Effect of Molecular Architecture of Long Chain Fatty Acids on the Dispersion Properties of Titanium Dioxide in Non-aqueous Liquids BY ANDREW DOROSZKOWSKI AND RONALD LAMBOURNE ICI Paints Division, Wexham Road, Slough, Berks Received 23rd November, 1977 The effect of varying chain length and chain branching on the adsorbed layer thickness of oligo- esters on titania in non-aqueous solvents has been studied and related to the degree of flocculation of the dispersions using viscometry. The oligoesters were a series of monodisperse condensates of 1Zhydroxystearic acid, up to the pentamer. Also, a series of branched esters in which the degree of branching was varied systematically was studied. The latter consisted of the valeric esters of mono-, di- and tri-hydroxystearic acid. It was found that the thickness of the adsorbed layer was not necessarily the criterion for good dispersion stability, but a complex function of surface concentration.This was dependent on main chain length, the size, position and number of branches and the solvency of the medium. The stabilisation of colloidal particles in non-aqueous media has received consider- able attention in recent years. It appears now to be generally accepted that stabilisa- tion in media of low dielectric constant is due to steric considerations involving loss of entropy and/or enthalpy and not to charge repu1sion1p2 except possibly for extremely dilute dispersions. There have been many theoretical publications on steric stabilisa- tion by adsorbed polymers, including statistical approaches, e.g., " loopy " adsorption and involving Flory-type thermodynamic considerations. These all depend on (segment density distribution, distance) calculations. These segment density distri- butions are obtained with difficulty theoretically and are inaccessible to physical measurement, at least for the present. Steric stabilisation, on the other hand, may be more simply studied by the use of simple fatty acids which offer the practical solution of known geometry and are, therefore, amenable to configurational considerations ; but they cannot be considered as polymers in lattice-type calculations.Rehbinder first studied the dispersion properties conferred by the adsorption of fatty acids in non-aqueous media. Ottewill and Tiffany4 also studied the adsorption properties of naturally occurring fatty acids.In this work we have made use of the ester linkage afforded by esterification of hydroxy fatty acids to construct in a controlled manner different species with various degrees of branching at constant chain length (extensibility), and attempted to relate structure to dispersion stability. Additionally we have kept the degree of branching constant and altered the extensibility (chain length). The latter has enabled the further appraisal of viscometric methods for determining the thickness of adsorbed layers on small particles.A. DOROSZKOWSKI AND R . LAMBOURNE 253 EXPERIMENTAL MATERIALS Stearic acid, '' Specially Pure ", (ex B.D.H.) m.p. 70 - 72 "C was used as received. 2- Hydroxypalmitic acid (ex Koch-Light) m.p.84-86 "C was used as received. PREPARATION OF HYDROXY ACIDS 1 2-Hydroxystearic acid was purified from commercial grade material (ex Prices, Bromborough) containing about 8.7% stearic and 0.9% palmitic acid. The methyl ester was prepared and recrystallised three times from petroleum ether (B.P. 100 - 120 "C). It was saponified, acidified and washed with distilled water, collected and stored in a vacuum dessicator over silica gel till required. The product (m.p. 82-83 "C) was analysed for OH value and purity was assessed by t.l.c., which showed the elimination of impurities. 