J . Chem. Soc., Furuduy Trans. I, 1982, 78, 2147-2156 Formation of Monodispersed Colloidal Cubic Haematite Particles in Ethanol + Water Solutions BY SHUICHI HAMADAT AND EGON MATIJEVIC* Department of Chemistry and Institute of Colloid and Surface Science, Clarkson College of Technology, Potsdam, New York 13676, U.S.A. Received 18th August, 1981 Cubic particles of colloidal haematite of narrow size distributions were prepared by aging ferric chloride solutions in water+ethanol mixtures. It was shown that a-Fe,O, formed by phase transformation from 8-FeOOH precipitated first. The kinetics of this conversion were followed at different temperatures, pH and chloride concentrations. The role of alcohol in the studied system is discussed and the rate of growth explained in terms of the theory of Burton, Cabrera and Frank.Numerous studies have shown that the precipitation of metal (hydrous) oxides from corresponding salt solutions is strongly affected by a variety of parameters, among which the pH, temperature, concentration of the reacting species and the nature of anions play dominant roles. Despite this sensitivity to experimental conditions, procedures have been established which result in the formation of colloidal dispersions of these materials consisting of particles exceedingly uniform in size and of different shapes, including spheres.' Specifically, ' monodispersed ' ferric basic sulphates are obtained by hydrothermal aging of ferric sulphate solutions2 and P-FeOOH and haematite from ferric chloride solution^.^ While P-FeOOH always appears in the form of acicular particles, a-Fe203 crystals of a variety of geometries can be prepared by minor changes in the experimental conditions.Only a few quantitative investigations on the effect of mixed solvents on the modification of the crystal habit have been reported in the literat~re.~ Since the composition and the shape of metal (hydrous) oxide particles depend on the medium from which they precipitate, it is to be expected that an addition of a miscible organic liquid to an aqueous solution should have a considerable effect on the properties of the resulting solids. Indeed, recently it was shown that on aging at elevated tem- peratures, ferric chloride solutions in mixed water + alcohol solvents may yield, under certain conditions, cubic haematite particles of great ~niformity.~ This process is investigated here in greater detail and the effect of the alcohol (ethanol) is discussed.It is shown that these dispersions are generated by phase transformation from a-FeOOH. The growth rate of cubic particles can be explained in terms of the theory of Burton, Cabrera and Frank. The availability of well defined dispersions of haematite is of considerable interest in various fundamental studies of this important material. EXPERIMENTAL MATERIALS Stock solutions of ferric salts at high concentrations (2.1-3.6 mol dm-3) were prepared by dissolving the corresponding salts in doubly distilled water at room temperature (except for t On leave of absence from Science University of Tokyo, Japan. 21472148 FORMATION OF CUBIC HAEMATITE PARTICLES ferric sulphate which was dissolved at ca.40 "C) and then filtering through a Millipore membrane of 0.2 pm pore size before storage. These solutions showed no visual changes on standing for 3 months. All chemicals used were reagent grade of highest purity. Millipore filters were carefully washed before use. SOL PREPARATION All sols were generated by heating solutions of ferric salts at specified temperatures (80, 90 or 99 "C) for varying periods of time (up to 200 h). The stock solutions were first mixed with the desired amounts of the corresponding acids (to prevent hydrolysis on dilution) and then with water and ethanol to give an alcohol content of 50% by volume: 40 cm3 samples of the thus prepared solutions were pipetted into 50 cm3 screw-cap culture tubes, tightly stoppered and placed in a preheated laboratory constant temperature oven or in an oil bath.The samples were spaced as far apart as possible to secure uniform heating. After aging, the tubes were quenched