Document gb07Mw92YMVB6yRj8gkRKdG4Q

19S Journal of The American Ceramic Society--Johnston Vol. 47, No. 4 Cbrysotile asbestos can be dispersed essentially to the col loidal state when agitated at high speed in water in the presence of colloidal alumina.. Presumably the coating of colloidal alumina, evidenced in Fig. 4, provides a strong positive repulsive charge which promotes separation of ulti mate fibrils. Such dilute (0,5 to 2%) viscous translucent sols <*an be dried to semitransparent asbestos sheets in which the alumina plays the additional role of binder. Kaolin and graphite, when coated with colloidal alumina, are adsorbed from dilute solution onto surfaces such as glass or cellulose and remain as a thin layer which is not easily rinsed off. An excess of colloidal alumina must be avoided in this case; only enough must be added to coat the clay or graphite, since if more is present, the glass or cellulose will be covered preferentially with the free colloidal alumina to the exclusion of the larger alumina-coated particles. The foregoing phenomena suggest means for depositing very uniform layers of alumina-coated particles of the order of Ip in thickness on a variety of surfaces. Effects on lubricity, electrical conductivity, light transmission, and other phe nomena might be expected, depending on the nature of the deposited material. Acknowledgment The writer wishes to acknowledge the cooperation of Vernon Keirstead in preparing the electron micrograohs. Jr April Oxidation-Reduction Equilibria in Iron-Containing Glass by W. D. JOHNSTON Research end Engineering Cen!er, Pittsburgh Corning Corporation, Pittsburgh, Pennsylvania The oxidation-reduction equilibrium between tion state of iron in the melt was then determined by chemical ferrous and ferric iron in Na-0-2Si0j glass melts analysis. was studied by equilibrating melts with various The glass used had the nominal composition Na*0-2Si0- -j- oxygen partial pressures. Exceedingly long ~2.5 wt% FejOj. Several master melts were prepared from equilibration times were required to obtain mean reagent-grade Na-COj, SiOj, and Fe*Oj in a platinum crucible. ingful results. The reaction appears to be dif Small quantities of the master melts (2 or 3 g) were placed fusion-controlled and may be expressed as in 5- or 10-cm* Morganite recrystallized aiumina crucibles or 20s*~ + 4Fe,+ ;=* 4Fei+ -f Oj. Data are pre in some cases, as mentioned later, in platinum crucibles. The sented in which the valence of iron varies from crucible was lowered by a platinum or molybdenum wire into predominantly +3 to predominantly +2. the hot zone of a vertical tube furnace under a controlled atmosphere. At the end of the experiment, the crucible was 1. Introduction quenched by raising it out of the hot zone. Samples were repeatedly reground and refired for periods of at least 20 hours ERY little quantitative work has been done to elucidate until a constant chemical analysis indicated that equilibrium Voxidation-reduction equilibria of variable valence ions in glass melts. The state of knowledge has been re had been attained. The furnace tube was constructed of mullite. The temperature of the furnace was controlled to viewed by Tress,1 who was forced to relate thermodynamic within =fclC as indicated by a Pt-PtlORh thermocouple en equilibria for pure oxide systems to what was presumed to be cased in a mullite sheath suspended inside the furnace tube an oxidation-reduction sequence in glass obtained from the beside the sample. The temperatures used were 1100, observation of glass colors. An attempt at a more quantita 1200, 1270, 1300, and 1450C. tive study has since been made by Baak and Horayak,1 who Atmospheres used included oxygen, air, oxygen-free COj, measured the iron oxidation state in air as a function of tem CO-COj mixtures, CO in equilibrium with carbon (the cruci perature in glass melts containing small additions of iron ble in this case), and Hi saturated with H-0 at 0C. The CO* oxide. In their work, the iron was almost completely tri- COi mixtures were metered using constant-head-type capillary valent. A more extensive investigation of the iron-oxygen equilibria in low-viscosity, high-iron silicate slags of metal lurgical interest has been made by Turkdogan and Bills.