9,lO-Dihydroxystearic acid was prepared from oleic acid (ex Hopkin and Williams) by the method described by Swern et al. ; the product had a m.p. 92 - 93 "C. 9,10,12-Trihydroxystearic acid was prepared from ricinoleic acid (reagent grade) by first acetylating the hydroxyl group.The double bond was then hydroxylated as described in the preparation of dihydroxystearic acid. The product was a white waxy solid, m.p. 105- 107 "C. ACID CHLORIDES Valeryl chloride was prepared from valeric acid (Puriss grade ex Koch-Light) using thionyl chloride as described by V0ge1.~ Stearoyl chloride was prepared from stearic acid using phosphorus trichloride as described by Young et alas Acid chlorides of 12-octadecanoyloxyoctadecanoic acid (dimer), trimer etc. were prepared from the corresponding dimer fatty acid using the same method as for stearoyl chloride. An alternative, simpler method using oxalyl chloride was also employed, which involved refluxing with the appropriate fatty acid and distilling off excess oxalyl chloride, leaving the acid chloride.(cf. Thionylchloride which produced black tars.) OLIGOESTERS The oligoesters were all prepared by condensing the appropriate acid chloride with the hydroxy fatty acid by heating the components at 120-130 "C and removing hydrogen chloride under reduced pressure. The oligoesters made with valeryl chloride were prepared by heating the appropriate reagents in the presence of lutidine. The reaction product was washed with dilute hydrochloric acid, water and then dried. Finally it was passed through a silica gel column to remove impurities. Purity was assessed by t.1.c. in the usual manner. Number average (M,) molecular weights were determined using a Mechrolab Vapour-phase Osmometer.Acid contents were determined by titration using alcoholic potassium hydroxide solution. The 16-mer oligoester was obtained by fractional precipitation of a 13% w/w solution of " poly (12 hydroxystearic acid) ", prepared as described by Walbridge9 in butyl cellosolve, using distilled water as precipitant. The turbid solution was warmed to clarity and allowed to reprecipitate slowly in a thermostatted water bath. The fraction collected accounted for about 9% of the polyhydroxystearic acid and had an M, of 4500, corresponding to the 16- mer hydroxystearic acid. SOLVENTS All solvents used were AnalaR grade except for the aliphatic hydrocarbon (white spirit 100) which was a petroleum fraction of low aromatic content having a boiling range 179- 200 "C.254 OLIGOESTER CHAIN EFFECTS ON TiO, DISPERSIONS PIGMENTS All dispersions were made from commercial grade titania which had been surface treated with SO2, A1203 etc., and was similar to that used by Crow1 and Malati;l* the B.E.T.nitrogen adsorption surface area was 13.4 m2 g-l. The surface area computed from electron micrographs at x 20 000 magnification was found to be 11 m2 g-l. METHODS Adsorption experiments were made with titania dried at 100 "C. The titania with the fatty acid solution was placed in glass jars together with I?' glass beads. The jars were continuously rotated on rollers for at least 24 h. The continuous phase was obtained by centrifugation of the dispersion and analysed for fatty acid content by simple titration with potassium hydroxide solution.Preparation of Dispersion for viscosity measurements. Dispersions were prepared by milling 200 g of titanium dioxide with 250 cm3 of dispersion solution in a + gallon stone jar with 500 g of r' steatite balls for 24 h. The suspensions so prepared were strained off and 5 cm3 pipetted samples were ashed in a muflle furnace at 500 "C to obtain accurate disperse phase volumes. The dispersions were diluted with the appropriate solution where required to obtain lower concentrations. VISCOMETRY The viscosities of the dispersions were measured on a