to room temperature and the particles were separated by centrifugation and washed with doubly distilled water. pH MEASUREMENT The acidity of each system was determined at room temperature (ca. 25 "C) with a Radiometer model PHM 26 pH meter before and after aging. An operational pH* unit (where * denotes pH of the mixed solvent systems) is defined as6 E - E(s) pH* = ~H*(s) + ~ 2.3 RT/F where pH*(s) is the value for a selected buffer solution of the same mixed solvent and E(s) and E are the corresponding e.m.f. values of the buffer and the sample solution, respectively.The pH*(s) values of buffers of mixed solvent systems6, were used to calibrate the pH meter. The pH* at higher temperatures (90 and 99 "C) in mixed solvents was evaluated from the expression8 (2) K,* (25OC) was taken from the work of Woolley et al.;9 for 50% (by volume) ethanol the extrapolation gave a value of pK,* = 14.67. Using for the heat of ionization of water in a 50% ethanol+water mixturelo AHo = 51.0 kJ mol-', and taking AH" to be constant over the temperature range of interest, pK,* at 90 and 99 "C was calculated to be 13.07 and 12.89, respectively. This estimate is based on the assumption that eqn (2) applies to mixed solvents. PH*(T,) = PH*(T,) x PK,*(T,)IPK,*(T,). ELECTRON MICROSCOPY For transmission and scanning electron microscopy the particles were redispersed in an ultrasonic bath and then deposited on corresponding sample holders.Particle size histograms were obtained from calibrated transmission micrographs. DETERMINATION OF FERRIC AND CHLORIDE CONTENTS The content of iron in the solutions and in the precipitated solids after aging was determined spectrophotometrically using 1,lO-phenanthroline as the complexing agent. For this purpose the particles were dissolved in hydrochloric acid. The chloride content in the solutions was analysed by the Volhard method. RESULTS FORMATION OF HAEMATITE PARTICLES IN WATER+ETHANOL MIXTURES To evaluate the precipitates formed by forced hydrolysis at 99 OC, ferric chloride solutions of concentrations from 1 x to 2 x 10-1 mol drnp3, which contained hydrochloric acid in concentrations between 5 x and 1 x 10-1 mol dm-3 and ethanol to give a final content of 50% (by volume), were aged for 48 h in a preheated laboratory oven. Precipitation occurred in all systems, except those containing theJ. Chern.SOC., Faraday Trans. 1, Vol. 78,part 7 S. HAMADA AND E. MATIJEVIC Plate 1 (Facing p . 2149)s. HAMADA AND E. MATIJEVIC 2149 -2.0- highest concentration of HCl. The solid phase always consisted of two kinds of particles, i.e. P-FeOOH and a-Fe,O,, which were identified by the X-ray powder diffraction technique.ll9 l2 Fig. 1 displays the FeC1,-HCl concentration domain covered in these experiments with the indication of the particle morphology in each system studied. The upper and the lower symbols refer to the P-FeOOH and haematite species, respectively.In most cases, P-FeOOH particles were rod-like with varying acicularity, whereas a-Fe,O, particles differed considerably in shape, depending on the I I R R S I N 5 E E I'E I I i N - I I d -0.5 I I I I I R C C S !I ? U F! RS R 5 5 R EC EC ic EC EC c E -1.0 R C R c R r! r! c ' R R R c c S S,R N c E N S I N 1 R I t i I N -3.5 -3.0 - 2.5 - 2.0 -1.5 -1 .o log( [ HC11 /mol dm-3) FIG. 1.-Concentration domain of FeCl, and HC1 in 50% (by volume) ethanol+water solutions which were aged for 48 h at 99 OC. The upper and the lower symbols indicate the morphology of the precipitated P-FeOOH and haematite particles, respectively: C, cubic; E, ellipsoidal; I, irregular; S , spindle; R, rod-like; N, no precipitation.concentration of the reacting components. Cubic particles of narrow size distribution were obtained under a limited set of conditions within the dashed boundary in fig. 1. The initial pH* of the solutions giving these uniform sols was 1.7; on aging the pH* dropped to 1 .O-1.4. Because of the great difference in their size, cubic haematite particles were readily separated from P-FeOOH needles by centrifugation (for 30 min at ca. 3000 r.p.m.) or by free settling (for 1-5 days). Plate 1 