in the stable and the flocculating systemsis involved. However, our paper contains two experimental observations which makeit reasonable to postulate that anomalous water properties and anomalous floccula-tion rates are related. These are the effect of the addition of electrolyte and the effect0 5 ~ 1 0 ' ' I * d ~number of particles per cm3 0 5x10'' l*lo'2FIG.1.-The dependence of the spin-spin relaxation rate 1/T2 on particleconcentration for the 0.8 u PVA par-ticles at 13, 21, 29 and 39°C.number of particles per cm3FIG. 2.-The dependence of the spin-lattice re-laxation rate 1/T1 on particle concentration forthe 0.8 p PVA particles at 13,21,29 and 39°C.of temperature on the results. The only difference between the stable and the flocculat-ing system is in the addition of electrolyte. The addition of enough electrolyte to havea large effect on the electrical double layers in the system has only a small effect on then.m.r. results, reducing molecular mobility slightly but not altering the general effectof particle concentrations on water proton relaxation rates.The flocculation rate measurements show that at low temperatures ( N 10°C) abarrier to flocculation exists and that this is reduced as the temperature is increased.The n.m.r.measurements show that the effect of the colloidal particles on the mobilityof water molecules varies with temperature in the same way. The measured relaxa-tion rate of water proton should vary linearly with the concentration of particlesprovided that the particles affect water independently. In fact, the relationship islinear only up to a certain concentration above which a cooperative effect becomesapparent. This is shown in fig. 1,2 and 3. The limiting concentration increases, andthe extent of the cooperative effect decreases, with increase in temperature. Thi140 GENERAL DISCUSSIONindicates that the effect of the colloidal particles in reducing the mobility of watermolecules decreases with temperature over the same temperature range that the barrierto flocculation decreases.FIG.3.-The dependence of the spin-spin relaxationrate 1 IT, on particle concentration for the 0.1 3 IJ. PVAparticles at 2, 13, 29, 39 and 46°C.I_-.number of particles per cm3Prof. J. Lyklema (Agric. Univ., Wageningen) said: In his study of anomalouswater structures, Derjaguin suggested that hydrophilic groups like -OH groups onthe silica or quartz surface would act as a kind of nucleus for the epitaxal growth of anumber of specially structurized water layers. As the zero point of charge is low thesurface of a hydrophilic silica in contact with water of about pH = 7 is negativelycharged, the charge becoming progressively more negative with increasing pH.Ascharged groups on the surface tend to orient the adjacent water dipoles one mightexpect a strong influence of the pH of the water on the structural properties ofadsorbed layers.Dr. F. M. Fowkes (Sprugue Electric Co., North Adarns, Mass.) said: The unusualwater behaviour reported by Derjaguin prompts me to mention some studies I madein 1949. Water-saturated hydrocarbons were cooled in a well-stirred completelyenclosed system with a built-in light-scattering apparatus. On cooling, water dropletsappeared at about 15°C and the forward (- 30") light-scattering of these droplets wasmeasured on cooling to - 80°C and back to + 20°C (in about 3 h).A sharp increasein scattering occurred at - 20 to - 3 1 "C (depending on the hydrocarbon present), butno further change was seen on cooling to -80°C. Warming produced a sharpdecrease at the identical temperature (k 0*3"), but no change in light-scattering wasever observed on warming through 0°C. This could be interpreted now as a manifes-tation of " specific " water, freezing and melting at - 20 to - 31" and never freezing at0°C. However, the dependence of the " freezing " temperature on the kind of hydro-carbon is difficult to explain without invoking hydrocarbon + water '' complexes ".Prof. G. M. Bell (Chelsea College, University of London) and Dr. S . Levine(Manchester University) said : We have examined the nature of the electrical forcGENERAL DISCUSSION 141between a pair of parallel uncharged plates of infinite thickness and low dielectricconstant, immersed in an aqueous 1-1 electrolyte.The results should have somebearing on the suggestion of Johnson et al. that an unexplained repulsive potentialexists between uncharged colloidal particles. The force we have calculated is causedby the ion-image self-atmosphere potentials discussed in our paper on the modifiedPoisson-Boltzmann equation. For anions and cations of the same size and polariza-bility in the 1-1 electrolyte, the image-self-atmosphere potentials of the two ionspecies are equal and consequently with uncharged plates there are no mean volumecharge density and no cavity potentials. We have so far only considered point-ionsand therefore neglected the exclusion volume of ions.It is convenient to discuss theFIG. 1.force P between the plates rather than the free energy of interaction.notation as in our paper, at the plate separation 2h,Using the sameThe first and second terms are the usual difference between the osmotic pressure at themedian plane and that in the electrolyte bulk. For plates of low dielectric constant,ions are repelled from the plates and the ion concentrations, and therefore the osmoticpressure, are lower at the median plane than in the bulk, giving an attractive forceterm. This pressure difference is commonly regarded as the only contribution, butwe have a second term in the form of an integral, which seems new. It arises becausethe local free energy density f&( l), due to the image-self-atmosphere potential,depends explicitly on the positions of the plates (x = +h) and not merely implicitlythrough local variables, such as the ion concentrations.This causes an extra repulsiveterm due directly to ion-image forces. We observe that the integrand is the derivativewith respect to the position x1 = - h of the far plate, of the function ~ D H ( 1) which isevaluated at position x on the side (0,h) of the median plane. It is found that theattractive term predominates at small separations and the repulsive term at large ones142 GENERAL DISCUSSIONThis is illustrated by curve I in the figure, which gives P (in units of 103 dyneslcmz ofplate area) as a function of K(h-d) for an electrolyte concentration 0.01 M ; IC isthe Debye-Huckel parameter and d = 2 A allows for distance of nearest approach ofthe ion to the plate wall.The dielectric constants of the aqueous and plate mediaare EO = 78.3 and E~ = 2.5 respectively. For comparison, curve I1 shows the Hamaker-van der Waals attractive force for infinitely thick plates, with constant A = 10-14erg cm.2 At large separations, the force P behaves asymptotically as llh3, i.e., as arepulsive van der Waals force. It is convenient to express this as B/48nh3, where Bis a power series in the electrostatic reflection coefficient f = (EO - E ~ ) / ( E O + E ~ ) ; for thevalues of EO and E~ in the figure, Bx2.9 x 10-14 erg cm2. (Strictly, our calculation ofthe image forces is valid for E ~ / E O = 0.) The constant B is independent of electrolyteconcentration. We have verified this by computing P at 0.1 M, although in this casethe neglect of ion size makes the result less reliable.It appears therefore, for valuesof A<B, the electrical repulsion due to ion-image forces is sufficient to overcome thevan der Waals attraction.Dr. G. A. Johnson (Unilever Research) said: In agreement with Bell and Levine itappears as though this proposed ion image force is insufficient to dominate van derWaals attractive forces for the systems which we have studied. In some of ourexperiments with 1-1 electrolytes, concentrations greater than the 0-25 M requiredto produce rapid coagulation were used which would give ion-image forces muchsmaller than those shown in Bell and Levine’s figure and hence these forces will notproduce a significant decrease in the rate of fast coagulation.Dr. G.D. Parfitt (University of Nottingham) said: The Coulter counter is provingto be a valuable instrument for the investigation of particle size distribution in disper-sions of particles whose effective diameters are larger than - 5 p. However, for smallerparticles and particularly in the sub-micron range, the investigation of which requiresthe use of small diameter orifice tubes, there is increasing evidence that the Coulterdata may lead to incorrect results. Cooper and I have made a detailed comparison ofparticle size distributions obtained with the Coulter counter model A and by electronmicroscopy, for dispersions of a series of Dow monodisperse latices of diameters inthe range 0.8-3-5 p. Using a 30 p orifice tube the Coulter counter gave distributionswhich are considerably wider than those from electron microscopy and show asignificant positive skew, i.e., towards larger particle size. Experiments with differentorifice tubes of nominally the same diameter and with different instruments gavesimilar results, and it was also shown that the skew was not a result of coagulation.Furthermore, on successive dilution of a latex dispersion the cumulative numbercounts over the complete size range were not simply related to those of the initialdispersion, deviations as large as 20 % being observed. It is concluded that theanomalous results are associated with the geometry of the small orifice tubes anddemonstrate that caution must be exercised in the interpretation of particle countsobtained for a coagulating dispersion.Dr. G. A. Johnson (Unilever Research) said: In reply to Parfitt, we have measuredthe variation of the total count with dilution for the P.V.T. latex system and found alinear dependence, within 3 %. In connection with other studies on P.V.T. latices,Willis and I have taken the initial particle size distribution and the coagulation ratedetermined from Coulter counter measurements and have computed from Smolu-chowski’s equations the particle size distributions after various times up to 50 % ofthe half-life of the system. These computed distributions were in good agreementwith those measured experimentally on the Coulter counter after the same periodsof coagulation

ISSN:0366-9033    

DOI:10.1039/DF9664200134

出版商:RSC    

年代:1966

数据来源: RSC

摘要:

hteractions between Spherical Particles of MonodispersePolystyrene LaticesBY A. WATILLON AND A.-M. JOSEPH-PETITUniversitC Libre de Bruxelles, Service de Chimie Analytique et Minkale,50 av. F. D. Roosevelt, Bruxelles 5, BelgiumReceived 6th June, 1966Reliable results concerning the interaction of lyophobic colloidal particles can only be obtainedfrom data concerning spherical and highly monodisperse systems. Polystyrene latices seem to beadequate. Coagulation kinetics gives the coagulation value, for which the Van der Waals attractioncounterbalances approximately the Coulomb repulsion. Furthermore from the electrophoreticmobilities, we calculated the zeta potentials which we show to be a reasonable approximation for thepotential governing the Coulomb repulsion. We considered also the effect of a viscoelectric correc-tion.Thus we obtained repulsion energies. Finally, we calculated under different conditions theprobable value of the Hamaker constant of polystyrene embedded in water and we compared theo-retical and experimental data.The curves giving the relation between the interaction energy and the distanceof two approaching particles can be obtained by combining a positive contributiondue to the double layer electrostatic repulsions and a negative one due to London-Van der Waals attractions. The latter always prevails at short distances. Therepulsion energies are sensitive to the nature and the concentration of the electro-lytes present in the liquid phase. Consequently, it is possible to find for a definitesystem a critical concentration of electrolyte, the coagulation value C1, whichcharacterizes the stability of the sol.With all the concentrations C<C1, the inter-action curves show a positive maximum while with C>C1 they always remain inthe negative region. The coagulation value separates the range of rapid coagulationwhere the time of coagulation is independent of the electrolyte concentration andthe range of slow coagulation where the rate is strongly influenced by the valencyand the concentration of the counterions.In rapid coagulation, the coagulation time T can be deduced from theSmoluchowski relation 1 :where nl(0) is the original number of primary particles and En,(t) the total numberof particles at the time t .In slow coagulation, the coagulation rate is slowed down,due to the presence of the interaction energy Vint, by a factor W which Fuchszdefined aswhere s = R/a and R is the distance between the centres of two approaching par-ticles the radius of which is a. Combination of experimental data related to re-pulsion energies and to coagulation values provided us information concerningthe London-Van der Waals attraction.14144 INTERACTIONS BETWEEN POLYSTYRENE LATICESEXPERIMENTALPREPARATION AND PURIFICATION OF THE LATICESFirst, we chose two monodisperse samples prepared by the Dow Chemical Company,*LS 055A and LS 067A, whose mean diameters, measured by electron microscopy,3 amountedrespectively to 188 mp and to 1,171 mp, with standard deviations of 7-6 mp and 13.3 mp.However, the particle sizes determined by Deielid and Kratohvil4 using different lightscattering techniques averaged 175 and 1,167 mp respectively. Furthermore, determinationsof particle diameters of selenium sols 5 by means of a counting set-up and by comparisonbetween measured and calculated extinction curves were in close agreement.These resultswere always smaller than electron microscopy data. Therefore, we only considered, in thiswork, sizes deduced from turbidimetric measurements. These took into account Heller andTabibian’s 6 recommendations in order to avoid errors occurring especially during thedetermination of large diameters .The other samples have been prepared by emulsion polymerization taking sodiumpersulphate as the initiator and two different emulsifiers.With sodium stearate, we obtainedfirst a latex of 60 m,u diam. ; by ulterior seedings, the particle size amounted to 245 mp.Similarly, with sodium dodecylsdphate, we prepared two latices, the diameters of whichwere respectively 56 and 18Omp. Electron microscopy provided us with the four corres-ponding standard deviations : 6, 12.7, 5.9 and 13.5 rnp. A considerable amount of soapremained in the liquid phase of the latex. For its purification, instead of using electrodialysis,which is time consuming, we passed our systems through mixed bed ion-exchange resins.Based on remarks of Schenkel and Kitchener,7 we measured the electrophoretic mobilities ontwo portions of the same latex, the former purified by prolonged electrodialysis ,the latter bypassage on a mixed bed.The measurements were carried out at the same electrolyte content.We avoided contamination from the desorbed polyelectrolytes diffusing slowly out of theresin by rinsing the column with conductivity water just before use. Under these conditions,no discrepancy between the two samples could be observed.KINETICS OF COAGULATION AND COAGULATION VALUEWe could follow spectrophotometrically the course of aggregation of the polystyrenelatices. They are colourless and two cases must be considered. (i) When the particle sizeis much smaller than the wavelength used for the measurement, the Rayleigh theory 8 can beused. Troelstra 9 showed in this case that the extinction coefficient y(t) increases linearlywith time.Practically, the extinction curves are concave to the time axis. This is attributedto the fact that agglomerates scatter far less than the corresponding massive particles.Therefore, Reerink 10 proposed to work at the very beginning of the coagulation. (ii) Whenthe particle size is not in the Rayleigh range, there are two possibilities for interpreting theaggregation process by the Mie 11 theory.The extinction curves of polystyrene latices had a maximum, the position of which wasrelated to the particle size; for larger wavelengths the curves were smoothly decreasing.During the coagulation, we observed a decrease of the extinction at the maximum and anincrease at large wavelengths.When the extinction maximum appears at short wavelengths, it was possible to measureand interpret the extinction increase at large wavelengths.Therefore by means of theSmoliichowski relations,we were able to calculate the relative populations of the particles of various q orders as afunction of time t, which we considered only between 0 and 0.5 T. These values were limitedto the sixth order and their summations were performed. For these short times, the resultsfrom eqn. (3) did not differ by more than 0.2 % from those obtained by means of eqn. (1).* Our thanks are due to Dr. J. W. Vanderhoff, who sent us the PSL samplesA . WATILLON AND A.-M. JOSEPH-PETIT 145034As in the Troelstra treatment for the Rayleigh range, we assumed that the multiple particleof q order is an isotropic sphere having a volume I/ = qu.At the wavelength of the measure-ment, it was possible to determine 12 or to compute from the Mie theory the extinction cross- - - - _. - _ _ - __ _. _ _ _ - ~/ /’,/‘ ’ ,/’// /’// /FIG. 1 .-Variation of the y(t)/y(O) - 1 function with the reduced time t/T calculated at the wavelengthA = 700 mp. I, Troelstra equation : y(t)/y(O)- 1 = 2t/T; 11, particles of 60 mp; 111, 175 mp ;IV, 245 mp ; V, 1,167 mp. - - - - linear variation.I 0 2 -,______-- H I,“sc dn>sections CZxt corresponding to particles of different orders. Thus we could derive the follow-ing expressions for extinction coefficients at times 8 and t respectively,y(0) = C y n l ( 0 ) (4146 INTERACTIONS BETWEEN POLYSTYRENE LATICESanda = 6y(t) = ki?fIlq(t).q= 1( 5 )As an example, fig.1 shows a comparison between the Troelstra equation, strictly valid forvery small particles and the evolution at il = 700 mp of the y(t)/y(O)- 1 function against t/T,calculated according to the Mie theory for 4 particle diameters : 60, 175, 245 and 1,167 mp.For the smallest particles (60 mp), even at short relative times, they are already outside theRayleigh range. The deviation from the Rayleigh range increases with time due to thegreater contribution of larger particles. Fortunately, the calculated function can be con-sidered as linear at short times. Finally we observed the lack of accuracy of this methodwhen applied to large particles (1.167 mp).For large particles, the calculations were performed at the wavelength of the extinctionmaximum (A = 500 mp) in the same manner as previously explained for the first possibility.Fig. 2 presents the function y(O)/y(t)- 1 against t/T in a Smoluchowski-type plot.Withintimes shorter than 0.2 t/T, the optical evolution can also be considered as linear. We con-cluded that for the two cases investigated we can study the aggregation process by workingat the very beginning of the coagulation such as in the Rayleigh range.DETERMINATION OF THE ELECTROKINETIC POTENTIALSWe measured electrophoretic mobilities using a moving boundary technique adapted foilyophobic colloids.13 Working always at very low particle volume fractions, (about lO-5),there was no surface conductance contribution to the measured conductance in the cell.Inorder to calculate the electrokinetic potential from the electrophoretic mobilities, the treat-ments of Overbeek,l4 Booth 15 and Wiersema 16 are available. As the potentials needed forthis work corresponded to concentrations not too far from the coagulation value, they werenever large. Strictly, the Overbeek equation fails for 5>25.7/u mV but for our systems, thethree methods gave the same results.RESULTSCOAGULATION VALUESAs the indifferent co-ion we chose the anion C10, which does not seem to beadsorbedNaOH.DifferentNaC104,on the particle surface. As peptizing electrolytes, we used HC104 orWe determined the coagulation value in presence of Naf or H+ ions.electrolyte conditions were used : pure NaC104, pure HC104, NaOH+HC104 + NaC104 and NaOH + Ba(C10&.Table 1 summarizes theTABLE CO COAGULATION VALUES OF THE DOW LATICESdiam. mp electrolytes175 10 mM/I. NaOHf NaC104175 NaC104175 10 mM/l. HC104+NaC104175 HCIO~1,167 10 mM/1. NaOH+ NaC1041,167 1 mM/1. NaOH+NaC1041,167 HC1041,167 10 mM/1. NaOH+ Ba(C104)ZC1, mM/1.90652517716526 12207513-5results obtained with the Dow latices prepared with a sulphonate as emulsifier.17As an example, fig. 3 shows the loglo W,loglo C relation obtained with the 175 mpdiam. latex in the previously described electrolyte conditions. Table 2 gives resultswhen sodium stearate has been used as emulsifier, and table 3 when sodium dodecyl-sulphate was used. Under the same electrolyte conditions, for highly purified systemsA.WATILLON A N D A.-M. JOSEPH-PETIT 147the coagulation value shifts can hardly be related to the emulsifier nature or to theparticle size. This is due to the fact that during the polymerization the soap present10- I 1FIG. 3 . P l o t of the Dow latex LS 055A experimental stability against respective cation concentration :I, 10-2 M/1. NaOH+NaC104 ; 11, NaC104 ; III, 10-2 M/1. HC104fNaC104 ; IV, HC104.TABLE 2.-cOAGULATION VALUES OF THE SODIUM STEARATE LATICESdiam. mp electrolytes C1, mM/I.60 10 mM/1. NaOHf NaC104 1,00060 NaC104 560245 NaC104 200245 1 mM/l. HClO4f NaC104 175245 HClO4 50TABLE 3.<OAGULATION VALUES OF THE SODIUM DODECYL SULPHATE LATICESdiam. mp electroiytes C1, mM/1.56 NaC104 52556 HClO4 115180 NaC104 455180 HC104 168339 NaC104 324339 HC104 111on the particle surface was not held constant from one seeding to another.Never-theless, we observed for each system that a decay of pH diminishes the stability inrelation to the gradual variation of the relative number of superficial ionized groups.ELECTROPHORETIC DATAThe electrophoretic mobilities have been measured with the same electrolytesas in the coagulation study. For some systems, many concentrations were in-vestigated up to the rapid coagulation range. As an example, table 4 shows thechange of mobility and l, potential against concentration for one of our various studieson the Dow LS 055 A latex. All the other data are summarized in table 5148 INTERACTIONS BETWEEN POLYSTYRENE LATICESTABLE 4.-ELECTROKiNETIC POTENTIALS OBTAINED WITH THE DOW LATEX Ls 055 A IN PRESENCEOF 10 mM/1.H a 0 4 AND DIFFERENT CONCENTRATIONS IN NaC104NaCIO4 concentrationmMI1.506080100120140160200250Ka70.375.986.093.1103-311 1.1118.4131.0151.5U, pcmlV sec2-402.3 12-162.041.921.821.671.541-39C, mV32.030.628.627.025.224-052 1 -0620.218.2TABLE 5.-vALUES OF THE HAMAKER CONSTANTS DETERMINED EXPERIMENTALLYemulsifier, expt. no. dim. mpDow 1sulphonate 23456sodium 7stearate 8sodium 9dodayl-sulphate 101751751,1671,1671,1671,16724524556180eIectrolytes C1, mM/1.10 mM/l. HC104+ NaC104 177HC104 16510 mM/1. NaOH+NaC104 2611 mM/l.NaOH+NaC104 220HC104 7510 mM/l. NaOH+ Ba(C104)~ 13.51 mM/l. HC104 + NaC104 175HClO4 50HClO4 115HC104 168C,mV21-521-635373426.52015.51420.6A X 1013 erg0.5 fO.10.6 10.11 -2 f0.21-4 f0.22.2 f0.31.6 rt0.20.7 f0.10-5 f0.10-4 f0-10.6 f0-1INTERACTION CURVESIt is possible to calculate 18 the Coulomb repulsion energy as a function of thereduced distance Rla separating two spherical particles when the surface potentialis known. Verwey and Overbeek have shown for planes that the electric potential$S at the limit between the Stern 19 and Gouy 20 layers remained practically con-stant during the approach, except at very small distances where the Van der Waalsattraction largely prevails. They concluded that $6 rather than $0 governs therepulsion against distance relationship.Nevertheless, in order to perform thosecalculations, as $6 is difficult to determine experimentally, different investigators 79 21-25approximated $8 to the electrokinetic potential 5. Furthermore, if we assume asLyklema and Overbeek 26 the presence of a highly viscous layer along the particlesurface, it is possible that during the approach of two particles the ionic redistributionbetween this layer and the external region could hardly occur. The potential 5rather than $8 could then be considered as governing the repulsion energy.The assumption that the atmosphere charge remains constant during the ap-proach seems in the present case a more suitable approximation. In fact, Verweyand Overbeek have shown that for a given configuration of two parallel planes theforce acting between the surfaces must be independent of the choice of the para-meter to be held constant during the variation of their distance.They showedthat the assumptions of constant potential or of constant charge lead to exactly thesame expression for the repulsion forceA . WATILLON AND A . - M . JOSEPH-PETIT 149In conclusion, the 5 potential used in connection with the constant potentialhypothesis is a reasonable way to estimate the repulsion energies. For our systemsthese energies have been calculated by means of the Derjaguin27 equation whenthe pctential was small. In a few cases the potential exceeded 257/umV and weused the G function tabulated by Verwey and Overbeek.18Hamaker 28 developed an equation relating the attraction energy between twospheres of a given radius and the distance separating their centres.Furthermore,Casimir and Polder29 pointed out that when the distance between two atoms isnot very short their mutual attraction will be attenuated. Overbeek 30 adaptedFIG. 4.-Interaction energy curves calculated for the Dow latex LS 055A+ 10-2 M/I. HC104+NaC104(particle radius 3 8-8 x 10-6 cm, A = 0 5 x 10-13 erg).these considerations to colloidal particles. For our purpose we used the generalequation of Hamaker corrected for the retardation.We combined the repulsion curves previously obtained with the attractioncurves related to different arbitrary values of the Hamaker constant.As anexample, fig. 4 gives a set of interaction curves constructed for A = 5 x 10-14 erg.There is no important shift of the particle distance corresponding to the potentialenergy maximum. Verwey and Overbeek 18 calculated the total interaction fortwo hypothetical cases. At constant potential they found that the interparticledistance of the maximum decreases with the ionic strength by compression of thedouble layer. At constant ionic strength, they showed that for a decay of thepotential there results an increase of the interaction maximum distance. In ourexperimental cases, the two effects counterbalance each other and the interactionmaximum corresponds practically to the same interparticle distance150 INTERACTIONS BETWEEN POLYSTYRENE LATICESCONSTRUCTION OF STABILITY CURVESThe stability curves were calculated by means of the Fuchs 2 function which weintegrated graphically. As an example, fig.5 summarizes the results correspondingto the combination of a series of experimental potential-concentration pairs, withthree arbitrarily chosen Hamaker constants. In our systems, where the potential( 50m M/1i ,-,-.-132mVI I I I!T2 5 x 1 a 2 lo-' 7~10~' 5JO-'loglo ma+] (equiv./l.)FIG. 5.-Stability curves obtained from integrations of the Fuchs functions (Dow latex LS 055A + 10-2 M/1. HC104+NaC104). I, A = 0-2 x 10-13 erg ; 11, A = 0.5 x 10-13 erg ; 111, A = 10-13 erg.decreases with increasing ionic strength, we could construct the same shaped loglo W/loglo C relation which has been calculated 18 assuming the potential did not varywith the ionic strength.For these curves it is possible to estimate the hypotheticalcoagulation values corresponding to the different arbitrarily introduced Hamakerconstants.DISCUSSIONDETERMINATION OF THE HAMAKER CONSTANTIn order to determine the most probable value of the Hamaker constant, weplotted in fig. 6 loglo (coagulation values) as a function of loglo (the correspondingHamaker constants introduced arbitrarily). The most probable value of theHamaker constant was then determined from this plot for the measured coagulationvalue. Table 5 summarizes the experimental conditions corresponding to all thecases investigated also given in fig. 6. For a given latex there is agood agree-ment between values obtained with different peptizing electrolyte conditions anA .WATILLON AND A.-M. JOSEPH-PETIT 151with counterions of different valencies (Na+ or Ba2f). For particles cf about thesame size (175, 180 and 245 mp) the measured Hamaker constant does not seem tovary with the nature of the superficial ionized group (sulphonic, carboxylic or1 i ____ 1 - 1--. _--L L---- I __ A__5& n 71KT2 3L12 lot a0 I SJOlog10 c1 (MA.)FIG. 6.-Representation of the Hamaker constants of polystyrene embedded in water according totable 5.sulphate). Finally, the results seem to be split in two groups, one with valuesranging around A = 1.5 x 10-13 erg corresponded to measurements on large dia-meter particles (1,167 mp), the other value was about A = 0.5 x 10-13 erg for allthe results obtained with small sized particles (55 to 245 mp).VISCOELECTRIC CORRECTIONWe also evaluated whether in our case the Lyklema and Overbeek26 consider-ation relating to the presence of a highly viscous layer near the wall has to be takeninto account.At the coagulation value we determined the repulsion curves whichcounterbalance exactly the attraction ones corresponding to the three Hamakerconstants : 0.5 ; 1 and 2 x 10-13 erg. As an example, for the set of experimentsno. 3 of table 5, the corresponding potentials were respectively 20.5, 3 1-3 and 46.7 mV.When we applied the viscoelectric correction to these values, we got cobs = 15.4,19.1 and 21-6 mV. Assuming that the potential is infinitely high near the wall,the zeta potential cobs should amount to 24.2 mV, which is still much smaller thanthe experimentally determined potential cexpt = 35 mV.We concluded then thatthe viscoelectric constant of water is overestimated by a factor of 10. For ourcalculation, we preferred to neglect this correction.COMPARISON BETWEEN MEASURED A N D CALCULATED VALUES OF THEHAMAKEK CONSTANTThe previously determined total interaction constant A corresponds to twosolid spheres embedded in a liquid phase. Overbeeklo proposed to relate thi152 INTERACTIONS BETWEEN POLYSTYRENE LATICESconstant to the individual Van der Waals constants of the solid phase A s and of theliquid phase AL by the equation,A = ( A t - At)2.Each term can be estimated from the expression,Ai = n2q;Ai, (7)where qi is the number of atoms per cin3 and lli the London constant of the substance.& can be calculated by means of the London,31 the Slater-Kirkwood 32 andthe Neugebauer 33 equations.These require a knowledge of the polarizabilities cc.They have been calculated using the Lorentz 34-Lorenz 35 equation where we intro-duced as refractive index the value extrapolated to zero frequency by means of aCauchy relation? We took nwater = 1.324 and npSL = 1.568 and obtainedawater = 1.43 x 10-24 cm3 and a p s ~ = 12-8 x 10-24 cm3. We made another estima-tion using for EPSL the polarizability of benzene determined by de Boer 37 (ccbenz. =16 x 10-24 cm3). In the London equation we took as lowest ionization potentialfor water the value hvo = 13.6 eV of hydrogen and for polystyrene the value hvo =8-5 eV of toluene respectively, In the Neugebauer equation we used the diamagneticsusceptibility of water x = - 2.1 5 x 10-29 e.m.u./molecule and for polystyrene wechose the value of ethylbenzene : x = - 12.5 x 10-29 e.m.u./molecule.TABLE 6.-CALCULATION OF THE HAMAKER CONSTANT OF POLYSTYRENE EMBEDDED IN WATERaPSL = 16 X 10-24 m 3 (de Boer)A(water) x 1013 ergequationA(PSL) x 1013 erg A x 1013 ergLondon 9.6 3.7 1.3Slater-Kirkwood 1 6 .8 6.0 2.7Neugebauer 12.2 5-6 1.3ccPSL = 12.8 x 10-24 cm3 (Lorentz-Lorenz)London 6 . 2 3.7 0-35Slater-Kirkwood 12.0 6.0 1 -05Neugebauer 9.8 5.6 0-55Table 6 summarizes the Van der Waals and Hamaker constants calculatedaccording to both assumptions.For the first, the results ranged between 1.3 x 10-13and 2.7 x 10-13 erg and were in agreement with the high experimental values. Thesecond set fall between 0.35 x 10-13 and 1.05 x 10-13 erg. It is difficult to makea choice between high and low experimental and theoretical values. Nevertheless,Schenkel and Kitchener 7 have shown that particles of 10 p diam. could aggregatein the secondary minimum of the energy barrier, for which a smaller Hamaker con-stant is sufficient to produce coagulation. For particles in the 1 p range it ispossible that our treatment which neglects this effect gives slightly too high valuesfor the Mamaker constant. Moreover in the calculation the use of the refractiveindex of polystyrene in order to obtain the polarizability seems to be a better ap-proximation than the choice of the benzene polarizability.Therefore we propose to consider as calculated Hamaker constant values between0.35 x 10-13 and 1-05 x 10-13 erg, and experimental values between 0-4 x 10-13 and0.7 x 10-13 erg.Finally, the value of Fowkes 38 (0-5 x 10-13 erg) was obtained ina completely different manner. In conclusion, we can consider that the experi-mental results are in fair agreement with calculated data153 A . WATILLON AND A . - M . JOSEPH-PETIT1 Smoluchowski, Physik. Z., 1916, 17, 557. Z. physik. Chem., 1917, 92, 129.2 Fuchs, 2. Physik., 1934, 89, 736.3 Vanderhoff, J. Opt. Soc. Amer., 1954, 44, 603 ; J. Appl. Physics, 1955, 26, 864.4 Dezelic and Kratohvil, J. Colloid Sci., 1961, 16, 561.5 Watillon and Dauchot, to be published.6 Heller and Tabibian, J.Colloid Sci., 1957, 12, 25.7 Schenkel and Kitchener, Nature, 1958, 182, 131; Trans. Faruduy Soc., 1960,56, 161.8 Lord Rayleigh, Phil. Mug., 1871,41, 107,274,447 ; 1897, 44, 28 ; 1899, 47, 375.9 Troelstra, Thesis (Utrecht, 1941).10 Reerink, Thesis (Utrecht, 1952).11 Mie, Ann. Physik., 1908, 25, 377.12 Heller, Pangonis and Jacobson, Tables of Light Scattering Fimctioizs for Splzericul Particles13 to be published.14 Overbeek, Adv. Colloid Sci., vol. 3 (New York, 1950).15Booth, Proc. Roy. SOC. A, 1950, 203, 514.16 Wiersema, Thesis (Utrecht, 1964).17 Vanderhoff, private communication.18 Verwey and Overbeek, Theory ofthe Stability of Lyophobic Colloids (Elsevier, 1948).19 Stern, Z. Elektrochem., 1924, 30, 524.20 Gouy, J. Physique, 1910, 9 (4), 457 ; Ann. Physique, 1917, 7 (9), 129.21 Douglas and Burden, Trans. Faruduy Soc., 1959, 55, 356.22 Watillon and Joseph, Proc. 3rd Int. Congr. Surf. Activity, 1959, 1, 145.23 Choudhury, Nature, 1960, 185, 308.24 Hunter and Alexander, J . Colloid. Sci., 1963, 18, 820.25 Ottervill and Wilkins, Trans. Faradby Soc., 1962, 58, 608.26 Lyklema and Overbeek, J. Colloid Sci., 1963, 18, 501.27 Derjaguin, Kolloid Z., 1934, 69, 155 ; Acta physicochim., 1939, 10, 333 ; Trans. Faraday SOC.,28 Hamaker, Physica, 1937, 4, 1058.29 Casimir and Polder, Nature, 1946, 158, 787; Physic. Rev., 1948, 73, 36030 Overbeek in Kruyt, Colloid Science, vol. 1 (Elsevier, 1952), p. 266.31 London, Z. Physik., 1930, 63, 245.32 Slater and Kirkwood, Physic. Rev., 1931, 37, 382.33 Neugebauer, Z. Physik, 1937, 107, 785.34 Lorentz, Wiedem. Ann., 1880, 9, 641.35Lorenz, Wiedem. Ann., 1881, 11, 70.36Bateinan, Weneck and Eshler, J. Colloid Sci., 1959, 14, 312.37 de Boer, Trans. Faraday Soc., 1936, 32, 10.33 Fowkes, hid. Eng. Chem., 1964, 56, 40.(Wayne State University Press, Detroit, 1957).1940, 36, 203

ISSN:0366-9033    

DOI:10.1039/DF9664200143

出版商:RSC    

年代:1966

数据来源: RSC

摘要:

Stability of Monodisperse Polystyrene Latex Dispersions ofVarious SizesBY R. H. OTTEWILL AND J. N. SHAWDept. of Physical Chemistry, University of BristolReceived 20th June, 1966A series of monodisperse polystyrene latex dispersions has been prepared covering a range ofparticle sizes from 600 to 4230A. The kinetics of flocculation of the latex dispersions by bariumnitrate solutionswere examined and theresults used to construct curves of the logarithm of the stabilityratio against the logarithm of electrolyte concentration. The slopes of the curves were calculatedtheoretically and compared with those obtained experimentally. The theory predicted an increaseof slope with particle size which was not observed experimentally. Experimental estimates of theHamaker constant give values ranging from 1-03 x 10-14 to 1-10 x 10-13 erg.Although the theory of colloid stability in the form proposed by Derjaguin andLandau 1 and Verwey and Overbeek 2 is now accepted by most workers in its broadoutline, experimental tests of the theory on a number of points are still required usingwell-defined systems.The theory was largely developed for the two simplest cases ofinteraction, namely, that between two flat plates and that between two spheres of equalsize. Experimental tests are therefore clearly more valid if carried out on systemsconforming closely to the theoretical models. The " flat plate " case has receivedconsiderable attention recently in experimental studies on thin soap fflms. The" sphere " case has not received such extensive and detailed investigation althoughsome studies have been reported of investigations on the flocculation of systems ofspherical monodisperse particles.3149 5 3 6 Recently, we have investigated in somedetail the preparation and characterization of a number of monodisperse systems.7~ 819So far, the system which appears to conform most clearly to an ideal model system isthat of polystyrene latex particles produced by the emulsion polymerization of styrene,using hydrogen peroxide as an initiator and a soap as an emulsifying agent.9 Electro-phoretic, infra-red, radiotracer and titration studies have shown that dialysis removesthe adsorbed soap and leaves a spherical polymer particle with a carboxylated surface.10Dispersions so produced are monodisperse and a range of particle sizes can be ob-tained.9 Commercial latices were rejected as experimental material since althoughthey are monodisperse, their surfaces are usually uncharacterized and the types andsources of stabilizers employed are usually unknown.The present paper gives a preliminary account of some stability experimentscarried out using a series of monodisperse polystyrene latex dispersions, in which theparticle diameters were varied from 600 to 4230 A.The kinetics of flocculation of thelatices using barium nitrate as the flocculating electrolyte has been investigated in anattempt to determine how the particle size influences the stability of dilute dispersions.EXPERIMENTALMATERIALSAll chemicals were of A.R.grade. The distilled water used was double-distilled materialand was always redistilled from an all-Pyrex still immediately before use.15R. H . OTTEWILL AND J . N. SHAW 155POLYSTYRENE LATICESThe latex dispersions were prepared by emulsion polymerization at 70°C using hydrogenperoxide as the initiator. The emulsion was stabilized by sodium laurate. Full details ofthe method of preparation have been given elsewhere.11 After preparation of the seed latexdispersion and growth to a suitable size, the adsorbed lauric acid could, within experimentalerror, be removed by dialysis at an alkaline pH. The latex particles were stabilized by carboxylgroups which were covalently linked to the particles, i.e., were an integral part of the polymerchains, rather than by adsorbed laurate ions.Particle size distributions were obtained bvelectron microscopy. The diameters of the particles obtained from the mode of the particlesize distribution curve are listed in table 1. The dispersions all have a high degree of mono-dispersity and there was only a small coefficient of variation on the modal diameters.TABLE 1 .-DIAMETERS OF POLYSTYRENE LATEX PARTICLESlatex modal diameter, A coefficient of variationdesignation electron microscopy on the modal diameter, %ABCDE600103024254230368016.08.05.44.63-8ELECTROPHORETIC MOBILITIES OF THE LATEX PARTICLESThe electrophoretic mobilities of the particles of latices A and B were determined by themoving boundary method 10 since latices of these sizes were not visible in the ultramicroscope.Observations on latices C, D and E were made using the ultramicroscopic microelectrophore-sis technique, employing the apparatus and methods previously described.12.13 The latexdispersion which had been dialyzed continuously against very dilute sodium hydroxide (ca.pH 8) was diluted with the appropriate concentration of barium nitrate solution and the pHof the solution adjusted to pH 8.0 by addition of sodium hydroxide.The pH of the dispersionwas measured before and after each measurement ; usually the values agreed to within 50.1of a pH unit. Concentrations of ca. 108 particleslml were employed for microelectrophoresis.All mobility measurements were carried out at 25 fO.1"C.DETERMINATION OF SOL STABILITYIf it is assumed that flocculation is a kinetic process which is described by the equations ofSmoluchowski,~~ then the total number of particles present at a time t after adding a floccu-lating electrolyte, is given byN , = No/(l + kNoZ), (1)where NO = the number of particles present in the initial stable dispersion; these are allassumed to be single particles.ThenN , = N I + N Z + N 3 + . . .,where N1 = the number of single particles after time t, N2 = the number of double particlespresent after time t, N3 = the number of triple particles present after time t, etc., k isa rate constant for the flocculation process which can be put equal to ko/W, where Wis the stability ratio obtained by considering slow flocculation as a diffusion process in apotential field.15 ko is the rate constant for rapid flocculation.Since discussion will becentred on curves of log W against log c, rather than the absolute values, we shall choose todefine a W&pt bywhere ko and k are the values determined experimentally in the rapid and slow regions offlocculation respectively156 STABILITY OF LATEX DISPERSIONSThe method employed for the determination of k depends on the size of the particles.Three methods have been employed in the present work and these were subdivided as follows :(a) PARTICLE COUNTING.-Eqn. (1) can be employed directly if the total number of particlespresent at set time intervals can be counted. This method is useful for particles which areeasily visible in the ultramicroscope and was found particularly useful for latex E.The particlecounter and methods previously developed were employed.16(b) LIGHT SCATTERING-RAYLEIGH REGION.-when the particle diameter is less than1/10, the turbidity 70 of a dispersion is given byz0 = AN,V;, (3)2, = ANoVi(l +2kNot), (4)where A = an optical constant and VO = the volume per particle. The change of turbiditywith time is given by 17918and hence the constant k can be obtained by observing the rate of change of turbidity withtime in the early stages of flocculation. This was carried out experimentally using an SP 600Unicam spectrophotometer equipped for automatic recording.19 An adder-mixer 19 wasused for adding the barium nitrate solutions to the latex dispersions.(c) LIGHT SCATTERING-ME REGION.-FOr particles having a diameter > l / l O , the Mietheory of light scattering has to be employed.20 Under these conditions the turbidity of theinitial monodisperse dispersion is given byand the turbidity at time t byzo = NonaTKtI, ( 5 )(6) 2, = N1na:K,, 4- N,na:K,,+ N,na;K,, .. .,where al, a2 and a3 are the radii of single particles, double particles and triple particlesrespectively. a2 was taken to be the radius of a sphere equivalent in volume to two singleparticles. a3 was similarly taken to be the radius of a sphere equivalent in volume to threesingle particles. Ktl, K,, and K,, were the total scattering coefficients of particles havingradii of al, a2 and a3. Measurements were made at a wavelength 1 of 5461 A (in uacuo)which corresponds to a relative refractive index for polystyrene in water of 1.20.Thus thevalues of the total scattering coefficients were taken from the tables of Pangonis, Heller andJacobson 21 computed for this value of relative refractive index. Assuming that the kineticsof flocculation were as given by Smoluchowski’s treatment,l4 the turbidity zt was calculatedfrom eqn. (6) for N, = 0.95 NO. The time taken for the turbidity to increase from 20 tozt, i.e., for NO to fall to N,, was observed and the rate constant k calculated from eqn. (1).The results for all latices were obtained using the Mie procedure and compared withthose found using eqn. (4). With latex A, which would be expected to contain particlesacting as Rayleigh scatterers, excellent agreement was obtained between the Mie and theRayleigh procedures.RESULTSSTABILITY OF POLYSTYRENE LATEX DISPERSIONSThe mobilities of the latex particles used all attained a constant negative valueabove pH values of cd.74,9 indicating that the zeta-potential of the particles remainedconstant above this pH. All flocculation experiments were therefore carried out atpH 8.0+0.5. Although a number of different flocculating ions have been examined,the results presented will be restricted to those obtained using barium nitrate as theflocculating electrolyte. The barium ion appears to be stable and as far as we couldascertain from experiments and from the literature, does not appear to hydrolyzeuntil well above the pH region employed. The determination of flocculation con-centrations and stability was also found to reach its optimum precision under theexperimental conditions employed with 2f valent electrolytesR.H . OTTEWILL AND J . N. SHAW 157A- 2 0- 2 0 -1.5DI I-2.0 -1.5log molar concentration of barium nitrateFIG. 1.-Log Wexpt against log c, curves, 0, latex A ; X , latex B ; 0, latex C ; A, latex D.log molar concentration of barium nitrateX , results obtained by particle counting.FIG. 2.--Log Wexpt against log c, curves for latex E. 0, results obtained by light scattering158 STABILITY OF LATEX DISPERSIONSThe results obtained for latices A, B, C and D are presented in fig. 1 in the form oflog Wexpt against log ce curves ; ce was taken as the molar concentration of bariumnitrate present.The results for latex E are presented in fig. 2 and a comparison isgiven of Wexpt obtained from light scattering experiments using the Mie procedurewith that obtained by direct particle counting. The agreement obtained is excellentand justifies the experimental light scattering procedure adopted. The curves are ofTABLE 2.-FLOCCULATION CONCENTRATIONS FOR BARIUM NITRATE SOLUTIONS'latex a@) d log Wexpt/d log ce flocculation conc.A 300 - 2.42 4-11 x 1011 1.88 X 10-2 MB 515 - 2.78 1-31 x 1011 2.82 x 10-2 MC 1213 - 2.71 9 . 9 0 ~ 109 1.88 x 10-2 MD 1840 - 2.70 2 . 7 0 ~ 109 1-78 x 10-2 ME 2115 - 1.59 1.55~ 109 1.41 x 10-2 Mthe same general shape as those found for the flocculation of a number of othercolloidal dispersions and can be represented by an equation of the form 22(7)The slopes of the curves and the flocculation concentrations are given in table 2.The flocculation concentrations were obtained by extrapolating the log Wexpt valuesdown to log Wexpt = 0, and reading off log ce at the point of intersection.The log Wexpt against log ce curves were linear over the range investigated andwithin experimental error no evidence of curvature was obtained.The variation ofslope with particle size expected on theoretical grounds 22 was not observed. In fact,the slopes of all the curves were remarkably similar with a possible trend, if any,towards smaller slopes with increase in particle size.log Wexpt = k2 - kl log c,.ZETA-POTENTIALS OF POLYSTYRENE LATICES I N THE PRESENCE OF BARIUMNITRATEThe mobility results obtained were converted into zeta-potentials using the theor-etical computations of Wiersema23 which allow for both the retardation and relaxationof the electrical double layer.The results obtained for all five latices are presented infig. 3 in the form of curves of zeta-potential against the logarithm of the molar con-centration of barium nitrate.THEORETICALTHEORY OF COLLOID STABILITYThe theory of colloid stability of hydrophobic dispersions most commonly acceptedat present is that propounded by Derjaguin and Landau 1 and Verwey and Overbeek.2In this theory the total potential energy V of interaction for a two particle system isgiven bywhere VR = the potential energy of repulsion and VA = the potential energy ofattraction.The potential energy of repulsion can be calculated according to theapproximate equation given by Reerink and Overbeek,22 namely,where E = dielectric constant of the medium, u = valency of the counter ion, a =v = vR+v,, (8)VR = 3.469 x 10'9e(kT)2(ay21v2) exp (-zu), (9R. H . OTTEWILL AND J. N. SHAW 159radius of the particle and z = Ica with IC = reciprocal Debye-Huckel double-layerthickness. y = [exp (oe$6/2k2') - l]/[exp (~e$~/2kT) + I], u = Ho/a, where HO =distance between the particle surfaces, e = electronic charge and = Stern potential.The energy of attraction for two spherical particles of equal size has been given byHamaker 24 in the form+2.0 In 1 VA = -- -A{+ 12 x2+2x x2+2x+1log molar concentration of barium nitrateX,latexB; 0,latexC; A,latexD; 0,latexE.FIG.3.-Curves of zeta-potential against log molar concentration of barium nitrate. 0, latex A ;where x = 242.0. A is the Hamaker constant which for particles of material 1immersed in a liquid medium of material 2, is given byA = (Jx - JX2)'. (11)Since the distances over which the attractive forces act for particles of the size usedin the present work are of the order of 100 A, the retardation effkct 25 has been neglec-ted. Under conditions also such that x< 1, eqn. (10) reduces toV A = - A / 1 2 ~ . (12)FLOCCULATION CONCENTRATIONIf the theoretical condition for flocculation is taken that V = 0, and that for amaximum in the potential energy curve aV/du = 0, then from eqn.(9) and (12) withwater as the solvent at 25"C,This indicates that the flocculation concentration is independent of particle radius andis only dependent on y2 (hence $a), A and 1.9 for the conditions under which eqn. (9)and (12) hold.Icfl,, = 2.04 x 10-5y2/~v2. (1 3)STABILITY RATIOTheoretically the stability ratio can be related to the potential energy of interactionby the expression 2Wtheor = 2.0 exp (V/kT)du/(u + 2)'. (14) Jo160 STABILITY OF LATEX DISPERSIONSUsing eqn. (9) and (12), Reerink and Overbeek 22 were able to show theoretically thatan equationcould be derived. The slope of the log Wtheor against log Ce curve was given bylog Wtheor = k; - k; log C,d log Wtheor/d log c, = - 2-06 x 107(ay2/v2). (15)Hence the slope was constant provided that $d remained constant and provided thiscondition was fulfilled, a considerable variation of slope with particle size would beexpected.An alternative approach in order to obtain a curve of log Wtheor againstlog ce is to evaluate the integral given in eqn. (14) for various concentrations ce ofelectrolyte at particular values of $6 and A.DISCUSSIONEFFECT OF PARTICLE SIZE O N THE FLOCCULATION CONCENTRATIONA small variation of flocculation concentration appears to occur with variation insize of the latex particles, the values apparently reaching a maximum for latex B(a = 515 A) and then decreasing. The variation is small but appears to be outsidethe limits of the experimental error on the results. However, the experiments, foroptical reasons, had to be carried out at different number concentrations and althoughthis is allowed for in the computation of k, some variation may occur if specificbinding of the barium ion occurs. Some additional experiments are required in orderto compare the flocculation concentrations for the different latices at constant totalsurface charge density of carboxyl groups.The small variation of flocculation concentration with particle size is in qualitativeagreement with the prediction of eqn.(13). The variation of log Wexpt with d, atconstant barium nitrate concentration (e.g., 10-2 M), indicates that there may be amaximum in the curve of log Wexpt against d, in agreement with the theoreticalprediction of Verwey and Overbeek.26It is of interest to compare the flocculation concentrations obtained with poly-styrene latices with those obtained using other sols.Some values for silver iodidesols, selenium sols, arachidic acid sols and octadecanol sols are given in table 3.TABLE 3 .-FLOCCULATION CONCENTRATIONS FOR VARIOUS SOLS OBTAINED WITH BARIUM IONSsol flocculation conc. ref.silver iodide 2-2 x 10-3 M 272 . 5 ~ 10-3 M 28selenium 2 - 9 ~ 10-3 M 5arachidic acid 1.3 x 10-2 M 29octadecanol 1 . 4 ~ 10-2 M 30The flocculation concentrations appear to fall into two groups, the low values forselenium and silver iodide and the high values for the organic materials, polystyrene,arachidic acid and octadecanol. This could mean that the Hamaker constants arelower for the organic materials or that solvation forces are involved with thesematerials. The present lack of precision in the experimental determination of theHamaker constants makes this a difficult question to resolve from stability studiesalone.HAMAKER CONSTANT FOR POLYSTYRENE LATEX PARTICLES I N WATERFollowing the procedure outlined previously,29 the refractive index data ofRateman, Weneck and Eshler 31 was used to calculate the dispersion frequency oR.H . OTTEWILL AND J . N. SHAW 161polystyrene, and an interpolated value of the refractive index at the dispersionfrequency. The values obtained were vo = 2.156 x 1015 sec-1 and no = 2.0857. Thepolarizability a0 was calculated from the Lorentz-Lorenz equation taking the densityof polystyrene as 1.057 and the molecular weight of styrene as 104.1.This gaveCIO = 2.06 x 10-23 cm3 mo1.-1, the London constant A = 4.55 x 10-57 erg cm6 mok2and hence A11 = 1.68 x 10-12 erg. Combination of this value of A11 with theLondon,32 Slater-Kirkwood 33 and Neugebauer 34 values of A 2 2 for water gave thetheoretical A values listed in table 4.TABLE 4.THEORETICAL HAMAKER CONSTANTS FOR THE POLYSTYRENE-WATER SYSTEMA11 x 1012 erg A22 x 1013 erg method for A22 A x 1013 erg1-68 3.66 London 4.781.68 5.95 Slater-Kirkwood 2.751 -68 5.61 Neugebauer 2.99This value of A11 for polystyrene is slightly lower than the value of 3.17 x 10-12 ergobtained by Schenkel and Kitchener ;35 however, the values used in their calculationswere those of liquid benzene rather than styrene.Experimental values of the Hamaker constant have been calculated by threedifferent methods as follows : (i) from the slope of the log Wexpt against log ce curvesusing the method of Reerink and Overbeek ;22 (ii) from eqn.(1 3) using the flocculationconcentration to evaluate Kflocc and the experimental zeta-potentials obtained at theflocculation concentration to calculate y ; (iii) from eqn. (1 3) using the flocculationconcentration to evaluate Kflocc and $a obtained from the slope of the log Wexptagainst log ce in combination with eqn. (15). The values of A obtained, together withthe values of t,bd, [ and Kflocc are listed in table 5.TABLE 5.-EXPERIMENTAL VALUES OF THE HAMAKER CONSTANT@lg A erg1 2 3 Kflocc, cm-l mv ' mV latexA 7 . 4 2 ~ 106 (3-0) 21.6 1.07 x 10-13 - 1 .1 0 ~ 10-13B 9 . 1 7 ~ 106 (1.0) 17.4 5 . 7 9 ~ 10-14 - 5 . 8 7 ~ 10-14C 7 . 4 2 ~ 106 6-5 2.8 2-89 x 10-14 1-16 x 10-14 2-02 x 10-14D 7 . 2 2 ~ 106 8.5 2.8 1 . 9 6 ~ 10-14 1 . 9 4 ~ 10-14 2 . 0 2 ~ 10-14E 6 . 4 5 ~ 106 12.0 2.0 1-15 x 10-14 3.74 x 10-14 1-03 x 10-14It was not possible to obtain precise values of the zeta-potential near the floccula-tion concentration using the moving boundary method and these values have not beenused for the calculation of A. The experimental values of A range from 1.03 x 10-14to 1.10 x 10-13 erg, i.e., over an order of magnitude which on the high side approachesthe theoretical value. One feature of column (1) is the apparent variation of A withparticle size. The values of A obtained by Watillon and Joseph-Petit 4 using acommercial polystyrene latex ranged from 2 x 10-14 to 1.1 x 10-13 erg, a range inclose agreement with that found in the present work.SLOPE OF THE log Wtheor AGAINST log Ce CURVESValues of d log Wtheorld log cb have been calculated from eqn.(15) and fromcurves of log Wtheor against log ce, where Wtheor was evaluated using eqn. (14).162 STABILITY OF LATEX DISPERSIONSY was calculated using eqn. (9) and (10). A series of curves were constructed of logWtheor against log ce at constant $8, for A values of 2.75 x 10-13 erg, 1-49 x 10-13 erg,5.23 x 10-14 erg and 1.01 x 10-14 erg. In addition, curves of log Wtheor againstlog ce were constructed using the values of zeta-potentials (taken as equivalent to $8)obtained experimentally at particular values of 2 ; hence slopes were obtained forlog Wtheor against log Ce curves of variable $6.The major part of this analysis wascarried out using an IBM 1620 computer. The A value of 5.23 x 10-14 erg was aboutthe mean of the range of the experimental values and results obtained using this valueare presented in table 6. The value of A chosen, however, did not greatly affect theslope obtained.TABLE 6.-cALCULATIONS OF d log Wtheorld log c, ( A = 5.23 x 10-14 erg)latex +8 eqn. (15) eqn. (14)A 2015variableB 2015variableC 2015variableD 2015variableE 2015variable-2.21- 1.20- 3.80- 2.30- 8.95- 5.50---- 13.57- 8.30-- 15.6- 9.40- 1.40- 0.76- 3.76- 2-73- 1.72- 9.00- 5.70- 3.30- 17.2- 10.4-21.8- 11.1- 36.7- 6.30- 7.10All the theoretical calculations predict an increase in slope with increase in particlesize.This, however, was not observed experimentally and the slopes obtained showedlittle change with an increase in particle radius over nearly an order of magnitude. Thesols used, moreover, contained particles which were spherical and monodisperse sothat neither polydispersity nor non-spherical shape can be invoked to explain thedisagreement, as might have been the case in the work of Reerink and Overbeek 22where increasing the radius of silver iodide particles form ca. 250 to ca. 2000 onlyincreased the slope from - 8.0 to - 11.0. The results of Watillon and Joseph-Petit 5on monodisperse selenium sols having particles of radius 775 A also appear to supportthe trend observed in the present work. They obtained an experimental slope of-2.5 for flocculation with barium chloride and a theoretical slope of - 10.9.The slopes obtained are only dependent on the flocculation experiments and not onelectrokinetic measurements. The former are precise and the agreement between thelight scattering method and particle counting for latex E (fig.2) would appear toconfirm the accuracy of the experimental kinetic results. Additional experimentalwork has also indicated that the change in valency of the counter-ion predicted byeqn. (15) is not confirmed experimentally when for the same latex dispersion u ischanged from 1 to 3. Clearly a discrepancy exists between theory and experiment.Part of the work described in this paper was carried out at the Department ofColloid Science of the University of Cambridge.We thank Dyestuffs Division ofImperial Chemical Industries Limited, and the Science Research Council for gener-ously supporting this workR. H . OTTEWILL AND J . N. SHAW 1631 Derjaguin and Landau, Acta physicochim., 1941,14,633.2 Verwey and Overbeek, Theory of Stability of Lyophobic Colloids (Elsevier, Amsterdam, 1948).3 Watillon, Romerowski and van Grunderbeeck, Bull. SOC. chim. Belges, 1959, 68,450.4 Watillon and Joseph-Petit, Abstr., A.C.S. Symp. Coagulation and Coagulant Aids, 1963, p. 61.5 Watillon and Joseph-Petit, 3rd Int. Congr. Surface Actiuity, 1960, 1, 145.6 Watillon and Gerard, 4th Znt.Congr. Surface Actiuity, 1964, preprint B/VL, 28.7Ottewill and Woodbridge, J. Colloid Sci., 1961, 16, 581.8 Ottewill and Woodbridge, J. Colloid Sci., 1964, 19, 606.9 Shaw and Ottewill, Nature, 1965, 208, 681.10 Shaw, Ph.D. thesis, (Cambridge, 1965.)11 Ottewill and Shaw, Kolloid-2. u. 2. Polyrnere, 1967, 215, 161.12 van Gils and Kruyt, Kolloid-Beih., 1937, 45, 60.13 Ottewill and Rastogi, Trans. Faraday SOC., 1960,56,880.14 Smoluchowski, 2. physik. Clzem., 1917, 92, 129.15 Fuchs, 2. Physik, 1934, 89, 736.16Ottewill and Wilkins, J. Colloid Sci., 1960, 15, 512.17 Troelstra and Kruyt, Kolloid-Beih., 1943, 54, 225.18 Oster, J. Colloid Sci., 1960, 15, 512.19 Ottewill and Sirs, Bull. Photometric Spectrometry Group, 1957, 10, 262.20 Mie, Ann. Physik, 1908, 25, 377.21 Pangonis, Heller and Jacobson, Tables of Light Scattering Functions for Spherical Particles(Wayne State University Press, Detroit 1957).22 Reerink and Overbeek, Disc. Faraday SOC., 1954, 18, 74.23 Wiersema, Thesis, (University of Utrecht, 1964).24 Hamaker, Physica, 1937,4, 1058.25 Overbeek, in Kruyt, Colloid Science (Elsevier, Amsterdam 1952) vol. 1, p. 270.26Verwey and Overbeek,z p. 177.27 Kruyt and Klomp5, Kolloid-Beih., 1943, 54, 484.28 Mirnik, Flajirnan, Schulz and Teiak, J. Physic. Chem., 1956, 60, 1473.29 Ottewill and Wilkins, Trans. Faraday SOC., 1962, 58,608.30 Johnson, Goldfarb and Pethica, Trans. Fmaday Soc., 1965,61,2321.31 Bateman, Weneck and Eshler, J. Colloid Sci., 1959, 14, 308.32 London, 2. Physik, 1930, 63, 245 : Trans. Faraday SOC., 1937, 33, 8.33 Slater and Kirkwood, Physic. Rev., 1931, 37, 682.34 Neugebauer, 2. Physik, 1937,107,785.35 Schenkel and Kitchener, Trans. Faraday SOC., 1960,56, 161

ISSN:0366-9033    

DOI:10.1039/DF9664200154

出版商:RSC    

年代:1966

数据来源: RSC

摘要:

GENERAL DISCUSSIONProf. J. T. G. Qverbeek (van’t HoflLab., Utrecht) (communicated) : In discussingthe interaction curves Watillon remarks that “ Verwey and Overbeek have shown that,for a given configuration of two parallel planes, the force acting between the surfacesmust be independent of the choice of the parameter to be held constant during thevariation of this distance”. This is only correct for an infinitesimal variation of thedistance. For a finite variation, the repulsion changes more rapidly at constantcharge than at constant potential.Dr. A. Watillon (Univ. Libre, Bruxelles) said : In reply to Overbeek, our remark isnot complete because we are lacking data concerning the repulsion energy betweenplanes calculated at constant charge. For our systems, it would be interesting to getthe same information for spheres at large rca (for the coagulation values, rca rangedbetween 31 and 1,009).For small rca values, we could estimate the differencebetween repulsion energies calculated 1 at constant charge and at constant potential.When rca is 1, 2, 3 and 5 at the reduced interparticle distance corresponding to theinteraction maximum (7cHo = 1.2), the energy difference ranged between 3 and 4 %.Furthermore, as our KHO value is large enough, we may use expressions (12) and(12’) of Derjaguin’s 2 paper derived from Levine’s 3 considerations. We calculatedUv, and UQ for two coagulation values. In order to estimate the approximationinvolved in this treatment, we determined the corresponding Vv interaction energyby means of the exact Derjaguin eqn.(8) valid for small potentials and large %a.Table 1 summarizes these data.TABLE 1expt. no. C1, mM/1. Ka VJI x 1013 ergs u9 x 1013 ergs UQ s 1013 ergs2 165 116 4-62 5.29 5.199 115 31 0.626 0.720 0.7 10HC104Although the discrepancy between Vy and Uy exceeds 10 %, we consider that thedifference between Uv and UQ (1-2 %) gives a good idea about constant charge andconstant potential approaches. The actual difference at the interaction maximumdistance is probably not important.Dr. J. F. Padday (Kodak Ltd., Harrow) said: The papers of Ottewill et al. and ofWatillon et al. appear to calculate Hamaker constants of interaction constantsbetween polystyrene and water in a manner not in accord with Hamaker’s equations.Hamaker 4 derived the total mutual energy E, of attraction between two particles asE = X ~ Q ~ L X ~ ( X , Y ) = A f ( x y ) (1)the constant A being the non-geometric factor of the energy.V0ld,5 Fowkes,6Haydon 7 and Overbeek 8 calculated values of the A12 constant for two particles of1 Verwey and Overbeek, Theory of the Stability of Lyophobic Colioids (Elsevier, 1948), p. 143.2 Derjaguin, Trans. Faraday Soc., 1940, 36,203.3 Levine, Proc. Roy. SOC. A , 1939, 170, 145, 165.4 Hamaker, Physica, 1937,4, 1058.5 Vold, J. Colloid Sci., 1961, 16, 1.6 Fowkes, Ind. Eng. Chem., 1964, 56, 40.7 Haydon and Srivastava, Tram. Furaday Sor., 1964, 60, 971.8 Overbeek, Colloid Science, cd. Kruyt (Elsevier, Amsterdam, 1952), chap. 12.16GENERAL DISCUSSION 165material 1, embedded in a liquid, of material 2, on the basis of Hamaker's moregeneral form of eqn.(I), where A is given by4 2 = n 2 2 cqlq2A12, (2)(3)1 2which when expanded givesAI2 = Ai+A2-2JAlA2 = (Af-A$)2.Good and Girifalco 1 and others studying the wetting of flat surfaces define A12 byA12 = JAiA2. (4)A confusion in terminology arises because it has been assumed that the constant A12has some fixed value. Hamaker (p. 1062) stresses that this is not so and that the valueof A12 may vary within certain limits. Following Hamaker, the A constants for somesystems (involving the same materials) are given in table 1. Example 1 must be usedto derive A from wetting data and example 4 must be used for flocculation data. TheA12 constant of polystyrene wetted by water is derived from the surface energy ofpolystyrene is 44-0 ergsjcm2(3) and the surface energy or surface tension of water.TABLE 1system energy of van der Waals' attractionA w p 1 =- 123two semi-infinite plates of material,1, in a vacuum 127td2 1 2 4two semi-infinite plates of material1 in fluid 2 12nd2-two large spheres of material 1 at smalldistances of separation in avacuumE = -A 7tE = --- r@1+4;Az- 2w2&21A R - n2qy1R E = - - -12d 12d4 two large spheres of material 1 at small AR n2R12d 12d distances of separation in fluid 2 = - = -[dA1+qiA2-a1q2'121where R = radius of the particle; d = the distance between the two surfaces of materials1 ; q1 and q2 = the number of dispersion centres per cm3 in the respective phases and A thecoefficients of 1/r6 of the London term for attraction between two dispersion centres.The value of &, the distance between polystyrene monomer units, is 6.0 A, thereforeA1, for polystyrene, is 12 x 10-12 ergs.The total surface free energy of water is 72ergs/cm2 which with an average distance of 3.1 1 A gives A22 to be 5-13 x 10-12 ergs ofwhich the contribution from dispersion forces is 1.59 x 10-12 erg following the dataof Fowkes 2 in which the dispersion contribution to surface tension is 2143 ergs/cmz.These values of Hamaker constants are considerably larger than those obtained by thecalculations of Watillon and Joseph-Petit and of Ottewill and Shaw. If Moelwyn-Hughes 3 formula is used for calculating dispersion forces as, for instance, by Schenkeland Kitchener,4 larger values are obtained which are in much better agreement withthe wetting data.Both Watillon and Ottewill appear to make a conceptual error in their calculationof A12 by using eqn. (3) above.It is that their value of A:! (of the water-water inter-action) is calculated for dispersion forces only whereas it should include all forces1 Girifalco and Good, J. Physic. Chem., 1957, 61, 904.2 Moelwyn-Hughes, Physical Chemistry, 2nd ed. (Pergamon, London, 1961), p. 392.3 Fowkes, I d . Eng. Chem. 1964, 56,40.4 Schenkel and Kitchener, Trans. Faraday SOC., 1960, 56, 161166 GENERAL DISCUSSIONinvolving attraction. The A2 of the term, 2JA2A1, should be that of dispersion forcesonly (here defined as A!), as only these forces will interact with polystyrene to anyextent.We may then calculate A12 according toA,, = Al+A,-2JA,Agto give a value of 8.4 x 10-12 from wetting data, a recalculated value of 4-7 x 10-12ergs from the (Neugebauer) data of Watillon and Petit, and a recalculated value of4.9 x 10-12 ergs from the (Neugebauer) data of Ottewill and Shaw. These recalculatedvalues may be large as it has been assumed that the Hamaker constant A2 for watermay be calculated directly from its surface tension, an energy term, which includes ahydrogen bonding force not likely to operate in full when d is larger than lOA.Further larger corrections to their calculations should then be made to take accountof the carboxyl group on the surface of the polystyrene latex.These groups will havethe effect of increasing the effective work of adhesion term 2JAlA; by a polar-polarinteraction term and hence decreasing the value of A12 possibly even to their valuesderived from flocculation data.Dr. R. H. Ottewill (University of Bristol) said : The value of 8.4 x 10-12 ergcalculated by Padday for the Hamaker constant of polystyrene in water is the highestvalue I have seen reported for any material. Moreover, it is two orders of magnitudehigher than the value obtained by Fowkes by essentially the same method. Even ourexperimental values only scatter over one order of magnitude ! We are all well awareof the limitations of the Hamaker approach and try to compensate for these indrawing conclusions.I have, however, even more reservations about Padday’sapproach. The surface free energy of a latex particle is not known with any certaintynor is the value that one should give to the distance between two monomer unitsimbedded in a polymer matrix. Moreover Padday’s calculation is limited to inter-action in surface layers whereas the Hamaker approach is a summation in depth andtherefore considers long-range forces. From an experimental point of view the valuesuggested also appears to be highly improbable.Dr. A. Watillon (Univ. Libre, Bruxelles) said: We do not agree with Padday thatour calculations of the Hamaker constant are not in accord with Hamaker’s 1equations. We followed exactly the same treatment as the investigators he mentionsin the case where only the dispersion contribution to the attraction is considered.Nevertheless it would be better, when the orientation of the fluid molecules plays animportant role, to take it into account. The question is to estimate the importanceof this effect in our case.We first consider the Van der Waals constant of solid polystyrene for which onlythe dispersion forces contribute.Padday’s calculated value A1 amounts to 12 x 10-12ergs. Without discussing the method he used to deduce this value from the surfaceenergy of polystyrene determined by Fowkes,2 we must point out that by using eqn.(18) of Fowkes’ paper, we got A1 = 0.63 x 10-12 ergs ; this result is 20 times smallerthan Padday’s value and in good agreement with our values obtained by means of theLondon, Slater-Kirkwood and Neugebauer equations. Moreover the value of A1obtained by Schenkel and Kitchener,3 A1 = 3.17 x 10-12 ergs, can be corrected bytaking into account the molecular weight of styrene rather than benzene ; the result(Al = 1-9 x 10-12 ergs) is not very different from our first assumption calculated bymeans of the Slater-Kirkwood equation : A1 = 1-68 x 10-12 ergs.For water the1 Hamaker, Physica, 1937, 4, 1058.2 Fowkes, Ind. Eng. Chem., 1964, 56, 40.3 Schenkel and Kitchener, Trans. Faraday SOC., 1960, 56, 161GENERAL DISCUSSION 167dispersion contribution to surface tension then gives A; = 0.31 x 10-12 ergs ratherthan 1.59 x 10-12 ergs. Following Padday’s assumption that the total surface freeenergy has to be taken into account we obtain A2 = 1.04 x 10-12 ergs.By combiningA1, A2 and A: as in eqn. ( 5 ) of Padday’s remark, we got A12 = 0.78 x 10-12 ergs ; thisvalue is 10 times smaller than his value, but still 10 times larger than the values weobtained assuming dispersion forces only. As the interparticle distance at theinteraction maximum is about KHO = 1.2 we probably do not have to include anypolarization of water or hydrogen-bonding forces.If we take Padday’s A12 value in order to compare calculated Hamaker constantswith experimental data, the polar-polar interaction due to superficial ionized groupsshould be able to reduce the attraction by a factor of 100. Accordingly the pro-bability of exact agreement with the measured data seems small. Moreover this effectwould be sensitive to the nature of the ionized groups. Our experiments have shownthat it was not the case.Prof.J. Lyklema (Agric. Univ., Wageningen) said : The fundamental double-layerpotential occurring in the stability formulas is the potential $8 of the outer Helmholtzplane. The problem is that cannot be directly measured. In several cases otherdouble-layer potentials can be obtained, viz., the total potential drop $0 (with respectto the z.P.c.) and the electrokinetic potential c. Hence the central problem is torelate Il/s either to $0 or to c.The first possibility, relating +a to 5, requires a theory of the slipping process.Two such theories have been published in the literature. According to Bikerman thedifference between 5 and $8 is due to surface roughness but this idea is not easy amen-able to theoretical calculations.Lyklema and Overbeek suggested that the visco-electric effect (the increase in viscosity due to the electrical field in the double layer) isdeterminative for the slipping process. However, the theoretical elaboration of thisidea involved a questionable extrapolation and some indirect evidence suggests thatfor water the viscoelectric effect was overestimated. In conclusion, neither one of thetheories can at present be used to relate $8 to c.The determination of $8 from $0 requires a knowledge of the amount of specific-ally-adsorbed counter-ions in the Stern layer. The only system for which this hasbeen done is mercury, but as reliable data on the stability of mercury sols are scantythis fact has little practical value for stability studies.The results of Ottewill and Shaw (table 5 ) show that for polystyrene lattices inwater the $a-value calculated from log W/log c plots is not particularly accurate andhence cannot be used to deduce something on the relation between $s and c.Similar deductions can be made from the work of Watillon and Mrs.Joseph-Petit.A promising possibility to obtain more insight in this relationship from experimentsis with the AgI-system. As from the high-temperature work the specific adsorptionpotential of cations and the degree of occupancy in the inner Helmholtz plane can bederived, $6 can be calculated. A provisional comparison of $8 thus obtained with cshows their difference to be about 10 %, $8 being the higher.The conclusion is thatfor stability studies equalization of $8 and fl is the relatively best approximationavailable.Dr. A. Watillon (Univ. Libre, Bruxelles) said : We agree completely with Lyklema’scomment. Nevertheless it is not only a lack of accuracy in the log W/log C determina-tion which prevents the comparison between $6 and c. As an example, data corres-ponding to the slope of our expt. 1 of table 5 and fig. 3 give $8 = 13-3 mV when 5changes from 24.1 to 20.2 mV. The $8 obtained are in every case much smaller thanthe corresponding c. The same observations can be deduced from other experimenta168 GENERAL DISCUSSIONdata.1-3 Moreover $a is considered as constant against the electrolyte concentration.Hence we preferred to use 5 rather than the $a values obtained from log W/log Cslopes in order to determine repulsion curves.Dr.J . Gregory (University College, London) communicated : The calculation of theHamaker constant for polystyrene in water is difficult, owing to the lack of accuratedispersion data. The calculated values of Watillon and Joseph-Petit and those ofOttewill and Shaw cover the approximate range 0.4 to 5 x 10-13 erg. Consequently,the selection of a “ correct ” value from the various experimental results is not easy.An alternative approach is to use the Lifshitz theory for two bodies in a liquid.4The long-range expression for the attractive force between two semi-infinite flatplates separated by a second medium iswhere d is the separation distance (d> 2000 A), ~ 1 0 and ~ 2 0 are the static dielectricconstants of the plates and the medium, respectively, and # is a complex functionwhose limiting value of 0.35 may be assumed for the polystyrene + water system. Theexpression of Overbeek 5 for the fully-retarded dispersion energy between two flatplates may be differentiated to give the corresponding force expression :where 1 is the “ characteristic wavelength ” of the dispersion interaction. Assumingthat expressions (1) and (2) lead to the same value of the force, and inserting numericalconstants, thenP = AL/79d4 (2)The values of 610 and ~ 2 0 may be taken as the squares of the limiting refractive indicesin the visible region (i.e., assuming no contribution to the dispersion force from lowerfrequency bands).Taking refractive indices quoted by Bateman, Weneck andEshler 6 : polystyrene, nlo = 1.5683, 810 = 2.455 ; water, n20 = 1.324, 820 = 1.758 ;and inserting these values into eqn. (3), AL = 7.32 x 10-19 erg cm. If the “ character -istic wavelength ” corresponds to the dispersion frequency calculated by Ottewill andShaw (vo = 2-156 x 1015 sec-I), then 1 = 1.39 x 10-5 cm and A = 5.3 x 10-14 erg.This value is close to most of the experimental results and similar to the value 5 x 10-14erg obtained by Fowkes 7 from wetting data.Dr. A. Watillon (Univ. Libre, Bruxelles) said : In reply to Gregory, this approachis interesting and the results agree reasonably well with our experimental data.Nevertheless the calculated range is probably narrower than the gap 04-5 x 10-13ergs.Indeed this one corresponds to three different estimations. Ottewill andShaw’s calculations (A12 = 2-75-4078 x 10-13 ergs) seems to be slightly too highbecause the refractive index at the dispersion frequency of polystyrene, assumed to benon-absorbing (n = 2*0857), was used in order to calculate the static polarizability.1 Watillon and Joseph-Petit, Proc. 3rd Int. Congr. Surf. Activity, 1959, 1, 145.2 Watillon and Joseph-Petit, Abstr. A.C.S. Symp. Coagulation and Coogulant Aids, 1963, p. 61.3 Ottewill and Shaw, this meeting.4 Dzyaloshinskii, Lifshitz and Pitaevskii, J. expt. Theor. Physics, U.S.S.R., 1959,37,229 ; Soviet5 Overbeek, in Kruyt, Colloid Science (Elsevier, Amsterdam, 1952), vol.1, p. 268.6 Bateman, Weneck and Eshler, J. Colloid Sci., 1959, 14, 308.7 Fowkes, Ind. Eng. Chem., 1964, 56,40.Physics JETP, 1960, 10, 161GENERAL DISCUSSION 169From the limiting refractive index in the visible region for polystyrene instead ofa = 20-6 x 10-24 cm3 mol-1 we obtain a = 12-8 x 10-24 cm3 mol-1 which yieldsinstead of A11 = 16.8 x 10-13 ergs, obtained by the London equation, the lower valueA11 = 6.5 x 10-13 ergs. The final 4 2 result then shifts from 4.78 x 10-13 ergs to0.39 x 10-13 ergs. Furthermore in calculating A12, Ottewill and Shaw combine A11values obtained from a London equation with A22 values for water obtained from theSlater-Kirkwood and Neugebauer equations. Combination of All and A22 valuesobtained by the same type of equation gives also lower results.The second estima-tion made in our paper uses benzene data (A12 = 1-3-1-7 x 10-13 ergs) and also seemstoo high. In conclusion, our direct calculations by means of optical data of poly-styrene and water (A12 = 0.35-1.05 x 10-13 ergs) are probably the most reliable valuesfor comparison with Gregory’s calculation.Prof. B. Teiak (Zagreb, Yugoslavia) said: We have been interested in thecharacterization of a sol with respect to the Schulze-Hardy rule, and we have performeda series of experiments with monodisperse samples from the Dow Chemical CompanyLS 055 A (175 mp diameter), and some similar samples of polystyrene latex dispersionswhich we have prepared (200 mp diameter). The results are presented in fig.1 and 2respectively. The coagulation values show the nearly linear relationship and also theinfluence of the counter ion size for the monovalent species. These are in full0 I 2 3-log (N) Me(N03)nFIG. 1 .-Concentration tyndallograms for Dow latex LS 055 A showing critical coagulating concentra-tions of K, Ba and La nitrate.LS455A = 5.461 x 10-5 g/ml; H N 0 3 = 10-4 N(pH-4); Me(N0& : K+, 0 ; Ba*+, 9 ; La3+, 0 .measured after 24 haccordance with our results obtained with sols in statu nascendi. Generally, we havenever obtained any effect with sols in statu nascendi which are not in accordance withresults found in classical hydrophobic systems.For the coagulation values determined by Watillon and Joseph-Petit, the stabilityconditions were radically changed by changing the medium from acid to alkaline.Therefore, the sols should be treated as different systems with different densities ofstabilizing ionogenic groups provided there are no accompanying changes in thephysical and chemical composition.Also, the two Dow samples should differ only inparticle sizes. Calculations using concepts and parameters which assume a muchbetter knowledge of the systems are therefore not warranted170 GENERAL DISCUSSIONFrom the results of Ottewill and Shaw, we may conclude that the influence of theparticle size can not be taken as an effective parameter for the flocculation values,although the vicinity of the pH value of the media to the critical concentration for theprecipitation of Ba(0H)z or BaC03 points to the necessity for precautions.0 I 2 3--log (N) Me(N03LFIG.2.-Concentration tyndallograms and dispersoidogams (DQT) for monodispersed polystyrenelatex showing critical coagulating concentrations of Li, Na, K, Ba and La nitrate.LS-2 = 2.615X 10-5 giml; HNO3 = 10-4 N(pH-4); Me(N03):: Li+, X ; Na+, 0 ; K+, ;Baz+, 8 ; La3+, 0 ; DQT (10min) ; turbidity (24 h) - - -Dr. A. Watillon (Univ. Libre, Bruxelles) said: In reply to TeZak, for spheres,combining repulsion equation for small interactions and attraction equation for smalldistances, Reerink 1 deduced a relation between the IC value at the coagulation and thedouble-layer potential :which gives for the coagulation concentration,The Schulze-Hardy rule can be expressed when the potential is sufficiently high asexp (ve[/2kT) - 1= exp (ue[/2kT)+ 1Then, the ratio of the K corresponding to uni- and divalent ions is I C ~ / K ~ = uz/vf = 4.In our case when the potentials are not high, K I I K ~ = y t u $ / y $ v f .Using the Cpotentials corresponding to expt. 3 and 6 of our table 5 , we obtain yfv$/y$vf = 1.92.Experimentally the corresponding ratio K ~ / I C ~ was 2.35. In the limit of the approxima-tions involved, the agreement is not bad. Furthermore, the stability conditions ofour sols were radically different in various pH-media, as shown in fig. 1 correspondingto the C-pH relationship of a Dow LS 067A latex (diarn. 1.167 p) where increasingdissociation of sulphonate groups with pH can be observed.The combination ofKcoag. = 2-04 x 10-572/AU2,C1 = 3.79 x 1 0 - 2 5 ~ 4 / A 2 ~ 6 .= 1.1 Reerink, Thesis (Utrecht, 1952)GENERAL DISCUSSION 171coagulation and 5 potential information corresponding to exactly the same experi-mental conditions takes implicitly into account the variations of counter ion sizes orspecificity and of any change in superficial charge density. It was our purpose toPHFIG. 1 .-<-pH relationship of a Dow LS 067 A latex (diam. 1 a 1 67 p).check by means of this treatment if, in presence of different numbers and types ofionized superficial groups and of different counterions, the same Hamaker constantcould be obtained.Dr. G. D. Parfitt (University of Nottingham) said: Of interest in connection withthe relationship between particle size and stability are some observations Picton andI have made on the rate of coagulation of graphitized carbon blacks in aqueous solu-tions of sodium dodecyl sulphate.On two carbons, Graphon and Sterling MTG,having significantly different surface areas (Graphon 78.9 m2 g-1, Sterling MTG7.7 m2 g-I), the adsorption isotherms, put on a unit area basis, are identical. In bothcases the zeta potential, derived from electrophoretic mobility data in solutions ofvirtually constant ionic strength (0- 1 M NaCl), increases with surfactant concentrationto reach the plateau value, below the c.m.c., of 62 mV for Graphon and 74 mV forSterling MTG. From particle counts obtained as a function of time after preparationof the dispersions (using ultrasonics), stability ratios were determined.For Graphon,Wincreases with zeta potential over the range 1 to 105 and the plot of log Wagainstzeta potential shows good agreement with that predicted by the DLVO theory using aradius of 125 .$, i.e., the radius of a primary Graphon particle, and A = 5 x 10-13 erg.With Sterling MTG, W = 1 for all surfactant concentrations, as expected since thetheory predicts, for 1250 .$ radius particles, a secondary minimum of significant depthin the potential energy diagram. However, electron microscopy showed that theparticles of Graphon in the stable dispersions were aggregates of primary particleswith a mean radius of approximately lOOOA. These results indicate that in thecoagulation of aggregates in dispersions for which rca is large, it is the radius of curva-ture of the primary particle and not that of the aggregate, which determines thestability of the system1 72 GENERAL DISCUSSZONProf. M.Mirnik (Zagreb, Yugoslavia) said: According to the paper of OttewiUand Shaw, the carboxyl groups covalently linked to the particle surface cause theelectric double layer. In AgI sols, the total potential drop $0 of the double layer isderived from the potentiometric potential of the Ag-AgI electrode in a galvaniccell, and a part of the total potential drop $8 is assumed to be active at the slippingplane and to cause the electrokinetic effect. In which way was the potential $d forthe polystyrene sol derived from $0 and what replaces the total potential drop, i.e.,the surface potential $0 of the AgI system in the polystyrene sol? Why is the samepotential $8 used in the two layers ? In the former, according to the DLVO theory, thedouble layer is built up by a charge with a Stern-Gouy distribution of counter ionsand a Helmholtz distribution of constituent ions.In the latter these are only coval-ently linked carboxyl groups and counter ions. Is there any knowledge of thecarboxyl charge density, i.e., the number of carboxyl groups per unit surface and/or per particle ?In all derivations of the electrokinetic potential starting from the Helmholtz-Smoluchowsky equation this potential is proportional to the charge density. Gener-ally, in the coagulation process the amount of the adsorbed stabilizing electrolytedecreases due to the decrease of the effective specific surface.The variation of thecoagulating electrolyte must therefore, in the vicinity of the coagulating concentration,cause a gradual decrease of the specific surface and so in turn a gradual decrease ofthe charge density. This may be considered as proved for the AgI sol. The variationof the electrokinetic potentials was presumably calculated in Ottewill’s paper assuminga constant specific surface and a constant charge density?The plot of electrokinetic potential against coagulating counter ion concentrationhowever is complex since it includes the variation of dispersity and the double-layerstructure. In low concentrations the structure of the stable state is measured whereasat and above the coagulation concentration the structure of the coagulated state will beformed.Consequently, in the coagulation concentration the electrokinetic potentialcannot be characteristic for a given sol and cannot be derived theoretically from anelementary definition. 1Schulz and Teiak have compared coagulation values of aged and dialyzed AgI solsand of those in statu nascendi. The former have a particle size of 200-700 mp thelatter of about 10mp. No significant influence was observed.:! There is alsoexperimental evidence that no difference exists between the coagulation values of Ag+ions (negative limit of stability) obtained on two analogous forms of AgI 3 ; andMatijeviC, Mathai and Kerker obtained the same coagulation values of highly chargedcounter ions on aged and fresh preparations of AgI.4 Ottewill and Shaw, however,observed constant coagulation values for polystyrene latex sols, independent of particlesize.But experimental evidence suggests that the particle size is not a first-orderparameter in contradiction to the DLVO theory and to all theories that use thecontinuum flat plate or sphere condenser double layer model as its basis.Dr.R. H. Ottewill (University of Bristol) said: In reply to Mirnik and to Teiak, allour experimental work has been carried out with polystyrene latices prepared in thelaboratory using sodium laurate as the emulsifier and hydrogen peroxide as theinitiator. The excess sodium laurate remaining after the polymerization can, within1 Mirnik, Croat.Chem. Acta, 1963, 35, 217.2 Schulz and Teiak, Avhiu. kern., 1954, 26, 187.3 Mirnik, Disc. Faraday Suc., 1954,18,224,226 ; fig. 1 and 2 (the fig. should be interchanged).4 Matijevic, Mathai and Kerker, J. Physic. Chem., 1963, 67, 1995GENERAL DISCUSSION 173experimental error, be removed from the latex by prolonged dialysis. We have avoidedthe use of commercial latices since the stabilizers are not usually specified and dialysisof these materials sometimes appears to be unsuccessful. Shaw 1 has shown that thecharge on our latex particles arises from carboxyl groups on the surface which appearto be an integral part of the polymer chains. The surface charge arises from ioniza-tion of these groupings and hence the electrostatic potential at the interface isgoverned by the surface charge density of the surface.From titration studies Shawfinds approximately one carboxyl group per 250 A2 ; this figure varies with the pre-paration. The surface charge varies with the pK of the carboxyl groups and with pH ;most carboxyl groups have pK values from 4.5 to 5.0. Our work was carried out inthe pH range S-0+0.5 because (a) the mobility of the particles reaches a constantvalue in this region and does not vary with pH.2 (b) As far as we are aware the Ba2+ion does not hydrolyse in this region.We have no evidence that the surface area of the latex preparation decreases in theregion of the flocculation concentration. We would agrce with Mirnik that our presentexperimental evidence does not indicate any significant change of flocculation con-centration with particle size.Prof. E.Matijevid (Potsdam, New York) said: There is no reason to assume thatthe variations in the flocculation concentrations with variation in the size of latexparticles are real (Ottewill and Shaw, table 2). The flocculation concentrations asreported in this work are obtained by a method which involves two graphical plots,the second being a double-log plot. The best accuracy one can expect is approxi-mately 5-10 % (e.g., one need only inspect fig. 2 and see that more than one linecan be drawn in the region of the slow coagulation, giving a range of flocculationvalues). Therefore, reporting flocculation results to 0.5 % as is done in table 2 is notjustified.Thus, except for latex B and possibly E there is no difference in the experi-mental flocculation concentrations. It would be interesting to repeat the resultsusing a newly prepared latex of a- 500 A to see whether the deviation in this case isrealistic.The comparison of flocculation concentrations as is done in table 3 is of littlemeaning. The flocculation concentrations depend on a number of factors and for thesame sol will vary with sol concentrations, concentration of the potential-determiningion, mode of sol preparation, etc. Sols of greatly different chemical composition willshow different dependence on all these factors not to mention specific effects. Thus,any conclusion based upon data as given in table 3 are arbitrary.Dr. R. H. Ottewill (University of Bristol) said: In reply to Matijevit, we have notclaimed an accuracy of 0.5 % for our flocculation values. The values in table 2 arethose obtained from the experimental curves reported in the paper. We have notquoted error values since these vary with particle size according to the optical densityrange employed in the measurements. Under the worst conditions the spread was+5 %. Also the sols used were monodisperse and the experimental results werereproducible.We agree with Matijevib that flocculation concentrations can depend upon a numberof factors such as sol concentration, concentration of potential-determining ions, etc.However, the data in table 3 were chosen for sols examined at initial zeta-potentials ofsimilar magnitude and similar sol concentrations. The difference in sol concentrationsbetween the inorganic material and the organic materials is an order of magnitude. On1 Shaw, J. N., Ph. D. Thesis (Cambridge, 1965).2 Shaw and Ottewill, Nature, 1965, 208, 681174 GENERAL DISCUSSIONthe basis of the D.L.V.O. theory the chemical composition of the flocculation particleenters primarily through the Hamaker constant. Calculations tend to show, evenwhen allowance is made for their inherent inaccuracies, that the Hamaker constants forinorganic materials (e.g., silver iodide, ca. 5 x 10-13 erg) are approximately an orderof magnitude greater than those for organic materials (e.g., polystyrene, ca. 5 x 10-14erg). The trend shown in table 3 therefore could have its origin in the variation of theHamaker constant, but one would not wish to make an issue of this point in view ofthe uncertainties in the Hamaker calculation