Weissenberg rheogoniometer, fitted with a 5 cm diameter, 3" cone and plate; housed in a 25 "C constant temperature room. The thickness of the oligoester fatty acid layer adsorbed on the titanium dioxide was deter- mined by measuring viscosities at a number of shear rates ranging from 380 to 4776 s-I and extrapolating to infinite shear rate in order to eliminate the effects of flocculation.The application of an empirical equation relating dispersion viscosity at infinite shear rate with continuous phase viscosity and disperse phase volume (DPV) enabled the calculation of layer thickness from the increase in hydrodynamic volume due to the adsorbed layer.5 The results of these measurements are presented in table 2. The dispersion stability of the oligoester fatty acids was assessed in the manner of Asbeck and Van Loo,ll and as used by us previo~sly,~ by plotting log (viscosity) against the reciprocal square root of shear rate as in fig. 1. The gradients of these graphs were used as a measure of flocculation and termed the " flocculation factors ", (see table 6).The smallest gradient indicates the least flocculation of the dispersion. The assessment of flocculation, as measured by the divergence from Newtonian behaviour, is disperse phase concentration dependent (fig, 2) and was, therefore, only made at similar disperse phase concentrations. RESULTS Fig. 1 shows that Ti02 particles, dispersed in a solution of the monovalerate of 12-hydroxystearic acid in aliphatic hydrocarbon are less flocculated than in the equivalent trivalerate solution, which in turn produces a better dispersion than in the equivalent divalerate solution. The most flocculated dispersion being that made using stearic acid as dispersant. The effect on dispersion stability and '' barrier thickness " with increasing chain length are shown in fig.2. Fig. 3 shows the effect of solvency on dispersions stabilised with " dimer " hydroxystearic acid. The adsorption isotherms for the various oligoester fatty acids are presented in fig. 5 and 6 . The Catalin version of H. A. Stuart's molecular models was used to construct models of the oligoester fatty acids from which their projected areas were measured in various orientations. In the case of the dimer, trimer, etc., the projected area in the normal orientation is taken to be that of the shorter chained monovalerate, divalerate analogue. The minimum areas quoted in table 1 were obtained by rotatingA . bOROSZKOWSK1 AND R. LAMBOURNE 255 I 1 .I-I 0.01 0.03 0.0 5 0-45 3 F~G. 1 .-Titania dispersed at high DPV (16.5%) and constant layer thickness in aliphatic hydrocarbon. Titania dispersed with: XII, stearic acid; XI, valerate of 2-hydroxypalmitic acid; 111, divalerate of 9,lO-dihydroxystearic acid; IV, trivalerate 9,10,12-trihydroxystearic acid; 11, valerate of 12-hydroxy- stearic acid.FIG. 2.-Log q against 0-3 plots of titania dispersed with varying chain length oligoesters in aliphatic hydrocarbon (note effect of DFV on flocculation in 2 and 5 dispersions). Titania dispersed with: 1, 16-mer hydroxystearic acid; l’, continuous phase of 1; 2, trimer at high DPV; 3, dimer at low DPV; 4, pentamer at low DPV; 5, trimer at low DPV. (Corresponding numbers “ prime ” refer to continuous phases.)256 OLIGOESTER CHAIN EFFECTS ON TiO, DISPERSIONS 0.01 0.03 0.05 L d l s 5 FIG.3.-Log q against 0-3 plots of titania dispsersed with dimer hydroxystearic acid in different solvents but at the same DPV. Titania dispersed with dimer hydroxy stearic acid in: (A), aliphatic hydrocarbon; (I?), n-butyl acetate; (C), xylol; and (a), (b), and (c) in corresponding continuous phases. 6 r 5 C 4 3 2 L--I_____L-, 0.01 0.03 0.05 &, + FIG. 4.