shows a transmission and a scanning electron micrograph of the separated cubic haematite particles. Such particles could be obtained under essentially the same concentration conditions (of FeCl, and HCl) as long as the temperature exceeded 80 OC and the ethanol content was > 40%.Additional experiments were carried out to investigate the effects of chloride ion concentration and of pH* on the formation of cubic haematite particles. For this purpose reference systems were chosen, which under the conditions shown in fig. 1 give uniform dispersions. In one series the concentration of C1- was adjusted with NaCl to between 5.7 x lo-, and 1.5 x 10-1 mol dm-, at a constant ferric chloride concentration (1.9 x lop2 mol drn-,) and a constant ionic strength of the supernatant liquid of 0.2 mol dmP3. The latter was adjusted with NaClO, after all the other ionic species in the solution were taken into consideration. In another series of experi- ments the ferric chloride concentration was varied between 1.9 x loP2 and 3.5 x lo-, mol dm-, at a constant C1- content of 0.2 mol dm-, adjusted with NaCl.In these experiments the initial pH* was kept at 1.6 _+O. 1. All the systems described21 50 FORMATION OF CUBIC HAEMATITE PARTICLES yielded cubic haematite particles (in addition to P-FeOOH) on aging at 90 O C for up to 150 h. When ethanol was substituted with methanol only irregularly shaped particles were generated. However, solutions containing propan-2-01 or t-butyl alcohol yielded cubic particles when the alcohol content was > 30% (by volume). Aging solutions of ferric salts other than the chloride (nitrate, perchlorate and sulphate) in the presence of ethanol did not produce well defined precipitates under the same conditions. Solids from systems containing nitrate and perchlorate anions mostly consisted of irregular haematite particles, while ferric basic sulphate formed in solutions of iron(1Ir) sulphate.RATE OF CONVERSION OF P-FeOOH INTO HAEMATITE The formation ofcubic haematite particles in acidified ferric chloride water + ethanol solutions was studied over extended aging times at two different temperatures (90 and 99 "C). In all cases p-FeOOH appeared first as the solid precipitate, while haematite resulted from a phase transformation process. The precipitation of 8-FeOOH was essentially completed after ca. 20 min of aging, whereas the formation and growth of cubic haematite particles proceeded over a longer period of time. The conversion of 8-FeOOH into a-Fe,O, was followed by chemical analysis of the amount of each of the two ferric compounds in the solid phase.For this purpose the precipitate was removed from the dispersion by ultracentrifugation of samples at 20000 r.p.m. for 1 h and subsequent filtration through a Millipore membrane of 0.08 pm pore size. The filtrate was completely free of solids, as verified by the absence 10, I 0 5 0 100 150 tlh FIG. 2 . 4 ~ ) Percentage of soluble Fe"' species remaining in the supernatant solution after aging at (0) 90 and (0) 99 "C for various periods of time. Initial solution contained FeCl, (1.9 x lop2 mol dm-,) and HCl (1.2 x mol dm-3) in 50% (by volume) ethanol+water mixture. (b) Percentage of iron as j?-FeOOH in (a). ( c ) Percentage of iron as a-Fe,O, in (a).s. HAMADA AND E. MATIJEVIC 2151 I I I 0 5 0 100 150 200 t l h FIG. 3.-Change in the length (l/pm) of the edge of cubic haematite particles as a function of time of agmg of 50% (by volume) ethanol+water solutions of FeCI, (1.9 x mol drn-,) and HCl (1.2 x lo-, mol dm-,) at (A) 80, (0) 90 and (0) 99 OC.O.* I 0 L--dldu 0.2 0.4 0.6- 0.8 1.0 1.2 1.4 UCtm FIG. 4.