* In their work, the iron valence was varied from predominantly divalent to predominantly trivalent A general review of the literature dealing with iron-containing glass has been made Received July 19, 1963; revised copy received November 8, by Weyl.4 In no case has a systematic investigation of the iron oxida tion-reduction equilibria been made on a high-viscosity glass 1963. The writer is research chemist. Research and Engineering Cen ter, Pittsburgh Corning Corporation. 1 H. J. Tress, "Thermodynamic Approach to Redox Equilibria melt in which the oxidation state of the iron was varied over a wide range. The purpose of the present work was to deter mine the conditions for the formation of the various valence states of iron and to permit an oxidation-reduction equation in Glasses," Phys. Chem. Glasses, l [6] 196-97 (1960); CeramAbstr., 1961, October, p. 238t. 1 T. Baak and E. J. Homyak, Jr., "Iron-Oxygln Equilibrium in Glass: Effect of Platinum on Fe*VF-e*+ Equilibrium." JAm. Ceram. Soc.. 44 [11] 541-44 (1961). to be written describing this process. II. Experimental Procedure Glass melts containing small quantities of dissolved iron oxide were equilibrated in various atmospheres. The oxida * E. T. Turkdogan and P. M. Bills, "Thermodynamic Study of FeO-FejOi-SiOj, FeO-Fe-Oj-Pjand FeO-FeiOi-SiOr-T*^* Molten Systems," J. Iron Steel Inst. [London), 186, 329-39 0957). * W. A. Weyl, Coloured Glasses. Dawsons of Pall Mall. London, 1959. 541 pp. (' fiowme mixtur t- Che: X total i: cr accord was de sulfate ; T added | * Analv? tion ti ? In son mined howevi subtra prefers \$' creasec creasec Jr the air bilitv t ysis. ( " was lo5 The a: Vr slightb , .v* expectc ; Plat yj'reducsr ; ;? tent ra _JL to alios UCt Ob'. $2' and ga be four It did < crucibl negligi' FeJ VF the rat The allow equili'o: found ' - ' iadicat ty. observ t tent, g} Hum is tydepii! c |*f. Points*" 194 Journal of The American Ceramic Society--IUr Vol. 47, No. * Measurements of these charges have been made by raising ~T,perature and permitting the charge to flow out of the n. This is an indirect approach and the nature of the and its distribution is difficult to deduce. A direct icdmique or optical technique would be desirable, harges can also arise due to artificial carriers injected & by ordinary light as discussed by Rohatgi* and by * Again direct measurements of charge densities ot been attempted. jets represent another, form of charge distribution ; in insulators. Electrets made of various organic organic compounds have been studied extensively but s not often included." The work of Gubkin and d4* indicates that a space charge can be frozen in as Pyrex-brand glass is cooled under a field. They state that the charge observed 30 minutes later at 50aC is opposite to the polarity of the electrode used to induce the charge. Acknowledgments The writer's thanks are due to various staff support groups at the Philco Research Laboratories for assistance in the preparation of this series. tt R. C. Nelson, "Sensitization of Photoconductivity in Glass by Dyes." J. Opt. See. Am., 50 1101 1029 (1960); Ceram. Abstr., 1961, January, p. 6d. 41 A. N. Gubina and G-1. Skanavi, "Some New Electrets from Inorganic Dielectrics," Soviet Pay's. JETP (English Transl.), 5 (U 140-43 (1957). Adsorption of Colloidal Silica on Alumina and of Colloidal Alumina on Silica by R. K. HER Industrial and Biochemical* Department, Experiments! Station, E t du Pont de Nemours &. Company, Incorporated, Wilmington, Delowsre The mutual adsorption of colloidal silica on alumina and of colloidal alumina on silica and silicate materials occurs in aqueous suspension at about pH 4. It is shown that the adsorption of colloidal particles on the surface of opposite charge is limited to essentially a monoparticle layer. An adsorbed layer of fibrils of colloidal alumina on the surfaces of silica, asbestos, graphite, and finely divided clay is shown in electron micrographs. The effect of the ad sorbed colloid on dispersibility of the substrate materials is discussed. 