ISSN:0366-9033    

DOI:10.1039/DF9664200164

出版商:RSC    

年代:1966

数据来源: RSC

摘要:

Coulombic and Stereochemical Factors of Colloid Stability ofPrecipitating SystemsBY Bozo TE~AK*Laboratory of Physical Chemistry, Faculty of Science, University of ZagrebandDept. of Physical Chemistry, Institute " Ruder BoSkovi6 ",Zagreb, Croatia, YugoslaviaReceived 15th June, 1966Applying the method of continuous variation of concentration of precipitating components thetypical precipitation bodies (PB) (as a plot of logarithm of concentration of the cationic componentagainst anionic one) for various precipitates are given. The boundaries of PB (precipitating/non-precipitating limits) reflect the existence of species controlling the transition from liquidus structuresto solidus phase (genotypical processes), and also phenotypical factors influencing the formation ofsub-systems.The main types of PB are named according to the dominant processes of transitionof solute species in the solid phase : (A) neutralization ; (B) ionic solubility ; (C) formation of ion-pairs and associates ; and @) unsymmetrical ionic reactivity. Examples of pure and mixed typesof PB are given. For an explanation of internal features PB, the precipitating systems are treatedas composed of five sub-systems : (i) formation of complex species (simple and monomers, as well aspolynuclears and polymers) ; (ii) embryonation (aggregates without core formation) ; (iii) nucleationand direct growth (the growing core) ; (iv) micellation (formation of primary particles with colloidindividuality) ; and (v) secondary aggregation (secondary structures of micro- and macro-particles),It is assumed that the transition from ideal liquidus structures (ILS) to ideal solidus structures (ISS)usually proceeds through stages characterized by the development of methorical structures and tex-tures (MS & T).A change of the shape of PB with the change of medium is demonstrated onprecipitating system of barium sulphate in various water+ethanol solutions. With respect to theflocculation-stabilization effects of large ions, non-ionic and ionic macromolecules on sols in statunascendi, and preformed sols, the adsorption capacity, the electrokinetic potential, and the exchangefraction of the sols are correlated with the micelle formation (surface tension) of the surface activesubstances (SAS).In general, the ways and means of characterization of precipitating system andsub-systems, with special emphasis on factors of colloid stability, are suggested and discussed.Colloid stability may be treated in a narrow or wide sense. If we started withthe interpretation of stability-instability relationship of the classical system of goldsol, then the main discussion should probably be centred on coulombic and physicalinteractions of electrolyt.es and particles following Faraday's observation 1 thatthe activity of salts cannot be explained in usual chemical terms. However, ina general analysis of colloid systems, we should consider the role of chemical inter-actions and structures of micelle, intermicellar liquid, and the critical region betweenthem 2 (the methorical3 layer).Moreover, by a wider treatment of colloid in-dividuality in dispersed as well as in condensed state, we must discuss not only colloid* in collaboration with H. Bilinski (Mrs.), M. Branica, T. Cvita3, B. Cernicki, R. DespotoviC,0. DeiZelib, N. DeZeli6 (Mrs.), H. Furedi (Miss), N. GaleSiC, M. Herak, Lj. JeftiC, J. Kratohvil,J. Mandjerelo-Radogevik (Mrs.), M. Mirnik, B. Novosel (Miss), E. Palib-Schiitz (Mrs.), N. Pavkovii(Mrs.), M. Petek (-Mrs.), J. Petres, B. Pokrit (Miss), V. Pravdib, Z. PuEar, I. Ruiid, Z. Selir, M.Slovenc (Mrs.), J. Sipalo-hljevib, D. Teiak (Mrs.), B. TomadiC, R. Wolf, M. Wrischer (Miss).17176 COLLOID STABILITY OF PRECIPITATING SYSTEMSsystems but also a number of sub-systems, where terms such as structures, texturesand architectures, and appropriate differentiation between genotypical and pheno-typical factors should be applied.Therefore there are many reasons for specifying the macroanalytical and micro-analytical distinctions of the constitutive, semi-constitutive and colligative factors 4in the same sense as we have tried to elucidate the critical phenomena of coagulationand flocculation during the General Discussion in 1954.To facilitate the discussion we are dividing the complex conceptual, experimental,and theoretical material under 10 headings, which, although different in their con-tent, may, nevertheless, represent an attempt to include in one coherent frameworkthe relatively large number of facts which should be taken into account if we wishto escape the danger of a single and partisan approach.EXPERIMENTAL BASISThe first useful generalizations in preparation of colloid systems are those of vonWeimarn,s but few people have investigated the appearance of a new phase fromsolutions in the similar systematic manner although analogous work gave a largenumber of experimental facts about melt-solid transitions.No doubt, the methodof observations of the effects of continuous variation of concentrations either ofthe main precipitating components or of the various ingredients in solutions isprobably the most powerful tool from both a practical and a theoretical viewpoint.We have used this method in macroanalytic and microanalytic sense for more than30 years,6 and now it is worthwhile to put the results from many laboratories intoan integrated scheme of colloid stability or individuality.Our method consists in observations (using preferentially the tyndallometrictechnique, but also by physical and chemical analysis of the solution and the pre-cipitate) of the changes caused in the systems obtained by rapid mixing of homo-geneous solutions containing various concentrations of components and constituentswhich may influence the appearance of new (usually solid) phase.These resultsare presented in precipitation diagrams (e.g., by a plot of logarithm of the normalor molar concentration of main cationic component against that of the anionicone), and serve as excellent means of orientation in respect not only of the geno-typical but also of the phenotypical characteristics 7 of the systems.HOMOGENEOUS SYSTEMSIt is incorrect to consider homogeneous, isotropic solutions as containing anassembly of simple molecules, atoms and ions.The ions in most cases are com-plexes where ligands are either molecules of solvent or other possible constituents.By varying the concentration the probability of finding a variety of species startingfrom the ions and molecules to individual complexes and monomers to formationof polynuclear complexes and polymers, should always be considered. When wechange the concentration of the same kind of components or constituents thesystems are not the same and sometimes not comparable. The effect may be acontinuous function, but also a discontinuous, abrupt one, and in saturated, andespecially in supersaturated, solutions the specific situations should be noted, orotherwise we make very rough simplifications.Thus, although in electrolyticsolutions we speak about simple ions and their coulombic interactions, in realitythere are a number of specific, more stereochernically accentuated, reactions leadingto a number of simple and polynuclear complexes and polymers, and these specieB. T E ~ A K 177could control profoundly the behaviour of the system as a whole or of its sub-systems. However, owing to optical homogeneity, the solutions may be treatedas representatives of ideal liquidus structures (ILS).TWO-PHASE SYSTEMSWith the appearance of the new, solid or quasi-solid, phase from solutions,there is also a chain of subsystems which cannot be formally defined in an easy manner.A direct transition from ideal liquidus structures (ILS) to ideal solidus structures (ISS)is a rare occurrence. In most cases the definition of ISS can be applied to relativelysmall parts of solids, usually of ultra-microscopical dimensions where the regularlattice pattern for constituent atoms, ions or molecules is found.Therefore, theideal two-phase concept for the precipitating systems should be always consideredas a first but inadequate approximation for the description of a number of struc-tures, textures and architectures which are more or less dynamically conditionedby many factors, some of which are genotypical, expressing strong chemical andphysical forces tending to give the most stable equilibrium states, and the pheno-typical ones reflecting the habit formation for ultra-micro, micro and macro units.THREE MORPHOLOGICAL REGIONSIn colloid chemistry and physics adequate attention was given to the structureof the boundary region between disperse phase and dispersing medium.Bothgroups of protagonists of the " chemical " and " physical " origin of charges oncolloid particles are right if we take into account marked changes in the latticewhich are caused by the adscrption of '' inert " atoms on metal surfaces as it isrevealed by low energy electron diffraction.8 In solutions also a marked rearrange-ment of crystal structure of colloid particles accompanied by special distribution ofpotential determining complexes may be expected.Such texture may influencethe structure and conposi tion of adjacent liquid layer to considerable thickness.Therefore it may be misleading to speak of ail interface as a two-dimensional surface ;instead, it would be better to use the term, methorical layer. Thus, usually thephases which meet at the critical interboundary layer may strictly be defined inthe bulk only, and the structure of this border region can be investigated by ob-servation and correlation of a large number of experimental facts determined byvarious methods which are carefully and systematically applied to all relevantchemical and physical factors. The meaning of the word ~ L E ~ O ~ L O S , 4 p~~"8opL'a (conznis,conJinis regio) indicates the significance of two or more boundaries and the regionbetween them.Thus, in precipitating systems, in addition to ILS and ISS, there are frequentlycolloid units in dispersed or condensed state which are characterized by methoricalstructures and textures (MS&T).It would be expected that in real systems allthese structures should manifest their combined mechanisms not only in the habitsof new phase but also in some macroanalytical results. Actually, proofs are givenby the types of precipitation bodies (PB) in general diagrams (plots of concentrationof one main precipitating component against the other one).FOUR TYPES OF PRECIPITATION BODIES (PB)The macroanalytical procedure for determination of the boundaries betweenprccipitating and non-precipitating systems from electrolytic solutions is easy t178 COLLOID STABILITY OF PRECIPITATING SYSTEMSperform experimentally with satisfactory precision.It is preferable to distinguishfour main types of precipitating systems; some of these types are familiar and arefound in pure or mixed forms. They may be named according to the dominantprocesses : (i) the neutralization (A-type) ; (ii) the ionic solubility (B-type) ; (iii)the formation of ion-pairs and associates (C-type) ; and (iv) unsymmetrical ionicreactivity (D-type). In fig. 1 are represented all four types of PB. The scheme(i) shows the precipitation as a neutralizatiop phenomenon where the boundariescomp[ex solubilityin excess of aniono -I -2 -3 -1 -5 -6 - 1 -e - 9 -m88M0 UI -11 -log conc.(N) cation log conc. (N) cationII09 g - 85 - 7 z - 6 4 - 5.r(nW0 - 4 0M - 3 0 - - 2I0 -I - 2 -3 - & - 6 - 6 - 7 - 0 - 9 -I0log conc. (N) cationregion of ionic.rroci.t.sc.,-I 3M0 4_Lr-11 -I? 0 - 1 -2 3 -4 -5 -6 -7 -8 -9 -10 -11 -12log conc. (N) cationFIG. 1 .-The types of precipitation bodies (PB) in general precipitation diagram : (A) the neutraliza-tion, (B) the ionic solubility, (C) the formation of ion-pairs and associates, @) the unsymetricalionic reactivity.along the equivalence line are the results of the high solubility of the complexesformed by an excess of the one or the other of precipitating ions, e.g., fluoresceinrhodamin B,9 thorium phthalate,lo aluminium hydroxide, and silver halides 11partially.The scheme (ii) is the ideal picture of systems dominated by the ionicsolubility product, e.g., strontium sulphate, barium chromate. The scheme (iii)gives the boundaries resulting from the formation of ion-pairs or other associatesbetween precipitating components or constituent, e.g., nickel dimethyl-glyoxime,and the usual coagulation phenomena. The scheme (iv) shows unsymmetricalprecipitation with anionic excess and also shows the limits controlled by complex(simple and polynuclear) formations, e.g., silver cyanideB . T E ~ A K 179complex solubility ofAgBr in excess of AgN03- 8I0 - I - 2 -3 -4 -5 -6 -7 -8 -9 -10log M (AgN03)FIG. 2.-Precipitation body of AgBr (Disc. Furuduy SOC., 1954, 18,223).violel flocks afler:I min.8 lhour l d a yc clear syslems5nc1 = red silk-like.- C sediment aner 1-2 days E3 4 2.-.1 a2ti c 3 3zIWenl-+2I2 3 4-log N (conc.AgN03)FIG. 3.-Precipitation body (general precipitation diagram : plot of AgNO3 ConcenlraLioIi againstconcentration of sodium erythrosinate)180 COLLOID STABILITY OF PRECIPITATING SYSTEMSThe diagrams of experimental systems are in most cases combinations of two,or sometimes of all, four types. Fig. 2-4 illustrate real systems: silver bromide,silver erythrosinate,lz and lead oxalate, all of which show that there are charac-teristic features inside the PB which-as more phenotypicaf ones-lead us to specialconsiderations of subsystems.log conc. Pb(NO&(N)FIG.4.-Precipitation body of PbC204 ; comparison of results from Cvitag (at 20°C, 5 min) with otherpublished results : Babkin, Zhur. Obsch. Khim., 1956, 26, 1025 ; Bottger, cit. Landolt-Bornstein,Physikalisch Chernisches, Tabellen, L[, p. 1187, 5. Aufl. (Springer Verlag, Berlin, 1923) ; Kapustinskii,Selivanova and Stakhanova, Trudy Moscow Khirn.-Tekhnol. Inst. im D. Mendeleeva 1956, no. 22, 30 ;Kolthoff, Perlich and Weiblen, J . Physic. Chem., 1942, 46, 561.FIVE SUBSYSTEMS OF PRECIPITATIONBy analogy to crystallization process from melts, it is customary also in pre-cipitation processes from solutions to distinguish the nucleation and the growthof nuclei. However, the multiplicity of factors already mentioned indicates thatit is necessary to differentiate at least five stages or subsystems: (i) the solutionalthough homogeneous is in a saturated or supersaturated state and is thus sensitiveto concentration and temperature variations in respect of resulting changes of corn-plex ions, polynuclear and polymer assemblies ; (ii) the embryos as stable or instableaggregates of precipitating species but without formation of the core with the moststable ISS; (iii) the nuclei as growing units with a core of ISS; (iv) the primaryparticles as units with relatively well expressed colloid individuality owing to thedevelopment of methorical layer with peculiar composition and structure (MS &T),but not the same as either solid (ISS) or solution (ILS); and (v) the secondarystructures which may be represented either by a sponge-like randomly aggregatedprimary particles, or by oriented growing blocks of ISS units giving mosaic crystal-line aggregates, or by recrystallized (through Ostwald’s and Tammann’s ripeningprocesses) aggregates having the appearance of homogeneous crystals.Table 1shows the composite scheme of these subsystems or transition stages from idealliquidus (ILS) to ideal solidus structures (ISS) with intermediate stages, mainlycolloid units characterized by methorical structures and textures (MS&T)B . T E ~ A K 181TABLE 1FIVE SUBSYSTEMS OF PRECIPITATION FROM ELECTROLYTIC SOLUTIONSI ILs IlDEAL LlQUlDUS STRUCTURELIQUIDSOLUTION~~~~~~~ ELEMENTAL: ATOM l:N MOLECULE - * .