-Histograms of cubic haematite particles precipitated in a 50% (by volume) ethanol + water solution of FeC1, (1.9 x mol drn-,) and HC1 (1.2 x lo-, mol drn-,) aged at 99 O C for (a) 16, (b) 40, (c) 64 and ( d ) 90 h, (i= average length of the edge of the cube in pm).21 52 FORMATION OF CUBIC HAEMATITE PARTICLES of a Tyndall cone. The P-FeOOH and a-Fe,O, particles were then separated as described in the experimental section. The time dependence conversion of rod-like P-FeOOH into cubic a-Fe,O, particles is shown in fig.2. It is clear that the rate of appearance of haematite is the same as the rate of disappearance of p-FeOOH from the mixture. The process was followed for periods longer than shown in this figure and the conversion continued until all the P-FeOOH had disappeared. As one would expect, haematite formation was considerably faster at the higher temperature. The concentration of unchanged ferric species in the supernatant solution remained unchanged during the phase transforma- tion process. The pH* dropped only during the first ca. 20 min and then remained constant. No change in the chloride content in the supernatant solution was noted during the entire duration of the aging experiment. The increase in the size of the cubic haematite particles with the time of heating at 80,90 and 99 OC is shown in fig.3. Except at the very beginning of the conversion process, the average length of the edge of the cubes grows linearly with time. The size distribution at different aging times is illustrated in fig. 4, which gives the histograms obtained from transmission electron micrographs for a sample kept at 99 O C . The apparent rate constants calculated from the linear portions of fig. 3 obey the Arrhenius equation from which the energy of activation for the conversion process of P-FeOOH into haematite is calculated to be E, = 110 kJ mol-l. Finally, fig.5 (a) shows that the apparent rate constant (R) of the phase transforma- tion is independent of the Cl- concentration. Fig. 5(b) indicates that R decreases with increasing acidity.The pH* was varied by altering the FeCl, concentration. Since /?-FeOOH precipitates rapidly, and haematite is formed by phase transformation, we assume that pH* is the only parameter affecting the rate constant in this case. - I s: (0 1 1 -1.3 -1.2 -1.1 -1.0 -0.9 -0.8 log( [ C1-I /mol dm-3 1 I I I I -1.3 -1.2 -1.1 log( [ H'I /mol dm-3 ) FIG. 5.--(a) Dependence of the apparent growth rate constant (R/pm h-*) of cubic haematite on the chloride concentration. Particles were precipitated in 50% (by volume) ethanol + water solutions of FeC1, (1.9 x mol dm-,) and HCl+NaCl (5.7 x 10-2-1.5 x LO-' mol dm-,) aged at 90 OC. The pH* of the supernatant solutions was 1.2. (b) Influence of pH* of the supernatant solution on the apparent growth rate constant (R/pm h-l) of cubic haematite particles precipitated in 50% (by volume) ethanol +water solutions of FeC1, (1.9 x 10-2-3.5 x mol drn-,) at a constant chloride content of 0.2 mol dm-, (adjusted with NaCl), aged at 90 OC.s.HAMADA AND E. MATIJEVIC 2153 DISCUSSION Precipitation of haematite from ferric chloride solutions may proceed either directly or through phase transformation. If a-Fe203 is formed by conversion from amorphous ferric hydroxide, nucleation and crystallization take place within the existing solid phase.13 On the other hand, a crystalline precursor (e.g. P-FeOOH) requires the original solid phase to dissolve first, followed by crystal growth of haematite.l49 l5 This work shows that cubic haematite particles in water +ethanol solutions of ferric chloride are generated through conversion from the originally precipitated P-FeOOH.There is strong evidence that the latter dissolved on longer aging at elevated temperatures with a simultaneous nucleation and growth of a-Fe20,. The effect of increasing acidity (fig. 5) also favours a dissolution mechanism. It is of interest to discuss the reasons why a simple alcohol exerts such a profound effect on the morphology of this common ferric oxide. Lower alcohols enhance the hydrolysis of ferric ions16* l7 and even more so the ferric chloride solute complexation.18-21 These complexes are known to be precursors in the formation of P-FeOOH. Thus, the first effect of alcohol is to promote the precipitation of the rod-like ferric oxyhydroxide. The generation of this solid phase is completed much more rapidly (within 10-20 min) in water+ethanol mixtures than in corre- sponding pure aqueous ferric chloride solution^.