1. Introduction t is the purpose of this paper to present examples of the I adsorption of colloidal alumina and of colloidal silica from aqueous sols onto the surface of particles of some typical ceramic raw materials and related substances and to discuss the effects of such interaction. It is well known that when two sols of opposite sign are mixed, mutual coagulation may occur.1 * 1Th*u*s when sols of colloidal silica and colloidal alumina are mixed,' there is a marked increase in viscosity owing to mutual flocculation. When, however, either one type of particle or the other is present in large excess, coagulation does not occur because the particles in the minority become covered with the particles that are in excess. For example, particles of colloidal gold are stabilized or protected by the adsorption of protein.* Thus there is a reversal of charge when one colloid is mixed with another protective colloid.1 However, the protective colloids usually have been organic in nature, e.g., proteins or gums. The coating of one type of inorganic particle by an other inorganic colloid is less well known, although the re versal of charge of day particles by adsorption of colloidal alumina has been reported by Bugosh.4 In the present paper it is shown that when a suspension of alumina is stirred into an excess of colloidal silica, each partide of alumina becomes coated with an adsorbed layer of particles of colloidal silica. The alumina partides which were pre viously positively charged take on a negative charge owing to the layer of adsorbed silica. Since the coated alumina partide now has the same charge as the surrounding silica partides present in excess, there is no flocculation. Similarly, when silica partides in a ball-milled silica slip are mixed with an excess of colloidal alumina, the surface of the silica becomes coated with a layer of alumina partides. This adsorption is demonstrated by chemical analysis and by changes in specific surface area. II. Experimental f 1J Materials Ball-milled powders of dense, fused alumina and of fused silica glass were kindly furnished by J. X>. Walton of the Georgia Institute of Technology. Amorphous silica I (not acid washed): This powder had been wet ball-milled, dried, and then dry-milled; specific surface area (nitrogen adsorption), 7.2 m* per g; alumina, 0.57% by weight. Received June 21, 1963; revised copy received October 28, 1963. The writer is a research manager, Industrial and Biochemical? Department, Experimental Station, E. I. du Pont de Nemours & Company, Incorporated. 1 H. B. Weiser, Colloid Chemistry, 2d ed.. Chapter 17. John Wiley & Sons, Inc., New York, 1949. 440 pp. * J. Th. G. Overbeek. "Specific Effects of Flocculating Ions" pp. 310-18 in Colloid Science: VoL I, Irreversible Systems Edited by H. R. Kruyt. Elsevier Pubfishing Co., Inc , Houston 1952. 389 up.; Ceram. Abstr., 1953, September, p. 169*. * H. G. Bungenberg de Jong, "Reversal of Charge Pheoomen; in Mixtures of Colloids"; pp. 321-34 in Colloid Science: Vol. II Reversible Systems. Edited by H. R. Kruyt. Elsevier Publish ing Co., Inc., New York, 1949. 753 pp.; Ceram. Abstr., 1951 May, p. 95i. 4 John Bugosh," 'BaymaV Colloidal Alumina," Am. Ink Make 40 (5] 54-55, 57-58,130-31 (1962). Fraction --* Solids recovered (%) Specific surface area (*/) Composition (?e) Silica Alumina Table 1. Adsorption of Colloidal Alumina on Silica Silica I with colloidal alumina Silica II with colloidal alumina Coarse Fine Colloidal Coarse Coarse 75 4.0 5.3 23 11.3 42 84 4.3 71.5 2.6 Silica II, untreated Fine Colloidal 8.7 15 6.0 20 97.7 1.2 92.2 4.82 8.31 97.40 0.85 98.2 0.52 97.4 0.9 0.71 Amorphous silica II {acid washed): The foregoing powder was slurried in concentrated nitric add for 24 hours, washed with distilled water by decantation until the pH of the suspension was 4.0, and kept as a suspension; spedfic surface area, 6.9 m! per g; alumina content, 0.50% on a solids basis (see also Table I). Alpha alumina: The powder as received was treated with concentrated nitric acid and washed as above; specific surface area, 2.3 m1 per g; silica content, 1.37% on a solids basis. Colloidal silica**: Aqueous sol containing 