( I ) COMPLEX: MONONUCLEAR MONOMERSOLUTIONS+ x % SATURATED 8,SUPERSATURATEDSOLUTIONS ION-PAIR ASSOCIATE POLYNUCLEAR POLYMER0-0 t x -v?vt%(11) EMBRYOSAGGREGATESOF COMPLEXCONSTITUENTS( I l l ) NUCLEIAGGREGATESWITH PARTS OF ISS4 (GROWING CORE)-I 2 STRUCTURE8 METHoRICAL (1V) MICELLESTEXTURE (PRIMARY PARTICLESSOLS IN AGGREWTES WITH COREAND METHORICAL LAYER STATUNASCENDI- \CORE ( V ) SECONDARY STRUCTURESSPONGE- LIKE SPHERULYTIC OR REALPR ECI P I TAT ESMICRO 8, MACRO AGGREGATES CRYSTALLINE AGGREGATES CRYSTALSSTRUCTURE, TEXTURE &ARCHITECTURE( MOSAIC CRYSTALS11 s s I IDEAL SOLIDUS STRUCTURESIX GROUPS OF EXPERIMENTAL METHODSFor observations and macroanalytical and microanalytical determinations ofstructures, textures and architectures of precipitating units, there are many director indirect methods and techniques. Among them may be mentioned : (i) opticalobservations (Tyndallometric,13 absorbometric, microscopic, fluorometric 9 andothers) of precipitation kinetics 14 including the determination of dispersity ; (ii)chemical analysis of composition of solutions and precipitates (microanalytic,spectrophotometric, potentiometric,l5 polarographic,16 conductometric, radio-chemical 17 and others) ; (iii) structural analysis of precipitates (electron micro-scopic,18 X-ray, n.m.r.and others) ; (iv) electrokinetical determinations of charge effect182 COLLOID STABILITY OF PRECIPITATING SYSTEMS(electrophoretic,lg electro-osmotic 20 and others) ; (v) peptization effects 21 (withor without addition of foreign agents); and (vi) kinetics of ageing (including solu-tions 22), ionic exchanges,23 and recrystallizations (followed by chemical or physicalanalysis, radiochemical determinations and others).The systems investigated inour laboratories are usually prepared in series with continuous variation of con-centration of the main precipitating components or other ionic or non-ionic sub-stances. Those characterizations were preferred by which the system could remainin situ or with as little disturbance as possible. In this respect, the direct opticalmethods are useful and therefore the tyndallometric method which allows quickand rough dispersity determinations 13 has been widely used.SEVEN GROUPS OF EXPERIMENTAL SYSTEMSAs stated previously, the precipitating systems from electrolytic solutions maybe grouped according to genotypical characteristics under four types of precipita-tion bodies (PB). Such a division gives the general division only, but for moredetailed practical characterization we have grouped various systems as follows :(i) silver salts (halides and other salts including those of the fluorescein group);(ii) barium sulphate, strontium sulphate, lead salts, various iodates, oxalates, fluor-ides, and others; (iii) chelates and metal-organic compounds such as nickel di-methylglyoxime ; (iv) ferric, aluminium, rare earth, thorium and uranium hydroxides,carbonates, phosphates and others; (v) the effect of continuous variation of con-centration of “ neutral ” electrolytes on the precipitation or coagulation (effects onthe sols in statu nascendi); (vi) the effect of continuous variation of concentrationof ionic and non-ionic macromolecules on sols in stutu ndscendi; (vii) the systemsunder complex conditions of so-called homogeneous precipitation.EIGHT GROUPS OF CHARACTERISTIC PHENOMENASome processes, structures, or subsystems, representing coulombic or stereo-chemical interactions, are naturally more easily expressed in one experimental systemthan in another, and therefore the synoptical approach is necessary. To find thedominant mechanisms it is preferable to distinguish phenomena as follows : (i) theformation of the so-called isoelectric or equivalent precipitating systems ; (ii) theformation of sols with stability of shorter or longer time; (iii) the adsorbabilityof common and foreign ions and other species, and the formation of “potential-determining complexes ’’ on colloid particles ; (iv) the applicability of Schulze-Hardy rule for the effect of “ neutral ” ions, and the equivalency of “ gegenion ”adsorption under conditions of coagulation ; (v) the antagonistic phenomena ofelectrolyte mixtures ; (vi) the effect of dielectric constant of the dispersing medium ;(vii) the effect of large ions and ionic and non-ionic macromolecules on flocculation-stabilization relationship ; and (viii) the phenomena of ageing of solutions and pre-cipitates, exchanges and the crystallization-recrystallization phenomena. Pre-ference should be given to an attack with a variety of experimental methods andtechniques.CONCEPTS, MODELS AND THEORIESIndividual findings, experimental facts, and a large number of valuable observa-tions may be easily lost if they are not correlated by more or less satisfactory con-cepts, models, and theories.However, in a complex field where many chemical,physical and physico-chemical factors intervene, the only satisfactory way is first tB . T E ~ A K 183gather facts according to some intrinsic and extrinsic principles which cannot beneglected in any comprehensive experimental and theoretical treatment. Thus, theunderlying concepts of our approach were to follow the " repeatable" steps byapplication of the step-by-step procedure of continuous variation of concentrationof components in respect of solute and solvent. Consequently, our aim was alsoto see how far the colloid state of sub-systems of embryos, nuclei, or primary particlescould be compared with the behaviour of classical colloid sols, lyophobic and lyophillicalike (behaviour of sols in stdtu nascendi).By this approach we have found useful : (i) the concept of coulombic (ion-pairformation), and stereochemical (complex species, specific adsorbability, and forcescontrolling the formation of crystal lattice) interactions ; (ii) the concept of idealand real crystals (the role of genotypical and phenotypical factors); (iii) the con-cept of critical discontinuities caused by the formation of polynuclear and polymerunits, embryos, nuclei, micelles, primary particles, and secondary structures ; (iv)the model of distinctive sites of discrete charge distribution (potential-determiningcomplexes) in the methorical region (MS&T) between solution (ILS) and solid(ISS) phases; (v) the model of ion-pair formation as a result of coulombic inter-action between " potential-determining complexes " and gegenions in correspondingstates of critical coagulation concentration (c.c.c.) ; (vi) the model of osmotic differ-ences of inicrocomponent effect in the methorical layer and in bulk solution toexplain the aggregation of particles during crystalline growth as well as during co-agulation and flocculation; (vii) the theories of Debye-Huckel and Bjerrum of theeffect of coulombic interactions in solutions of strong electrolytes ; (viii) the theoriesof forces controlling the formation of various complex species, co-ordinations andconfigurations resulting in critical concentrations for embryonation, nucleation,crystal growth (direct and indirect), ion-exchange and recrystallization ; and (ix) thetheories elucidating the mechanism of interactions of ultramicro and micro particlesincorporating various structures and architectures.CONCLUDING REMARKSA comprehensive theory which includes all important concepts, models andpartial theories, and gives the right predictions of experimental results, shouldnecessarily be composed of so many parameters that its practical value is doubtful.Therefore, a " grand-design " approach may comprise three paths leading to someuseful general and special points of view, namely, (a) the macroanalytical proceduresusing step-by-step investigations of relatively large regions which should be consistentwith the nature of the subject and the method; (b) the microanalytical proceduresapplied to various systems which may interact between themselves producing somecomplicated but clearly definable results ; such investigations with various systems,methods and techniques should apply some step-and-repetitive-variations ; and( c ) the synthesis of results of a large number of systematically-carried out experi-ments with current concepts, models and theories in such a way that the unity ofthought, expressions and facts is preserved.In conclusion, to illustrate the value of such general statements we would liketo give one example of (a) and (b). (a) Barium sulphate is probably one of the moststudied systems ; the effect of changing the composition of water + ethanol mediais presented very well by four precipitation bodies (PB) in fig.5. The character-istics of our PB may also give the reason for the discontinuities in a plot of incuba-tion periods against the concentration which were found by several authors 25 ;fig. 6 illustrates the results of Nielsen.2184 COLLOID STABILITY OF PRECIPITATING SYSTEMS60 . 8 W z51- - 3l --log N (conc. Ba(NO3)2)I - 1 -3 -5 -7log N (conc. Ba(N03)~)_ _Ih3456-P -hETHANQ-,234561 - 1 -3 - 5 - 7log N (conc. Ba(N03)2)I - I -3 - 5 -7log N (conc. Ba(N03)~)FIG. 5.-Effect of media (waterSethano1) on PB of BaS04 (at 20°C, 1 h) ; broken line inside PBgives the change in precipitation rate : (i) water, (ii) 20 % ethanol in water, (iii) 30 % ethanol inwater, and (iv) 40 % ethanol in water.FIG.6.-(from Nielsen, Acta Chc:n. Scand., 1961, 15,441). The following quantities are shown as functionsof the initial concentration c of barium sulphate. (1)number concentration of crystals formed ; (2) inductionperiod ; (3) quotient of the former two ; for c>O.O1M (log c> -2) this is assumed to equal the rate ofhomogeneous nucleationB . T S ~ A K 185(b) As an example of the correlation of various systems having a close corres-pondence, we may consider the behaviour of silver halide sols in statu nascendi,and the behaviour of hydrophobic sols in general.In respect to interactions be-tween sols and macromolecules (ionic and non-ionic), fig. 7 shows critical changes700-30 JTRITON X-305 1 31 I 13 4 5 6 7-log CM SASFIG. 7.-The influence of surface-active substances (SAS) Triton x 305 (T) and n-myristylaminenitrate (M) on : A, surface tension (G in dynes/cm) of SAS solution ; B, adsorption of Eu3+ on theAgI sols (y in mg equiv. Eu/gr mole AgI) ; C, fraction exchange for I- (F,) ; D, electrokineticpotential (E in mV) ; E, intensity of scattered light of the AgISI- system (5x 10-4 M AgI). Thecoagulation region corresponds to the maximal change in adsorption (B), exchange (C) and F potential(D). The stabilization of the AgI sols (E) probably could be ascribed to polymerization of SAS (A).occurring in pure and mixed systems with a change of surface-active agent con-centration.The concepts, models and theories we have developed27 as well as those putforward by others should be evaluated according to the rule: the observationleads, the experiment decides, and the theory describes.1 Faraday, Phil.Mag., 1857, ser. 4, 14, 512.2 Guggenheim, Discussion on Surface Chemistry (Bordeaux, 1947) (Butterworths, London, 1949),3 Ostwald, Kolloid-Z., 1942, 100, 2, footnote 40. Teiak, Arhiv. kern.. 1949. 21. 93.p. 11.4Teiak et al., Disc. Faraday Soc.,-1954, 18, 63, 194, 199, 214,221, 223, 313, 367; Mirnik, ibid.,204, 207, 224, 227.5 Von Weimarn, Kolloid-Z., 1908, 2, 301 ; Chem. Rev., 1925, 2,217.6 Teiak, Glas.Hem. DruJtva Kr. Jugosl., 1932, 3,147 ; 1933, 4,137 ; Kolloid-Z., 1934, 6$, 60 ;Z . physik. Chem. A , 1935, 175, 219, 284; ibid. B, 1936, 32, 46, 52; ibid. A , 1942, 190, 237;1942,191,270 ; 1943,192,101186 COLLOID STABILITY OF PRECIPITATING SYSTEMSNiggli, Lehrbtich der Mineralogie und Kristallchemie, 3rd ed., Teil I (Gebriider Borntraeger,Berlin-Zehlendorf, 1941), p. 450,535,570.8 May, Ind. Eng. Chem., 1965,57, 7, 19.9 Teiak and Teiak, Croat. Chem. Acta, 1964,36, 59.10 Bilinski, Ph.D. Thesis (Zagreb, 1964).11 Teiak, Disc. Faraday Soc., 1954, 18,223.12 PaliC-Schiitz, Teiak and Teiak, Croat. Chem. Acta, 1964, 36, 133.13 Teiak, Kolloid-Z., 1936,74, 16; 2. physik. Chem. A, 1936, 175, 284.14 Kratohvil, Teiak and Vouk, Arhiu. kern., 1954, 26, 191.DeieliC, DeieliC and Teiak, J.Colloid Sci., 1963, 18, 888.Kratohvil and Teiak, ibid., 1954,26,243. Herak, Herak, Teiak and Kratohvil, ibid., 1955, 27, 117. Cernicki and Tekk, Croat.Chenz. Acta., 1956, 28, 175; 1957, 29, 7. Kratohvil and Teiak, ibid., 1957, 29, 63. Hecak,Herak, Kratohvil and Teiak, ibid., 1957,29,67; Wolf and Teiak, ibid., 1957,29,461. Cer-nicki and Teiak, ibid., 1958, 30, 1. Biffl and Teiak, ibid., 1958, 30, 9. DeieliC and Teiak,ibid., 1958, 30, 119. MatijeviC and PavkoviC, Kolloid-Z., 1958, 159, 1. MatijeviC and ZoriC,ibid., 1958,161,97. Teiak, Fiiredi and Branica, Proc. 2nd Conf: Atomic Energy, 1958,28,250.15 Mirnik and Teiak, Arhiv kem., 1951, 23, 1. Mirnik and DespotoviC, Croat. Chem. Acta 1960,32,139. Mirnik and Teiak, Arhiu kem.,1949, 21, 109; 1951, 23, 44, 59.Mirnik, FlajSman, Schulz and Teiak, J. Physic. Chem.,1956, 60, 1473. Mirnik, FlajSman, and Teiak, Kolloid-Z., 1962,185, 138.16 PravdiC, Branica and PuEar, Electrochem. Technol., 1963,1,312. Branica, PravdiC and PuTar,Croat. Chem. Acta, 1963,35,281. Petek and Branica, J. Polarogr. Soc., 1963, 9, 1. JeftiC andBranica, Croat. Chem. Acta, 1963, 35, 203, 211. Branica and PravdiC, Proc. 3rd Int. Con$Polarogr., 1964, 435. Petek, JeftiC and Branica, ibid., 1964,491.17 Herak and Mirnik, Kolloid-Z., 1960, 168, 139 ; 1961, 179, 130 ; Croat. Chern. Acta, 1962, 34,153 ; Kolloid-ZL, 1965, 205, 129 ; Croat. Chem. Acta, 1965, 37, 79.18 Wrischer and Cernicki, Croat. Chem. Acta, 1958, 30, 163. Herak, Kratohvil, Herak andWrischer, ibid., 1958, 30, 22. Mirnik, Strohal, Wrischer and Teiak, Kolloid-Z., 1958, 160,146.Mirnik and Teiak, Trans. Faraday Soc., 1954, SO, 65.19 Mirnik, FlajSman and Teiak, Croat. Chem. Acta, 1956, 28, 167.20 PravdiC and Mirnik, Croat. Chem. Acta, 1958, 30, 113.1958, 30, 207.32, 1, 75. Pravdid, ibid., 1963, 35, 233. PravdiC, JoviC and Mirnik, ibid., 1963, 35, 239.Mirnik, PravdiC and MatijeviC, ibid.,Mirnik and PravdiC, ibid., 1958, 30, 213. PravdiC and Mirnik, ibid., 1960,21 Teiak, Arhiv kern., 1951, 23, 30.22 Teiak and Wolf, Arhiv kern., 1953, 25, 39. MatijeviC and Teiak, J. Physic. Chem., 1953, 57,951. Teiak, MatijeviC, Kratohvil and Fiiredi, Proc. Sympos. Coord. Chem. (Copenhagen,1954), p. 92.23 Mirnik, Kolloid-Z., 1959, 163, 25. Mirnik andDespotoviC, ibid., 1962,180, 51 ; Croat. Chem. Acta, 1961, 33, 107. Wolf, Mirnik and Teiak,Kolloid-Z., 1965, 205, 111, 118. DespotoviC and Mirnik, Croat. Chem. Acta, 1965, 37, 155,163,169.MatijeviC and Janauer, J. Colloid Interface Sci., 1966, 21, 197.Mirnik and VlatkoviC, ibid., 1959, 163, 32.24 Watson, Cardell and Heller, J. Physic. Chem., 1962, 66, 1757.25Nielsen, J. Colloid Sci., 1955, 10, 578, fig. 3.26 Nielsen, Acta Chem. Scand., 1961, 15, 441.27 Teiak, Kolloid-Z., 1932,59, 158 ; 2. physik. Chem. A , 1942,191,270 ; 1943,192, 101 ; Arhiv.kern., 1947, 19, 9, 19; 1948, 20, 15; 1949, 21, 96; 1950,22, 26. Mirnik, Nature, 1961, 190,689 ; J. Physic. Chem., 1961, 65, 1635 ; Croat. Chem. Acta, 1962, 34, 97 ; Kolloid-Z., 1962,185, 66 ; 1963, 35, 217 ; Nature, 1963, 199, 555 ; Kolloid-Z. u. Z. Polymere, 1965,205,97

ISSN:0366-9033    

DOI:10.1039/DF9664200175

出版商:RSC    

年代:1966

数据来源: RSC

摘要:

Stabilization of Lyophobic Colloids by Hydrolyzed Metal Ions*BY EGON MATIJEVI?, STANKA KRATOHVIL AND LYNDEN J. STRYKER~Dept. of Chemistry and Institute of Colloid and Surface Science,Clarkson College of Technology, Potsdam, New York, U.S.A.Received 7th June, 1966The coagulation and reversal of charge of silver bromide and silver iodide sols (aged and in statunascendi) by chromium(II1) nitrate and hafnium tetrachloride has been investigated as a functionof pH. Turbidity, mobility and adsorption measurements were performed and the results inter-preted in terms of the hydrolysis of chromium(I1I) and hafnium ions.The numerous experimental studies of the stability of lyophobic colloids in thepresence of electrolytes deal mainly with physical parameters of the systems and aredesigned to test various theoretical models.Few meaningful investigations havebeen made in which the sol stability is related to the chemical properties of thecoagulating and the stabilizing ionic species. The chemical changes of these ionsexert profound effects upon the stability of lyophobic colloids which, in manycases, may greatly exceed the phenomena caused by variation of physical properties.The chemical aspects of coagulation were recently discussed by Stumm and Morgan1while a number of studies were carried out in our laboratory in which the stabilityof lyophobic colloids in the presence of various complex forming ions has beensystematically investigated. In particular, the polyvalent metal ions are susceptibleto complexing either by co-ordination with anions or by hydrolysis.The latterprocess influences the stability of negative sols in two ways: (d) the hydrolysisis almost invariably associated with a change in the charge of the cationic (coagulating)species, and (b) the hydrolyzed species are responsible for the charge reversal.If all the chemical and thermodynamic information (i.e., the type and the con-figuration of the hydrolyzed species and their corresponding hydrolysis constants)were known, one should in principle be able to predict the effects of a given poly-valent metal ion upon a lyophobic sol as a function of pH and salt concentration.Despite a considerable amount of work on hydrolysis, in many cases there is littledefinitive information on hydrolyzed species. Even the most sophisticated methodspresently in use are not sufficiently sensitive to detect extremely small quantities ofa hydrolysis complex, especially in mixture with large concentrations of other species.If undetected ions are present as polymers of high charge, these species may havegreater effect upon sol stability than the rest of the ionic medium.Conversely,studies of sol stability in such media can be used as a method for detection of thehighly charged hydrolyzed species. The coagulation method has been used suc-cessfully in the elucidation of the hydrolyzed species when the hydrolyzed productof a metal ion is essentially monodisperse, e.g., for aluminum.2-4 When thehydrolysis process is complicated (i.e., for ferric salts) no relationship between* supported by the U.S.Army Research Office (Durham), Grant No. DA-ARO-D-31-124-G656.1- The support of L. J. S. by a NASA Traineeship is acknowledged.18188 STABILIZATION OF LYOPHOBIC COLLOIDSthe presently known composition of the electrolytic solution and sol stability couldbe established,s indicating inadequate knowledge of the specific hydrolysis system.As a further example of the relationship between the stability of a lyophobicsol and the chemical constitution of the electrolytic medium, the coagulation andthe reversal of charge by chromium(II1) and hafnium ions, are discussed in thispaper, and related to the nature of the metal ion species. In addition, electro-kinetic mobility of sol particles is given as a function of concentrations of chromium(111) and hafnium salts over a range of pH values.Since the hydrolyzed ions arepreferentially adsorbed by the colloid,69 7 the mobility of the latter is strongly in-fluenced by the composition of the electrolytic s0lution.4~ 5,79 8 Sols of reversedcharge sometimes show considerably higher mobilities than the original sols ofthe opposite charge. In this paper the relationship of the mobility to sol stabilityin systems containing hydrolyzed ions is discussed. Except at the zero point ofcharge, no correlation of mobility to stability could be established, and thereforeelectrophoretic mobility cannot be used as a criterion for sol stability. Finally,the adsorption of hafnium on a silver iodide sol was measured as a function of pHusing a radioactive tracer technique.The relationship of adsorption to stabilityand mobility is discussed.EXPERIMENTALSilver iodide and silver bromide sols, aged and in statu nascendi, were used. The criticalcoagulation concentration (c.c.c.) and the critical stabilization concentration (c.s.c.) weredetermined from turbidity measurements as described previously.4~ 9 Aged sols wereprepared according to Ottewill et aZ.10 Electrophoresis measurements were carried out in aMattson type cell.1~~ 3The adsorption of Hflg1 on silver iodide sols was measured with a Tracerlab GammaGuard Fully Automatic Well Scintillation Console System. In a given run a number ofsystems were prepared in test tubes containing a final volume of 25 ml of a silver iodide solin statu nascendi (1 x 10-3 M AgI, 4 x 10-3 M excess I-) and a constant concentration ofHfC4 to which a known amount of Hf181 was added.The pH in each test tube was system-atically varied using HNO3 or KOH. The systems, so prepared, were allowed to stand for12-1 5 h and then centrifuged at 20,000 rev/min in a Beckman Preparative Ultracentrifugemodel L-2. An aliquot of the supernatant solution from each test tube was analyzed forradioactivity and compared with activities of equal volumes of known isotope solutions.The centrifuged solid was washed with approximately 5 ml of a 4 x 10-3 M KI solution,centrifuged again, the supernatant removed, and the activity of the solid determined. In allcases the adsorption of hafnium on the ultracentrifuge tubes was measured separately andtaken into account in the calculation of the final adsorption results. The pH measurementswere made using calibrated glass electrodes and a Beckman model G pH meter.All chemicals were of the highest purity grade.Chromium(II1) nitrate was a violet salt.Fresh stock solutions were prepared from solids at least every 10 days. No aging effects atroom temperature were observed even when compared with solutions kept at 3°C during thisperiod of time. " Aged " solutions of chromium(II1) nitrate were obtained by heating2 x 10-4 M solutions for 4 days in glass stoppered containers. Solutions heated longer showedno more change in the C.C.C. or C.S.C. This aging time needed to reach an " equilibrium " inhydrolyzed chromium(II1) salt solutions is consistent with observations made by Hall andEyring 12 (other workers have indicated that equilibrium is not established by refluxing solu-tions for 27 13 or even 133 days 14).At the concentrations employed the chromium(II1) salt solutions were colorless.However, it is assumed that " fresh " and " aged " solutions correspond to the well knownviolet and green solutions, respectively.Aqueous solutions of hafnium tetrachloride aremuch more stable at room temperature and no aging was detected in a period of 6 months.The chloride salt was employed since the isotope was available in this form. Small amountsof chloride ions do not affect the properties of silver bromide or silver iodide s0ls.lE. MATIJEVIC, s.KRATOHVIL AND L. J . STRYKERRESULTS189In solution the relative aniounts of different chromium species change with pH, age,and temperature. One would expect these changes to affect the interaction betweenthese solutions and a lyophobic colloid. Fig. 1 represents the " [Cr(NO3)3]-pM "I, ___2 0 3.0 40 5.0 60- 50P H3FIG. 1.-The log [Cr(N03)3]-pH domain for a silver bromide sol in statu nascendi at 25", usingfresh solution of chromium(II1) nitrate. Shaded area represents the coagulation region. Emptycircles give the critical coagulation concentrations (c.c.c.) and the blackened circles the criticalstabilization concentration (c.s.c.). Concentrations : AgBr 1.0 x 10-4 M ; pBr- 3.4.STA BI L I ZA'TION REGION b3FIG. 2.-The log [Cr(NO&] - pH domain for a silver bromide sol in statu nascendi using an agedchromium (In) nitrate solution.0, c.c.c., 0, C.S.C. Concentrations AgBr 1.0 x 10-4 M ; pBr- 3-4.Chromium (111) nitrate solution aged for 4 days at 90". Coagulation experiments were carried outat 25°C190 STABILIZATION OF LYOPHOBIC COLLOIDSI IFIG. 3.-The log [HfC14]-pH domain for an aged silver iodide sol using a fresh solution of hafniumtetrachloride at 25°C. 0, @, C.C.C. ; D,., C.S.C. ; A,A, ZPC (zero point of charge). Concentra-tions : AgI 5.0 x 10-5 M ; PI- 3.6. Empty symbols represent experiments carried out with a gradientof Hfcl4 and blackened symbols were obtained using a gradient of pH.-2.01 -4.0 -I I-40 1 I40 t- I I-4ok I I II .o 30 50 7.0PHFIG. 4.-Mobilities of an aged AgBr sol (AgBr 1.0 x 10-4 M ; pBr- 3.4) in presence of four differentconcentrations of chromium(I1I) nitrate as a function of pH. Squares denote the mobilities of thesame sol in absence of Cr(NO&.Corresponding values for C.C.C. and C.S.C. are indicated by arrows.Hatching represents the coagulation rangeE. MATIJEVIC, s. KRATOHVIL AND L. J . STRYKER 19 1domain for a negative silver bromide sol in Stdtu nascendi using freshly pre-pared solutions of chromium nitrate. Empty circles give the C.C.C. as obtainedfrom experiments using a gradient of Cr(N03)3 at approximately constant pH,and blackened circles give the C.S.C. determined from experiments in which theconcentration of the chromium(II1) salt in a series of test tubes was kept constantand pH varied systematically.Shaded area shows the coagulation region. Atlower pH values (< -4.4) no reversal of charge takes place regardless of the con-centration of chromium salt.Fig. 2 gives the results obtained with the same sol but with solutions of chromiumnitrate aged for 4 days at 90". The solutions were then cooled to room temperature- 4 0 1 "I-444-PHFIG. 5.-Mobilities of a silver iodide sol in statu nascendi (AgI 1.0 x 10-4 M ; PI- 2-7) in presence offour different concentrations of HfC14 as a function of pH.and all experiments carried out at 25°C as before. There was a tremendous effectof aging upon the coagulation and reversal of charge properties of these solutions.Aged solutions reverse the charge over the entire pH range (squares) at concentra-tions which are approximately 10 times smaller than the coagulation concentrationsof the fresh solutions.The C.C.C. boundary of aged solutions is shifted to consider-ably lower concentrations of the chromium salt. Below the dashed line the repro-ducibility is poor but even the highest values for C.C.C. are for at least two orders ofmagnitude below the C.C.C. of fresh solutions.Fig. 3 presents results obtained with fresh solutions of hafnium chloride usingan aged silver iodide sol. Solutions containing hafnium species reverse the chargeof AgI particles over the entire pH range. At higher pH values there is a shar192 STABILIZATION OF LYOPHOBIC COLLOIDSincrease in the C.S.C. indicating that more hafnium salt is needed to reverse the chargethan at lower pH.Empty and full symbols designate experiments carried out witha gra.dient of hafnium chloride and pH, respectively. The agreement is good.Triangles denote the zero point of charge as determined from electrophoretic measure-ments.Fig. 1-3 were obtained from turbidity data. In order to explain the results,electrophoretic measurements were performed on various sols. The concentrationof the coagulating electrolyte was maintained constant and the pH was varied.This corresponds to measuring mobilities at a particular horizontal cross-sectionof fig. 1-3. The concentration of coagulating electrolyte was changed and successivecross-sections investigated.5 -2.0 -W -3 3 -40 "' -I 1PHaged AgI : 5.0 x 10-5 M PI- : 3.6FIG. 6.-Mobilities of an aged silver iodide sol (AgI 5-0 x 10-5 M ; PI-3.6) in presence of five differentconcentrations of HfC14 as a function of pH.Blackened circles denote duplicate runs. Squaresare mobilities of the same sol in absence of HfC14.Fig. 4 presents the mobility against pH of an aged negative silver bromide solin the presence of chromium nitrate solutions which were aged independently for4 days at 90°C. Again the actual measurements were carried out at 25°C. Mobil-ities of the sol in the absence of the chromium salt are given by squares. Shadingsrepresent the coagulation region as obtained from turbidity measurements on thesame systems. The stabilization (c.s.c.) and coagulation (c.c.c.) limits are indicatedby arrows.The unique effect is that the sols are of reversed charge (positive) atlower pH values and become negative again at higher pH. This is the oppositeof what is normally expectedE . M A T I J B V I ~ , s. KRATOHVIL AND L. J . STRYKER 193Fig. 5 presents similar results using freshly prepared hafnium chloride solu-tions and a negative silver iodide sol in statu nascendi. Again each curve representsmeasurements in which a constant amount of HfC14 was added but the pH wasadjusted systematically by addition of HNQ3 or NaOH. Essentially the sameeffects are observed as with chromium nitrate solutions.Fig. 6 is based upon measurements on an aged negative silver iodide sol whosemobilities are given by squares. To indicate the reproducibility of the results,blackened points are for duplicate runs made with a second sol preparation andanother hafnium chloride solution.It is apparent that aged sols and sols in statunarcendi give the same general effects although in detail the behaviour is different.Fig. 3-6 show that in all cases coagulaiion takes place at about the electrophoreticzero point of charge (z.P.c.). However, there appears to be no correlation betweenthe mobility and the extent of the coagulation range or the mobility and the C.S.C.and C.C.C. values beyond the Z.P.C. It is unexpected that at higher pH values thereversal of charge does not occur. This would seem to contradict the claim thathydrolyzed species are responsible for this phenomenon.DISCUSSIONCOAGULATION AND REVERSAL OF CHARGE BY CNROMIUM(III) NITRATEThe chemistry of aqueous chromium(lP1) solutions has been studied extensivelybecause of its application to the tanning of leather.A review of possible complexspecies due to hydrolysis of chromium(II1) salts was given by Stiasny 16 althoughthe formulation of the majority of the complex ions in his papers was essentiallybased upon speculation and only indirect evidence.Recently, the introduction of new experimental techniques, such as ion-exchangechromatography 13% 1 7 , 1 8 and conductometric titration with ammonium molyb-date,l2 has led to a more specific formulation of the complex chromium species.The nature of the species which are present depends on the treatment of the chromiumsalt solutions, i.e., on the mode of addition of base, the time and temperature ofaging, the concentration of the stock solution, etc.In summary, it appears thatthe addition of a base to a fresh solution of a chromium salt, which does not con-tain a “ penetrating” anion, rapidly leads to the formation of CrOH2-t species.(Water of hydration is not considered in this and the subsequent discussion.) Slowaging of the solutions at room temperature or forced aging at elevated temperaturesresults in the formation of olated polymeric species. Most experimental evidencesupports the conclusion that at first only dimers 129139 17-20 [(CrQH):+ or Cr2QHs+]and trimers 12918 [Cr3(QH)i+ or Cr3(OH)i+] are produced, which upon prolongedaging and in the presence of higher concentrations of base may further polymerize.We will attempt to explain the coagulation and the reversal of charge in terms ofthese species. Since nitrate ion does not complex with chromium,l2 only hydro-lyzed species need to be considered.Fig.1, which represents data obtained with freshly prepared solutions ofchromium nitrate, can be interpreted assuming that no polymerization has takenplace in the solution. At the low pH values (< 3), the C.C.C. corresponds to that oftrivalent ions21 and no reversal of charge takes place. This would then confirmthe existence of Cr3f as the predominant species. At higher pH values the C.C.C.increases somewhat indicating the formation of species of lower charge. Thehydrolysis constant for CrQHzf in the blue chromium(ll1) solutions was given as9 x 10-5 and 1.6 x 10-4, respectively.% 23 Using the first value, it can be calculatedthat at the C.C.C.at pH 3 -9 and at pH 5 -90 % of total chromium should exist194 STABILIZATION OF LYOPHOBIC COLLOIDSin the hydrolyzed form. If this were the case the C.C.C. would have to be somewhathigher at pH>4. Coagulation data would indicate that at pH 4-55 -70 %of chromium is in the hydrolyzed form. It appears that the C.C.C. of freshly pre-pared solutions of chromium salts can be explained assuming only the first hydro-lysis step but a smaller hydrolysis constant. This is in line with observations ofLaswick and Plane13 who found that even after prolonged refluxing of chromiumsalt solutions more than 70 % of chromium remains in the unhydrolyzed form.Once sufficient CrOH2+ is present in the solution, the charge is reversed due to the ad-sorption of hydrolyzed ions on sol particles.This is represented by blackened circles.Chromium nitrate solutions aged at elevated temperatures show a tremendousdifference in their interaction with the same lyophobic colloid (fig. 2). The C.C.C.is considerably lower and the solutions reverse the charge over the entire pH range.All this indicates the presence of polymerized chromium(lI1) species of highercharge. At pH<3.5 complex ions of charge higher than +4 must be present.The species Cr2QH5+ and Cr3(0H),7+ as suggested by Hall and Eyring 12 andCr3(0H)z+ as given by Finholt 18 could well account for the low C.C.C. BetweenpH 3-5 and 5.5 the C.C.C.remains essentially constant at a value corresponding toa +4 charge species, Thus, it appears that over this pH range CrZ(0H):f is thepredominant species. This is reasonable because this ion contains more OH perCr than the complexes assumed to exist at lower pH.The reversal of charge contributes additional evidence for the presence of complexspecies over the entire pH range. As mentioned, the increase in the C.S.C. at higherpH seems at first unusual. This can be explained if one assumes the existence ofsoluble uncharged species such as monomeric or polymeric Cr(OH)3 at higher pHvalues. Molecular species of this kind may adsorb on colloid particles but cannotreverse the charge. Mobility measurements as well as adsorption of hafniumsupport this assumption.The electrophoresis measurements were carried out with aged chromium solu-tions and charge reversal is found at lower pH at all concentrations of chromium(II1)nitrate used in these experiments.At higher pH values the sol remains negative,which means that chromium species have lost the ability to reverse the charge. Thisis only possible if the hydrolyzed species are neutral. A decrease in mobility withpW was observed with thorium 7 but that was at least partially due to the formationof a thorium hydroxide precipitate. The solubility product24 of Cr(OH)3 isestimated to be -10-30 and all C.C.C. data were obtained in unsaturated chromiumhydroxide solutions.COAGULATION AND REVERSAL OF CHARGE BY HAFNIUM TETRACHLORIDEHafnium ions appear to be hydrolyzed except at very high dilutions and extremelyhigh acidities.259 26 The ultracentrifugation method supplied considerable evidencethat the hydrolysis species (at HClO4 > 0.2 M) are monodisperse, either trimericor tetrameric with + 1 charge per monomer unit.27 Other experiments have ledto essentially the same results although the Hf: OH ratio varies in the suggestedcomplex species.28-30 The degree of polymerization decreases with dilution andincreasing acidity.There are no reports on the type of hydrolyzed species presentin solutions of very low concentrations and intermediate pH values-the con-ditions which are of particular interest in coagulation studies.The strong hydrolysis of hafnium solutions is confirmed by the reversal of chargewhich takes place over the entire pH range used in the experiments (fig.3). TheC.C.C. for tetravalent ions for a similar silver iodide sol is - 10-6 M.21 Since thE. MATIJEVIC, S . KRATOHVIL AND L. J . STRYKER 195C.C.C. of hafnium is considerably below this value * at pH<6, the hydrolyzed speciesmust be of charge higher than +4. The continuuus increase in C.C.C. with pHsuggests that more than one species is present and that at higher pH a complexof lower charge or an uncharged molecule is formed. Electrophoresis experimentsindicate that the latter is the case. As shown in fig. 5 and 6 in all instances thesol remains negative at higher pH values and no reversal of charge is observed.The only explanation is that under these conditions a neutral species is formed.The existencc of soluble Hf(OH)4 has been suggested by Peshkova and Ang.30 Theyalso gave the hydrolysis constants for species HfOH3+, Hf(OH)$ +, Hf(OH)l, Hf(OH)4,Hf3(OH)z+ and Hf4(0H): + as follows : K1 = 1-33, K2 = 0-59, K3 = 0.38, K4 = 0.30,K3?4 = 2 .3 4 ~ 104, and K4,8 = 1-01 x 108. Using the first four constants we havecalculated the composition of the solutions along the C.C.C. curve. A few examplesare given in table 1. Similar results were obtained for the concentrations along theTABLE 1 .---COMPOSITION OF SOLUTIONS OF HfC14 ALONG THE C.C.C. CURVE (fig. 3) AS CALCULATEDUSING HYDROLYSIS CONSTANTS GIVEN BY hSHK0VA AND ANG30IHfC141tot PH %IHf4+l %WfoH3+1 %IHf(OH)pl %tHf(OH):I %[Hf(OH).+]2-88 x 10-7 2-80 0 0 0 0.20 99.804-17 x 10-7 4.00 0 0 0 0.01 99.995-89 x 10-7 5.00 0 0 0 0 1001.00 x 10-6 6.00 0 0 0 0 1002-51 x 10-6 6-50 0 0 0 0 100C.S.C.curve and also when the constants K3,4 and &,8 were considered. Althoughthe conditions under which these constants have been determined were differentfrom those used in our experimmts, the above calculations indicate that hafniumis predominantly present in the molecular form. At least a small amount of a highlycharged species (such as Hfd(OH):+) must still exist in these solutions which is re-sponsible for the coagulation and the reversal of charge effects. Only 1/10 % orless of the total concentration of Hf in the form of such highly charged complexeswould be needed to coagulate and/or reverse the charge of the sol.The solubilityproduct constant 28 for Hf(OHJ4 is given as 4 x 10-26 which means that all of thehafnium over the entire domain in fig. 3 is present in soluble form.Table 2 gives adsorption data on a silver iodide sol as obtained with four differenthafnium tetrachloride concentrations over a range of pH. The results are expressedTABLE 2.-ADSORPTION OF HAFNIUM IONS ON A SILVER IODIDE SOL in Stall4 nascendi (AgI 1 '0 X M , PI- 2.4) AT DEFERENT pHFOR FOUR DIFFERENT CONCENTRATIONS OF ADDED HfC14[HfC14]tot pH 1.64-.69 2*00-*02 2*48-*50 2.92-3.00 3*42-.50 4*18-*32 5*06-*18 5*68-*92 6.60-*705 X 10-5M Hf,&/Hftot 0.45 0.25 0.34 0.41 0.65 0.83 0.97 0.98 1-001 X 10-5M Hfad,/Hftot 0.93 0.96 098 1.00 0.99 0.98 0.96 0.95 0.805 X 10-6M Hf&Hftot 0.99 0.98 1.00 0.97 0.98 1.00 0.92 0.91 0.862X 10-6M Hfads/Hftot 1.00 1.00 0.99 0.97 0.88 0.95 1.00 0.99 1.00as ratio of the adsorbed to the total added hafnium ion concentration. Except atthe lowest pH values and at the highest concentration of HfC14 (5 x 10-5 M) allof the hafnium is adsorbed on the silver iodide sol. This confirms that the hydroxylgroup in the hydrolyzed species is responsible for the adsorption.If the hydro-lyzed species are ionic, the reversal of charge may take place. Neutral species* The polymerization of the species means a decrease in actual molarity of the coagulating speciesby a factor which is equal to the number of monomers in the complex. Therefore, the C.C.C. infig. 3 are in reality smaller than indicated by the ordinate196 STABILIZATION OF LYOPHOBIC COLLOIDSadsorb equally strongly but cannot restabilize the sol by charge reversal.Theadsorption results support, therefore, the conclusions made in the discussion of theturbidity and mobility data.Our examples show that the coagulation and reversal of charge phenomenacan be explained on the basis of the hydrolysis of counterions.1 Stunm and Morgan, J. Amer. Water Works Assoc., 1962, 54,971.2 Matijevic and Teiak, J. Physic. Chem., 1953, 57, 951.3 Matijevik, Mathai, Ottewill and Kerker, J. Physic. Chenz., 1961, 65, 826.4 Matijevid, Janauer and Kerker, J. Colloid Sci., 1964, 19, 333.5 Matijevii: and Janauer, J. Colloid Interface Sci., 1966, 21, 197.6 MatijeviC?, Abramson, Ottewill, Schulz and Kerker, J.Physic. Chem., 1960, 64, 1157.7Abramson, Jaycock and Ottewill, J. Chem. Soc., 1964, 5034, 5041. Herak and Mirnik,8 Matijevid and Stryker, J. Colloid Sci., 1966, 22, 68.9 Teiak, Matijevib and Schulz, J. Physic. Chem., 1951,55, 1557 ; Matijevik and Kerker, J. Physic.10 Ottewill and Rastogi, Trans. Faraday SOC., 1960, 56, 866 ; Horne and Ottewill, J. Phot. Sci.,11 Mattson, J. Physic. Chem., 1928, 32, 1532; 1933, 37, 223.12 Hall and Eyring, J. Amer. Chem. SOC., 1950, 72, 782.13 Laswick and Plane, J. Amer. Chem. SOC., 1959, 81, 3564.14 Fjerrum and Faurholt, 2. physik. Chem., 1927, 130, 584.15 Cernicki and Teiak, Croat. Chem. Acta., 1956, 28, 13.16 Stiasny and Balinyi, Collegium, 1927, 682,86.17 Ardon and Plane, J. Arner. Chern. SOC., 1959, 81, 3197.18 Finholt, US. Atomic Energy Comm. UCRL-8879, 1960, 66 pp.19 Souchay, Bull. SOC. chim., 1948, 15, 143.20 Faucherre, Bull. SOC. chim., 1954, 253. Schaal and Faucherre, Bull. SOC. chim., 1947, 927.21 Matijevi6, Broadhurst and Kerker, J. Physic. Chem., 1959, 63, 1552.Kolloid-Z., 1965, 205, 129.Chem., 1958,62,1271.1958, 6, 39.Biffl and Teiak, Croat. Chem. Acta.,Stiasny, Gerbereichemie (Th. Steinkopff Verlag,1958, 30, 9.Dresden, 1931), pp. 334 ff.Matijevik, Schulz and Tebk, Croat. Chem. Acta, 1956, 28, 81.Teiak, Matijevi6 and Schulz, J. Physic. Chem., 1955, 59, 769.22 Bjerrum, Z. physik. Chem., 1907, 59, 336.23 Lamb and Fonda, J. Amer. Chem. SOC., 1921,43, 1154.24Bjerrum, 2. physik. Chem., 1910, 73, 724. Oka, J. Chem. SOC. Japan, 1938, 59, 971.25 Larsen and Wang, J. Amer. Chem. Sac., 1954, 76, 6223.26Ryabchikov, Marov, Ermakov and Belyaeva, J. Inorg. Nucl. Chem., 1964, 26, 965.27 Johnson and &am, J. Amer. Chem. SOC., 1956,78, 3937. Johnson, Kraus and Holmberg, J.28 Larsen and Gammill, J. Amer. Chem. Soc., 1950, 72, 3615.29 Muha and Vaughan, J. Chem. Physics, 1960,33,194.30 Peshkova and Ang, Zh. Neorgan. Khim. (English trans.), 1962,7, 1901.Amer. Chem. SOC., 1956, 78, 26