^ The fast precipitation of P-FeOOH is also indicated by the sharp decrease in pH* which occurs only during the initial period of aging. The phase transformation of p-FeOOH to a-Fe203 involves no change in pH*, as verified by the experiments.Furthermore, it is not affected by the amount of chloride ions in the solution (at least not within the concentration range studied). The monodisperse nature of the final product is probably due to the slow dissolution of P-FeOOH, which allows for uniform growth of haematite particles after nucleation by incorporation of solute complexes. Thus, the necessary supersaturation in the particle-forming species is maintained until the entire amount of the initial solids is used up.The influence of alcohol on the conversion of P-FeOOH to haematite is most pronounced with respect to particle morphology. The resulting a-Fe203 particles of cubic morphology showed two strong X-ray powder diffraction peaks at d = 0.366 and 0.184 nm, which are characteristic of haematite, while the normally observed strongest peak at 0.269 nm was much weaker.12 Such structure modification reflects a different growth mechanism at crystal faces, which depends on the nature of the solute/solvent interactions. The latter are greatly affected by the presence of alcohol molecules. The influence of mixed solvents on the crystal morphology and on the crystal growth mechanism was examined in terms of the surface entropy factor, a, which is defined as22 24 a = 4 ~ / k T (3) (4) where 4ss, q5ff and dsf represent potential interaction energies between solid blocks, fluid blocks and solid/fluid blocks, respectively.Eqn (3) may be approximated by a relationship which shows that a is proportional to the entropy of phase transition, AS. 23 where k is Boltzmann’s constant and E = 0.5 4,,+0.5 4ff-@,f An estimate of the surface entropy factor can be obtained fromz4 a = 4y/kT ( 5 ) F A R 1 702154 FORMATION OF CUBIC HAEMATITE PARTICLES in which the edge energy y is given as y = d2F (6) F = AHs/2ba2 (7) where d is the height of the monomolecular step for a given crystal face and F is the surface energy25 AH, is the enthalpy of dissolution, a is the lattice spacing and b is the ratio of the binding energy of a molecule in a kink site to the energy required to form a new surface by cleavage of the crystal.Eqn ( 5 ) is strictly applicable only for a cubic lattice, but should offer estimates for other The surface entropy factor a < 3.2 should indicate an inherently rough surface with no energy barrier to crystal growth at low supersaturation. If 3.2 < a < 4.0, the surface is assumed to be smoother but with surface nucleation still possible. Finally, values of a > 4 are characteristic of smooth surfaces, in which case the crystal growth at low supersaturations is only possible in the presence of steps and the diffusion law of Burton, Cabrera and Frank should be applicable.26 Addition of alcohol to an aqueous solution influences q5ff and dsf and causes a given face to grow by a different mechanism.The calculation of a is difficult in general2' and even more difficult for haematite particles in ethanol + water mixtures because of insufficient information in the literature on the necessary parameters. However, a rough estimate yields for this system a very high a value (> 30). Apparently highly hydrated solute species in the presence of alcohol undergo dehydration, causing an increase in entropy. Alcohol acts as an impurity, destroying the normal interfacial structure and providing an easier transition from solution to crystal, thus affecting the growth rate. According to Burton et a1.26 and Bourne et a1.,28 at high supersaturations (0 4 ol, where 0 and o1 are the supersaturation and the critical supersaturation) a crystal may grow linearly according to : In the former case b assumes values between ca.2 and 3. where R is the linear growth rate constant, is the retardation factor for entry of a molecule into a step, l2 is the volume of the molecule, no is the number of molecular positions per unit area on a given surface, h is Planck's constant and AGdesolv. is the activation free energy for desolvation of the constituent species and its entry into