30.0% SiO*; ' pH, 9.8; ratio SiOj/NaiO, 95; specific surface area, 200 m5 per g of silica; approximate particle diameter, 15 tom Colloidal alumina^: Dispersible powder; AlOOH, 83.1%; acetic add, 9.8%; sulfate ion, 1.7%; moisture, 5.0%; parf: tides fibrillar, approximately 5 m^t in diameter; spedfic sur- face area, 27o m1 per g; pH of 4% sol in water, about 4. - (2) Procedure The colloidal alumina powder was agitated in distilled water at a concentration of from 1 to 5% in a high-speed blendor foT * a few minutes and the dispersion was left to stand overnight. The colloidal silica was diluted with distilled water just before use and addified to pH 3 or 4 with acetic add. The general procedure was to pour a dilute, weakly acidic suspension of the coarser particles into an excess of colloidal solution of the smaller partides of opposite charge at the same pH while the mixture was intensively stined. The coarser particles were then recovered by settling or centrifugation, resuspended in addified water, and centrifuged to wash out excess colloid. To coat the fine amorphous silica powder with colloidal alumina, a 20 to 30% slurry of the silica in water adjusted to pH 4 with acetic add was nm into a dilute sol of colloidal alumina at the same pH, with violent agitation in a blendor, for 1 or 2 minutes. At least 5 g of colloidal alumina solids per 100 g of silica were present to ensure an excess of alumina over that required to coat the silica. The mixture was per mitted to stand while the coarser and finally the finer frac tions of silica powder had settled out and had been removed. The colloidal alumina remaining in the supernatant fluid was discarded. The silica fractions were repeatedly suspended in water at pH 4 and recovered by centrifuging until the wash water was free from colloidal alumina. A sample of silica II was similarly suspended in water and separated into fractions as a control. The fine alumina powder was coated with colloidal silica by slowly adding the alumina suspension (pH of about 3) to a 5% sol of colloidal silica previously acidified with acetic add. In this case 15 g of silica solids per 100 g of alumina powder were present. . A variety of powders were coated with.colloidal alumina by suspending them in water at about pH 4,'adjusting with acetic add, adding the suspension to an excess of a sol of colloidal ,, alumina at about the same pH, and separating the coated ''-'material from excess colloidal alumina and washing, as prer'-viously described. When the material to be coated contained i-'substantial amounts of exchangeable cations, a preliminary rr washing with dilute acetic add was necessary to obtain a y suspension that would remain at pH 4. II!. Results (1) Silica Coated with Colloidal Alumina Analyses of the recovered fractions of silica powders are given in Table I. It should be pointed out that fractionated silica powder, derived from a fused-alica glass, contained, on the average, about 0.7% Al;Oj. In the coated samples, the adsorbed colloidal alumina is thus indicated by the amount of alumina in excess of the original amount. The most striking observation is that colloidal alumina was adsorbed on silica I but not on silica II which had been treated with concentrated nitric add. Such treatment un doubtedly removed metal ions such as aluminum from the surface. Chemical analysts showed that some alumina was still present, but this no doubt was in the interior of the pat tides. It was also observed that the silica II powder was not peptized by the colloidal alumina but remained flocculated and settled rapidly so that there was no separation into finer and coarser fractions. On the other hand, silica I, which adsorbed colloidal alumina, was peptized. The lower adsorption of colloidal alumina on pure silica than on an aluminosilicate has been previously noted in this lab oratory. For example, P. C. Yates has observed that colloidal alumina is strongly adsorbed at pH 4 on commercial heatdeaned glass fibers, which contain some alumina, but to a much less degree on add-treated silica fibers. It is postulated that for the colloidal alumina to be adsorbed, the silica surface