ISSN:0366-9033    

DOI:10.1039/DF9664200187

出版商:RSC    

年代:1966

数据来源: RSC

摘要:

Electrokinetic Study of Dispersions of Clay in HydrolyzedAluminium SolutionsBY E. S. HALLWater Research Association, Medmenham, Marlow, Bucks.Received 14th June, 1966Values are presented of the <-potential of dispersed clay particles, in the presence of hydrolyzedaluminium salts, that were obtained for a study of the water-treatment Coagulation process. Thevalues of < are determined by the concentration of polynuclear hydroxy-ions that appear to have ageneral affinity for aqueous interfaces. Their concentrations depend upon the solubility of theprecipitated Al(OH)3, which at low concentrations is present only as a relatively insoluble layer onthe clay surface. As the iso-electric point falls from pH 8 to 5, the rate of coagulation of the dispersedparticles and the solubility of the hydroxide decrease and the strength of the aggregates formed bycoagulation increases.The clarification of water often involves the addition of aluminium salts, forthe purpose of co-precipitating aluminium hydroxide and dispersed particles andhigh-molecular weight organic substances.The solubility of aluminium hydroxideformed by the hydrolysis of Al3f ions is greatest at the instant of formation anddecreases subsequently by (i) rearrangement to more stable crystalline forms 1, 2 ;(ii) re-dissolving of the more soluble portions to form polynuclear hydroxy-ionsand basic salts or complexes with certain anions 3-5 ; (iii) interaction with manysurfaces, particularly those of clay, to form relatively insoluble adsorbed hydroxidelayers .6One model is based upon the assumption 7 that the initially precipitated hydroxideexists in a form in which there is the least amount of chemical bonding that willgive an infinitely extended structure.If the bonding occurs through Al-OH-A1links, similar to those in bayerite,* an arrangement is obtained having the form ofa chain of Al(OH)3 units in which each A1 atom shares two OH groups.According to this model, a reduction in the solubility of the precipitate withtime is brought about by a loss in the availability of aluminium, resulting fromcross-linking between chains, dissolution of portions that are not cross-linked ortheir adsorption on surfaces as relatively insoluble layers. These processes alsobring about a reduction in the volume of the precipitated hydroxide.In thissystem it is likely that much of the cross-linking is reversible and occurs throughhydrogen bonding. Cross-linking also occurs with the formation of 4-@-A1bonds2 and these result in a permanent reduction in solubility. In the presenceof such a system, dispersed particles are coagulated by cross-linking between adsorbedhydroxide layers.Packham 9 showed that the coagulation of kaolinite dispersions, having particleconcentrations not greater than 107 cm-3, was most rapid when aluminium hydroxideprecipitated most rapidly. This occurred at pH values at which only small con-centrations of polynuclear ions or complexes could have formed and when thevolumes of precipitated hydroxide were much greater than that of the dispersed19198 CLAY DISPERSIONSphase.Under these conditions amorphous aluminium hydroxide evidently pre-cipitated in its most soluble form. Coagulation was also observed when theamount of hydroxide was small in relation to the surface area of the clay, but onlyat higher particle concentrations. In this case the hydroxide was a comparativelyinsoluble layer on the surface of the clay.Under the first set of conditions, vigorous agitation caused the effective re-dispersal of clay particles, whereas in the latter case the aggregates were strongerand denser. It appeared that the quantity of hydroxide precipitated, in relationto the surface area of the clay, affected not only its solubility and rate of coagulation,but also the mechanical strength of the aggregates formed.In terms of our model,the mechanical strength of aggregates is determined by the concentration of A1atoms that are cross-linked. This is smaller in the more voluminous system.It has been concluded, therefore, that for water clarification the smallest additionof aluminium salts, necessary to bring about coagulation in a convenient time, leadsto a form of hydroxide that is strongest and least soluble for the particular system.The strength of the aluminium hydroxide determines the resistance of the aggregatesto breakdown by the shearing forces inherent in all solid-liquid separation processes.In the present paper it is suggested that the solubility, and hence the strength ofaluminium hydroxide, can be defined in terms of the electrokinetic potential, (.EXPERIMENTALThe electrokinetic potential 5 of adsorbed aluminium hydroxide on kaolinite and mont-morillonite dispersions was determined by microelectrophoresis at 25"C, according to themethod previously described.' The dispersion media were solutions of sodium chloride+bicarbonate+ carbonate of varying pH value, but having a reasonably constant ionic strength(111~ ~4.3 x 10-7 cm).The clays used were Speswhite, a sodium form of kaolinite having aparticle diameter mainly greater than 1 p and the sodium montmorillonite, Fulbent 570,which had a particle diameter approximately in the range 0-1-0.5 p. The electrophoreticmobility of these clays would not be greatly affected by the relaxation time of the electricaldouble-layer.10 The electrokinetic potential was therefore computed from the Smoluchowskiequation,= 4 n t p / X ~ .Solutions of aluminium sulphate, nitrate and an aged, partially hydrolyzed form of the latterwere added to 0.050 g/l.dispersions of clay in the appropriate dispersion media and allowedto stand at 25" for 15 min before measurements were made. Partial hydrolysis of aluminiumnitrate was affected by adding 2-5 M quantities of NaOH to one of Al3f in order to producepolynuclear ions. Because these are slow to form, the immediate reaction was the precipita-tion of aluminium hydroxide. Most of this had subsequently re-dissolved after the suspen-sion had been stored for 16 h at 37°C. A potentiometric titration of the resultant solutionindicated that at least half of the A1 had been converted to polynuclear species.Most of theremainder appeared to be present as a relatively insoluble form of hydroxide.RESULTSThe results obtained from the addition of various amounts of aluminium sulphateto kaolinite and montmorillonite dispersions are given in fig. 1. The descendingportion of each curve represents the range of precipitation of aluminium hydroxide,large concentrations of which were clearly visible under a microscope. As theconcentration of hydroxide was reduced in relation to the surface area of the dis-persed clay, the iso-electric point of the descending portion diminished progressivelyfrom about pH 8 to 5. The condition where the concentration of hydroxide wassmall, in relation to the surface area of the dispersed phase, was demonstrated mainlE.S. HALL 1990-0400 * 0 3 00'ij 0-020- 5u0-010by montmorillonite. The surface area of the kaolinite was not sufficiently large toachieve wide variations in the solubility of the adsorbed hydroxide layer, exceptwhen it was placed in competition with a 0.030 g/l. dispersion of polysilicic acidI I I I I- --- -I I I I I0*03(002(0'013+v(34- 0.01'- 0.02 1- 0.03'- 0.04FIG. 1 .-The electrokinetic potential of aqueous dispersions of clays in the presence of hydrolyzedaluminium sulphate. 0.05 g/l. kaolinite and the following Al concentrations : A, 6.0 x 10-4 M ;B, 1.5 x 10-4 M ; C, 6-0 x 10-5 M ; D, 1.5 x 10-5 M ; E, 7.5 x 10-5 Mfpolysilicjc acid (0.03 g/l.SiO~)0.05 g/l.montmorillonite and the following Al concentrations : F, 1.5 x 10-4 M ; G, 9.0 x 10-5 M ;H, 6.0 x 10-5 M ; J, 1.5 x 10-5 M ; K, no Al.FIG. 2.-Th200 CLAY DISPERSIONS(curve E). At negative values off: the slope of the curves were similar for all con-centrations of hydroxide. When the latter was reduced below that giving an iso-electric point at pH 5, the curves progressively flattened out to that for the purePHFIG. 3.-The electrokinetic potential of aqueous dispersions of kaolinite (0-05 g/l.) in the presence ofaged, partially hydrolyzed aluminium nitrate. A, 7.5 x 10-4 M ; B, 1.5 x 10-4 M ; C, 3.0 x 10-5 M ;D, 1-5 x 10-6 M Al.3 - 0WuOgO50- I 1 I I0.040 -0.030-0.020 -0'010-- 9 - 8 -7 -6 - 5 - 4 -3log10 CAI1FIG.4.-The inaxirnum electrokinetic potentials of aqueous dispersions of kaolinite (0-05 g/l.) givenby aged, partially hydrolyzed aluminium nitrate.clay. Fig. 2 shows that larger positive values of 5 were obtained with aluminiumnitrate. Ageing of aluminium nitrate solution in a partially hydrolyzed state pro-duced higher values and these were reasonably constant over certain pH rangeE . S. HALL 201(fig. 3). These values are plotted in fig. 4 against the total concentration of Aladded to each dispersion.DISCUSSIONThere is a clear correlation between the iso-electric points for the descendingportion of the 5 against pH curves and the form of the precipitated aluminiumhydroxide. As the volume of hydroxide became large compared with that of thedispersed phase, the iso-electric points tended towards about pH 8.When thehydroxide concentration was low, however, a limiting iso-electric point at aboutpH 5 was observed.given by aluminium sulphate,nitrate and the aged, partially hydrolyzed solution of aluminium nitrate are evidencethat the value of f is determined mainly by the concentration of polynuclear ionsof aluminium. The highest values of 5 were obtained from a solution in whichpolynuclear ions had been given the maximum opportunity to develop (fig. 3).The values given by unaged aluminium nitrate (fig. 2), were less because of theslow rate of formation of polynuclear ions.3 In the presence of sulphate, whichforms basic salts or complexes with partially hydrolyzed forms of the aluminiumi0n,43 5 the values of f were still further reduced (fig.1).The work of MatijeviC et aZ.11 with silver halides has shown that the peptizinginfluence of partially hydrolyzed aluminium solutions is not confined to claysurfaces. The emulsification of oils has also been reported 12 and we have ob-served that foaming occurs more readily in such solutions. Evidently polynuclearions of aluminium have a general affinity for aqueous interfaces without possessinga hydrophobic function. The structures of the polynuclear ions of aluminiumhave not been elucidated, but all of the suggested formulae 39 1 1 3 13 can be basedmost reasonably upon the Al6 hexagon found in bayerite. Probably, such a unitcould associate with an ordered array of water molecules that exist at a water sur-face14 in an analogous manner to the association between the aluminium andsilicate layers in clays.8The descendiog portion of the curves in fig.3 refers to a state of equilibriumbetween polynuclear ions and their hydrolysis product. The corresponding portionsof the curves in fig. 1 and 2 may refer to equilibrium conditions, but hydroxidesof continuously varying stability are involved for each curve. This occurs becauseprogressively greater proportions of the hydroxide are consumed to form poly-nuclear species as the pH is reduced. However, at higher pH values the formationof a small concentration of polynuclear ions is achieved reasonably quickly andwould not significantly affect the structure of aluminium hydroxide newly adsorbedon a given area of clay surface.This would be mainly determined by the quantityof hydroxide precipitated. Therefore, over a sufficiently high pH range, the dis-persion medium is in quasi-equilibrium with a given form of the hydroxide. Theequilibrium would not be truly reversible because subsequent changes in the com-position of the dispersion medium would result in permanent changes in the formof the hydroxide. The similarity between the gradients in fig. 1 and 3 suggeststhat quasi-equilibria with constant forms of hydroxide occur for most negativevalues of 5.The comparative magnitude of the values ofThe quasi-equilibria conform approximately to the equation= constant - 0.028 pH, (1)for a given form of adsorbed layer.Expressing the equilibrium between polynuclea202 CLAY DISPERSIONSions and the adsorbed hydroxide layer by the stability constants,andwhere &.ads. is the solubility product of the adsorbed hydroxide, it follows that[ conforms to a Nernst-type equation,[Al,,(0H)~/s”n’+]/[Al3+ Jn[0H-]”.’” = K , (2)+1coH-13 = K a d s , (3)0.056nC = constant + - log,, [AI,(OH)Fi?+]. (4)In fig. 4 the slope of the line corresponds to a value of n of about 7-2, compared withthe values of 6, 8 and 13 that have been suggested previously.3s 1 1 s 13Whereas the Nernst equation, expressed in terms of either Al3f or OH-, wouldpossibly describe the inner potential 15 of this system, one might ask why a similarrelation should hold good for the electrokinetic potential.A feature of the rela-tionship is that 0*028/ln 10 V is such a high proportion of RT/F. At 25”C, 0.028F/RT In 10 equals 0-47, which is similar to estimated values of [/$*,I6 where t,ho isthe outer potential. If indeed $0 follows the Nernst equation, a constant arrange-ment of water molecules at the interface is indicated.If the constant in eqn. (4) is dependent only upon the affinity between the poly-nuclear ions and the water structure, eqn. (1) can be put in the general form,C = constant + 0.056 log,, -0.028 pH. ( 5 )Eqn. (5) indicates a shift in the iso-electric point of the adsorbed hydroxidelayer from 8.0 to 5.0 as corresponding to a reduction in loglo Ks.ads. by 1.5. Takingthe value of Ks.ad& in the former case to be 10-32-4917 it follows that at low concentra-tions of precipitated hydroxide, the solubility product of the adsorbed layer attainsa limit of 10-33.9, which corresponds nearly to the stability of the crystalline hydratedoxide, bohmite.179 18 A similar conclusion was reached by Ragland and Coleman 6who observed that the hydrolysis of 10-3 M aluminium chloride in the presence oflarge concentrations of clay could occur at pH values as low as 3.4.Bohmite hasa structure in which each A1 atom is shared by four oxygen atoms and two hydroxylgroups.8 An adsorbed layer of hydroxide having similar stability would not there-fore possess any reactive chains such as were postulated for the most soluble form.Matijevie et aZ.19 have observed that another hydrolysis product of Al3+, formedby the heating of aluminium nitrate at 90’61, is a powerful coagulant for silver iodideat pH values less than 4, where the unaged salt has no effect. Probably, the productof this ageing also has a chain structure, for we have found that evaporation ofsuch a solution brings about the precipitation of thin wafers having the structureshown by fig.5. This material gives the X-ray diffraction powder pattern forbohmite. From fig. 1 it is reasonable to expect molecular chains, having a solu-bility similar to that of bohmite, to exist at pH values less than 4 as a positivelycharged dispersion. These chains are strong, but since they still possess hydroxylgroups, they probably can bring about coagulation in a manner similar to that en-visaged for amorphous aluminium hydroxide.The existence of such units in aprecipitate formed at room temperature is unlikely, but the possibility that they maycontribute to the coagulation process under the more acidic conditions cannot beentirely discounted.The author thanks Dr. G. C . Bye of Sheffield University for obtaining electronmicrographs and X-ray diffraction patterns of samples and the Director of theWater Research Association for permission to publish this workFIG. 5.-An electron niicrograph of bohmite formed by refluxing 1 . 8 0 ~ 10-3 M aluminium nitrate for24 h, followed by evaporation in the pH range 3.2-3-6 (48,000 x magnification).[To face page 202E. S. HALL 2031 Fricke and Mehring, 2. anorg. Chern., 1933,214,269.2 Bye and Robinson, Kolloid Z., 1964,198, 53.3 Brossett, et al., Acta. chem. scand., 1954, 8, 1917.4Denk and Bauer, 2. anorg. Chem., 1952,267, 89.5 Marion and Thomas, J. Colloid Sci., 1946, 1, 221.6 Ragland and Coleman, Soil Sci. Soc. Proc., 1960, 457.7 Hall, J. Appl. Chem., 1965, 15, 197.8 Wells, Structural Inorganic Chemistry (Oxford University Press, 2nd ed., 1950).9 Packham, J. Colloid Sci., 1965, 20, 81.10 Overbeek, Kolloid chem. Beih., 1943, 54,289.11 Matijevii, et al., J. Physic. Chem., 1961, 65, 826.12 Istomina, Sbornik Nauch, Rabot Moladykh Uchenykh (Tomsk, 1960), 102.13 Biedermann, Svensk Kemisk Tidskr., 1964,76, 19.14 Drost-Hansen, Ind. Eizg. Chem., 1965, 57, 18.15 Overbeek, Colloid Science (ed. Kmyt) (Elsevier, Amsterdam, 1952), 1, 124.16 Davies and Rideal, Interfacial Phenomena (Academic Press, London, 1961), p. 145.17 Latimer, Oxidation Potentials (Prentice-Hall, Englewood Cliffs, N.J.), 2nd. ed. (1952).18 Pourbaix, Atlas of Electrochemical Equilibria (Pergamon Press, Oxford, 1966), p. 169.19 Matijevik et al., J. Colloid Sci., 1964, 19, 333

ISSN:0366-9033    

DOI:10.1039/DF9664200197

出版商:RSC    

年代:1966

数据来源: RSC