-Log t j against D-4. plots of titania dispersed with dimer, acetate of 12-hydroxystearic acid and oleic acid in aliphatic hydrocarbon at similar DPVs. Titania dispersed in aliphatic hydrocarbon using: (A), dimer 12-hydroxystearic acid (16.5% DPV); (B), acetate of 12-hydroxystearic acid (15% DPV); (C), oleic acid (15% DPV).A . DOROSZKOWSKI A N D R . LAMBOURNE 257 TABLE 1 .<OMPARISON OF MOLECULAR AREAS OBTAINED FROM MODELS AND EXPERIMENTALLY DETERMINED MOLECULAR AREAS USING ADSORPTION ISOTHERMS.molecular areas from molecular area from models/A2 adsorption isotherm/A2 minimum inter- compound cross- mediate aliphatic reference section configur- molecule hydro- butyl no. adsorbate area ation" flat carbon xylol acetate I 9-octadecanoic acid (cis form) (oleic acid) octadecanoic acid (monoval- erate of hydroxystearic acid) I11 9: 10-dipentano- yloxyocta- decanoic acid (divalerate of dihydroxy- stearic acid) IV 9: 10: 12-tri- pentanoyloxy- octadecanoic acid (trivalerate of trihydroxy- stearic acid) V 12-octadecanoyl- oxyoctadecanoic acid (" dimer " hydroxystearic acid) VI 12-octadecanoyl- oxyoctadecano- yloxy octa- decanoic acid (" trimer " hydroxystearic acid) I1 12-pentanoyloxy- - - 29 50 90 45 - 36 50 130 68 160 190 40 I 180 120 38 50 - 250 56 64 70 123 40 120 300 70 109 - a Assuming maximum number of carbonyl groups adsorbed with minimum -CH2- adsorption.Oleic acid included for comparison, assuming that the carboxyl and double bond only adsorbed.258 OLIGOESTER CHAIN EFFECTS ON TiOz DISPERSIONS $ 4 CSI 0 9 3 0, a U 0 a CI 0 5 2 1 1 2 3 4 concentration g per 100 cm3 FIG, 5.-Typical adsorption isotherms. 0, trivalerate; x , divalerate; 0 , valerate; all of hydroxy stearic acid. the individual atoms to give the densest packing allowed by the models and assuming attachment to the substrate through the carboxyl group.12*13 The results are sum- marised in table 1 along with the experimentally determined values of area per molecule obtained from the plateau region of the adsorption isotherms.I m X x n 0 1 2 3 4 concentration g per 1o0cm3 FIG. 6.-Typical adsorption isotherms. 0, Dimer 12-hydroxystearic acid in aliphatic hydrocarbon; x , Dimer in xylol; 0, Dimer in n-butyl acetate. DISCUSSION The oligoester fatty acids may be divided into three groups for ease of reference: group 1 : consisting of the acetate, valerate and stearate esters of 1Zhydroxystearic acid; where the size of the branch chain is varied, but not its position of attachment to the main chain. group 2 : consisting of mono-, di- and tri-valerate of 1Zhydroxystearic acid, 9,lO- dihydroxy- and 9,10,12-trihydroxystearic acids, respectively, where there is an increase in the number of branch chains, and the valerate of 2-hydroxy-A .DOROYZKOWSKI A N D R. LAMBOURNE 259 palmitic acid. All branch chains are of equal length but their positions of attachment differ. group 3: dimer, trimer etc. of 12-hydroxystearic acid where the chain length is uniformly increased, as is the branching. TABLE 2.-ADSORBED LAYER THICKNESS MEASUREMENTS ON TITANIA USING DIFFERENT OLIGO- ESTERS COMPARED WITH CHAIN LENGTH MEASUREMENTS ON THE MOLECULAR MODELS. chain com- length adsorbed pound measured apparent layer ref. from DPV actual thickness no. adsorbate model/A qmlqo DPV IA I I1 111 IV V VI VII VIII IX oleic acid monovalerate of 12- h ydr ox ys t ear ic acid divalerate of dihydroxy- stearic acid trivalerate of tri- hydroxystearic acid " dimer " hydroxy- stearic acid " trimer " hydroxy- stearic acid " tetramer " hydroxy- stearic acid " pentamer " hydroxy- stearic acid " 16-mer " hydroxy- stearic acid 1.94 24 2.21 24 2.21 24 2.21 39 2.21 54 1.42 70 85 1.93 256 (64)b 2.7 16.5-15.8 18.3-17.4 18.3-1 7.4 18.3-17.4 11.65-11.07 9.76-9.28 14.98-1 5.75 20.3-21.35 15 16.5 16.5 16.5 9.0 7.2 10.68 15.60 10-20 10 -20 10-20 10-20 40-50 50-60 70-90 50-70 The two values quoted depend on which correction term is used in determining DPV [see ref.