the adsorbed layer. Assuming that 0 is reasonably constant at three different temperatures (80, 90 and 99 "C), eqn (8) can be expressed as follows: C = In PQnoa(k/h). (10) The value of AGdesolv. can thus be estimated from eqn (9) using the experimental values of the growth rate at different temperatures.Fig. 6 shows that the rate data given in fig. 3 yield a linear relationship when plotted in terms of eqn (9), from which AGdesolv. is calculated to be 110 kJ mol-l. Note that this value is very close to several listed total evaporation energies, W, for crystal growth from the vapour phase of different materials.26 Using AGdesolv. = 110 kJ mol-1 and reasonable values for the other parameters in eqn (lo), /? is estimated to be 3 x This value is lower than that obtained for crystal growth from the vapour phase (0.1-1. 1),26 which is to be expected if chemical reactionss. HAMADA AND E. MATIJEVIC 2155 -27.5 - 28.0 n - ;C -28.5 - I m 5 . - I -29.0 v E - - 29.5 -30.0 I I I 2.70 2.75 2.80 2.8: 1 0 3 ~ 1 ~ FIG. 6.-Plot of the data shown in fig. 3 according to eqn (9).at the crystal/solution interface take place, as is the case when ferric solute complexes are condensed to form an 0x0-bridge network. The effects of alcohols on crystal growth from aqueous solutions studied before24$ 2 8 y 29 involved simpler systems than that described in this work. It is therefore of interest that the same fundamental principles may be applied to a case as complicated as the conversion of P-FeOOH into a-Fe,O,. The existing theories give a reasonable explanation for the formation of haematite particles of cubic morphology. This work was supported by the Electric Power Research Institute, contract no. RP-966-2. E. Matijevid, Acc. Chem. Res., 1981, 14, 22. E. Matijevid, R. S. Sapieszko and J.B. Melville, J. Colloid Znterface Sci., 1975, 50, 567. E. MatijeviC and P. Scheiner, J. Colloid Interface Sci., 1978, 63, 509. J. W. Mullin, Chem. Ind. (London), 1980, 9, 372. S. Hamada and E. MatijeviC, J. Colloid Interface Sci., 1981, 84, 274. R. G. Bates, Determination of p H (John Wiley, New York, 1964), p. 222. ’ M. Paabo, R. A. Robinson and R. G. Bates, J. Am. Chem. SOC., 1965, 87, 415. R. T. Lowson, Aust. J. Chem., 1974, 27, 105. E. M. Woolley, D. G. Hurkot and L. G. Hepler, J . Phys. Chem., 1970, 74, 3908. lo G. L. Bertrand, F. J. Millero, C. Wu and L. G. Hepler, J. Phys. Chern., 1966, 70, 699. l1 J. D. Bernal, D. R. Dasgupta and A. L. Mackay, Clay Miner. Bull., 1959, 4, 15. l2 V. Kastalsky and M. F. Westcott, Aust. J . Chem., 1968, 21, 1061. l3 W. R. Fischer and U. Schwertmann, Clays Clay Miner., 1975, 23, 33. l4 R. J. Atkinson, A. M. Posner and J. P. Quirk, J. Znorg. Nucl. Chem., 1968, 30, 2371. l5 R. J. Atkinson, A. M. Posner and J. P. Quirk, Clays Clay Miner., 1977, 25, 49. G. Popa, C. Luca and E. Josif, Z. Phys. Chem. (Leipzig), 1963, 222, 49. 70-22156 FORMATION OF CUBIC HAEMATITE PARTICLES l7 E. J. Bowers and H. D. Weaver, Proc. Indiana Acad. Sci., 1961, 17, 101. la K. Bridger, R. C. Pate1 and E. Matijevid, Polyhedron, in press. Is G. Wada and Y. Kobayashi, Bull. Chem. SOC. Jpn, 1975, 48, 2451. 2o G. D. Brykina, N. L. Filippova and T. A. Belyavskaya, Zh. Neorg. Khim., 1976, 21, 2936. G. S. Murty and M. N. Sastri, J . Indian Chem. SOC., 1977, 54, 783. 22 K. A. Jackson, in Liquid Metal and Solidijication (Am. SOC. Metals, Cleveland, Ohio, 1958), p. 174. 23 P. Bennema and G. H. Gilmer, in Crystal Growth: An Introduction, ed. P. Hartman (North Holland, 24 J. R. Bourne and R. J. Davey, J. Cryst. Growth, 1976, 36, 278. 25 W. J. Dunning, in Physics and Chemistry of the Organic Solid State, ed. D. Fox, M. M. Labes and A. Weissberger (Interscience, New York, 1963), vol. 1, p. 412. 26 W. K. Burton, N. Cabrera and F. C. Frank, Philos. Trans. R. SOC. London, Ser. A , 1951, 243, 299. 27 B. Simon and R. Boistelle, J. Cryst. Growth, 1981, 52, 779. 28 J. R. Bourne and R. J. Davey, J. Cryst. Growth, 1976, 36, 287; 1977, 39, 267; 1978, 43, 224; 1978, 29 J. R. Bourne, R. J. Davey and J. McCulloch, Chem. Eng. Sci., 1978, 33, 199. Amsterdam, 1973), p. 263. 44, 613. (PAPER 1 / 1335)