(S)]. Calculated value of r.m.s. end-to-end length in aliphatic hydrocarbon. By comparing the experimentally derived areas occupied per molecule, obtained from adsorption isotherms, with the projected areas obtained from the molecular models (see table 1) it is concluded that all the oligoester fatty acids studied, with the possible exception of oleic acid, are adsorbed with the carboxyl group down and the major axis of the molecule perpendicular to the adsorbing surface.This is in keeping with adsorption studies of stearic and other fatty acids by Shenvood and Rybicka and others,I2J3 who concluded that fatty acids were attached to the surface by both ionic and hydrogen bonding. This orientation is also borne out by viscometric studies on the adsorbed layer thickness. If the London-van der Waals forces of attraction, which are responsible for causing the disperse phase to flocculate, are considered, then the energy of attraction between the titania particles is dependent on the effective Hamaker constant (Al2). This is greatest in aliphatic hydrocarbon and least in butyl acetate.However, contrary to expectation, a greater degree of flocculation of the TiOz has been observed in xylol or butyl acetate compared with aliphatic hydrocarbon. This we attribute to the lower surface coverage (or reduced segmental concentration) of the oligoester fatty acids in the better solvents. Since the adsorption (or partitioning) is probably not wholly ionic, surface cover-260 OLIGOESTER CHAIN EFFECTS ON TiOz DISPERSIONS TABLE 3 .-LIST OF HAMAKER CONSTANTS. attractive potential attractive potential (Al2) x 10l3/erg Y = cm r = 0.5 x cm Hamaker const ant (V) in kT ( V ) in kT mat er id ( A 1) (of rutile in) at separation of at separation of rutile 18" - lOA 20A 3 0 A lOA 20A 30A aliphatic H/C 4.4.F 4.6 93 46 32 46 21 16 xylol 5.2t 3.8 78 39 27 39 20 13 n-butyl acetate 35.3p 2.9 58 29 20 29 15 10 * Average quoted by Visser [ref.(14)] V = Ar/l2hO. t Calculated from refractive index using Gregory's approach [ref. (1 S)]. age might be expected to be influenced by the solubility parameter of the solvent. The solubility parameters and the corresponding solvent " fractional polarity " (re- lated to hydrogen bonding potential of the solvent) are quoted in table 4. The differences in the cohesive energy density (CED) are not, however, large and it is doubtful if differences in solvency (as reflected in CED element) are responsible for the variation in surface coverage. However, the fractional polarity may play an important part in determining whether or not the acidic species are more or less readily displaced by solvent, leading to a significant reduction in surface coverage in the case of butyl acetate solutions.TABLE 4.-LIST OF SOLUBILITY PARAMETERS. solubility parameter [ref. (16)] fractional polarity CED element xylol (mixed isomers of xylene) 8.8 0.001 butyl acetate 8.6 0.167 tetradecane 8 .o 0 fatty acids 9* 0.2-0.3 f * Calculated using Rheineck's approach [ref, (17)]. Estimate. EFFECT OF CHAIN BRANCHING AND SEGMENT DENSITY I N THE ADSORBED LAYER ON DISPERSION STABILITY The size of the branch chain appears to be very important with respect to dispersion stability for, although the acetate and valerates of 1Zhydroxystearic acid have the same " site " density (38 A2 molecule-l) on the substrate, by increasing the acetate side chain length by just three -CHz- links a very large increase in stability was obtained, (fig.4 and table 6). Increasing the side chain to 18 carbon atoms on the other hand, decreased the site occupancy, since the measured area per molecule is greater, but changes in the surface layer thickness have also occurred, compensating for the de- crease in surface density. This indicates that surface concentration is very important in influencing dispersion stability. The acetate of 12-hydroxystearic acid was marginally superior to oleic acid as a dispersant, and both were very much better than stearic acid. (Note flocculation factors, table 5.) All three fatty acids are perpendicularly oriented to the particlePLATE 1 .-Viewing left to right : dimer, divalerate, trivalereate, valerate and acetate of hydroxystearic acid, valerate of 2-hydroxystearic acid.PLATE 2.-Above: trimer, below: pentamer. [To face page 260A . DOROSZKOWSKI A N D R . LAMBOURNE 261 TABLE 5.-cOMPARSION OF " LOCALISED " SEGMENTAL VOLUME WITH FLOCCULATION FACTOR IN ALIPHATIC HYDROCARBON. ____ ~ segments 8 A penetra- com- relevant per floccu- tion pound volume no. of unit lation segments per mol. no. adsorbate /A3 segments volume factor unit volume wt./vol. I oleic acid 45 x 8 9 0.025 6.9 0.022* 0.32 I1 valerate of HSA 38 x 8 63-5 0.036 1.2 0.036 0.51 111 divalerate of DHSA 56 x 12 9+5+5 0.028 3.7 0.025 0.44 IV trivalerate of THSA 70 x 12 9+5+ 0.029 2.4 0.029 0.43 X acetateof HSA 38 x 8 6+2 0.026 6.1 0.026 0.46 XI valerate of palmitic acid 38 x 20 14+5 0.025 7.0 0.021 0.47 5 f 5 * Assuming vertical orientation.7.0 1.2 n u) al .Q a 0 0 d c 0 U ZJ --. * U 6.t c c .- c d g 3.7 L YI 2.4 FIG. 7.-Schematic representation of adsorbed layer showing the position of regions of increased segment density relative to the adsorbing surface with various oligoester acids. (a) Valerate of 2- hydroxypalmitic acid, flocculation factor 7.0; (b) valerate of 12-hydroxystearic acid, flocculation factor 1.2; (c) acetate of 1Zhydroxystearic acid, flocculation factor 6.1; (d) divalerate of 9,lO- hydroxystearic acid, flocculation factor 3.7 ; (e) trivalerate of 9,1O,lZhydroxystearic acid, flocculation factor 2.4.262 OLIGOESTER CHAIN EFFECTS ON TiO, DISPERSIONS surface and, since the stearic acid stabilised dispersion is the most flocculated, then interpenetration of the straight CIS chain must be more probable than with branched chains.The effect of chain branching is perhaps best illustrated by reference to fig. 7, (which deals with the case of groups 1 and 2 acids). Branching gives rise to increases in segment density in regions varying in remoteness from the main chain ends. Thus, flocculation is decreased when the region of increased segment density is closest to the outermost part of the adsorbed layer [cf. fig. 7(a) and 7(b)]. The acetate of 12-hydroxystearic [fig. 7(c)] contributes only a small region of increased segment density, in comparison to the valerate, [cf. fig. 7(b)], and the efficiency is only marginally different to oleic acid or the valerate of 2-hydroxy- palmitic acid, fig. 7(a). The di- and tri-valerates of the appropriate hydroxystearic acids [fig.7(d) and 7(e), respectively] exhibit intermediate levels of efficiency of dis- persion stability due to the positions of the branches. Using the same models, the increase in segment density arising from interpenetration can be calculated. Thus, in table 5, a correlation is shown to exist between the " flocculation factor " (an experi- mental measure of the dispersion stability) and the effective segment density allowing for interpenetration of the adsorbed layer up to 8 A. In the treatment localised concentration effects are thus taken into account. If the total (or average) surface concentration of adsorbed molecules is considered, no correlation appears to exist. The importance of local configurational effects has perhaps been overlooked in the more sophisticated statistical treatment of polymer adsorption, which may explain why the statistical treatments have so far not agreed with experimental data.TABLE 6.-LIST OF FLOCCULATION FACTORS (SLOPE OF LOG 77 AGAINST D' PLOT) SHOWING EFFECTS OF DPV, SOLVENT ENVIRONMENT, CHAIN LENGTH AND CHAIN BRANCHING. chain length measured chain from pound (calc. in at high DPV at low DPV com- length viscometry flocculation factor? ref. from aliphatic aliphatic aliphatic no. fatty oligoester acid models) H/C H/C xylol H/C xylol I XI1 X I1 I11 IV XI V VI VII VIII IX oleic acid stearic acid acetate of HSA valerate of HSA divalerate of DHSA trivalerate of THSA valerate of 2-HPA dimer of HSA trimer of HSA tetramer of HSA pentamer of HSA 16-mer of HSA 20 10-20 6.9 24 20 22.2 24 20 6.1 24 20 1.2 24 20 3.7 24 20 2.4 24 20 7.0 39 45 4.5 54 55 2.5 1.8 70 85 80 256(64)* 60 0.6 - - - 4.4 - - - - 6.1 1.7 2.2 7.7 0.4 0.7 2.4 - 0.2 - - - - - - * r.m.s.in aliphatic H/C. 7 slope of log q against 0-3 plot.A . DOROSZKOWSKl AND R . LAMBOURNE 263 EFFECT OF CHAIN LENGTH O N ADSORBED LAYER (GROUP 3 OLIGOESTERS) The agreement between the experimentally determined adsorbed layer thickness and the corresponding theoretical length of the terminally adsorbed oligoester fatty acids is very close, except in the instance of the 16-mer hydroxystearic acid. In this case it is suggested that the 16-mer, because of its size is behaving more like a true polymer, by adsorbing in a typically coiled manner. The adsorbed layer thickness is, therefore, very much smaller than the linear length of the polyester, and might even be the same as its r.m.s.length in free solution, 64 A (see table 2). Thus the viscometric technique to determine the adsorbed layer thickness on small particles as described by the authors5 is validated, despite the approximations made, and gives values of adsorbed layer thickness consistent with molecular dimensions. Although the degree of flocculation of the dispersion decreases regularly with increase per unit length of oligoester in the group 3 series, there is also a concurrent change in chain spacing which effects the surface concentration. Hence it is difficult to isolate the effect of chain length of stabilising molecules on dispersion stability apart from an overall qualitative appreciation that there is an improvement. It is, however, clear from the results that dispersion stability is not simply a function of the adsorbed layer thickness of the stabilising species but a complex function of surface concentration. The latter is dependent on the main chain length, the size, position and number of branches and the solvency of the medium. D. W. J. Osmond, Disc. Faraday Soc., 1966, 46, 314. G. R. Feat and S. Levine, J. Colloid Interface Sci., 1976, 54, 34. P. Rehbinder, 2. Phys. Chem., 1930, A146,63. R. H. Ottewilf and J. M. Tiffany, J. Oil Coloiir Chem. ASSOC., 1967, 50, 877. A. Doroszkowski and R. Lambourne, J. Colloid Interface Sci., 1968, 26, 214. D. Swern, G. N. Billen, T. W. Findley and J. T. Scanlan, J. Amer. Chem. Soc., 1945,67,1787. A. Vogel, Practical Organic Chemistry (Longmans, 1961), p. 367. a C. G. Youngs, FI. Epp, B. M. Craig and H. R. Sallans, J . Amer Oil Chem. Soc., 1957,39,107. D. J. Walbridge, Dispersion Polymerisation in Organic Media, ed. K. E. J. Barrett (Wiley and Sons, N.Y., 1975), p. 108. V. T. Crow1 and M. A. Malati, Disc. Faraday Soc., 1966, 42, 301. l1 W. K. Asbeck and M. Van Loo, fnd. and Eng. Chem., 1954,76,1291. l2 A. F. Sherwood and S. M. Rybicka, J . Oil Colour Chem. Assoc., 1966, 79, 648. l3 J. J. Kipling and E. H. Wright, J. Chem. SOC., 1967, 3535. l4 V. Visser, Ado. Colloid Interface Sci., 1972, 3, 331. l5 J. Gregory, Adv. Colloid Interface Sci.. 1969 2, 296. l6 J. L. Gardon, J. Paint Tech., 1966,38,43. l7 A. E. Rheineck and K. F. Lin, J. Paint Tech., 1968,40, 61 1 .