must bear a negative charge. As has been previously discussed by the writer,* the presence of alumino silicate ions on the surface of silica maintains a negative charge on the surface even at a pH as low as 4, whereas on pure silica there is little adsorption of hydroxyl ions or development of a surface charge at this pH. However, except after being treated with strong add, most siliceous surfaces are contam inated with traces of iron or aluminum which provide sur face sites that remain negatively charged down to pH 3 or 4. The thickness of the adsorbed layer of colloidal alumina on the surface -of different fractions of silica I can be calculated from the percentage of adsorbed alumina and the surface area of the coated material. Since the writer has previously shown that the spedfic surface area, as determined by nitrogen adsorption, is not appredably changed when a colloidal alumina sol is dried to a powder, it is also likely that the sur face area of the fibrils is unchanged when the fibrils are adsorbed randomly on the surface of a silica partide. Con sidering the fine fractions of coated and uncoated silica I, the increase in colloidal alumina content of the coated fraction is 4.82% -- 0.9%, or a gain of 3.92% AljO*. This corresponds "Ludox" HS; registered trademark for Du Boat colloidal silica. t "BaymaV; registered trademark for Du Pont colloidal alumina. * R. K. Her, Colloidal Chemistry of Silica and Silicates, Chap ter VIII. Cornell University Press, Ithaca, New York, 1955. 324 pp.; Ceram. Abslr., 1956, May, p. 107f. ..-s -t *: -j-[ r 1. :c L :';C c.;. 7 s a -:M?. ; 4 A. *??-/?? Fig. 1. Alpho atumino particles coated with colloidal silica. i :i . ll \, - ; t u i.;; f to 5.5% by weight of colloidal alumina which has a specific surface area of 275 m* per g. Thus the colloidal alumina contributes an area of 15.2 m* per g to the coated powder, which has a measured specific area of 23 ml per g. The difference, 7.8 m1, is due to the area of the silica itself which constitutes 92.2% of the sample. The specific surface area of the silica is thus 8.2 m* per g. The colloidal alumina is known to dry to a porous film hav ing a density of about 1.0 g per cm*. Assuming that the alumina is spread evenly over the silica surface, the thickness of the alumina coating is calculated to be about 5 mu. This is about the thickness of a colloidal alumina fibrillar particle; thus the silica surface is covered bv about a monolayer of alumina particles lying flat on the surface. Similar calculations for the coarse fraction of coated silica I indicate that it has a specific surface area of 1.5 ms per g and that the alumina coating is 4.3 mu thick. The colloidal frac tion contains a higher proportion of alumina, indicating the greater difficulty of separating the extremely fine coated silica from the excess of colloidal alumina panicles; the calculated coating thickness is about 10 edm. (2) Alumina Coated with Colloidal Silica The alpha alumina coated with colloidal silica was recovered in an 84% yield; the finer material was lost in washing. The recovered silica-coated alumina had a specific surface area of 4.1 m5 per g and contained 2.78% silica and 96.5% AljOj. By a calculation like that used for the foregoing silica powder, the thickness of the adsorbed silica layer is estimated, based on the fact that the colloidal silica has a specific surface area of 200 mJ per g and dries to a porous solid having a density of 1.0 g per cm*. By difference the specific surface area of the recovered alumina is 1.3 ms per g, corresponding approxi mately to 1m particles. The calculated thickness of the adsorbed layer of colloidal silica is 10 mM- Since the average particle size of the colloidal silica is 15 mM, the thickness of the coating thus corresponds approximately to a single layer of silica particles. This is borne out by the electron micrograph in Fig. 1. (3) Colloidal Alumina Adsorbed on Other Materials It has been observed that colloidal alumina is not strongly adsorbed on surfaces if the sol is too strongly acidified, e.g., below pH 2. Under these circumstances, it is believed that some alumina passes into solution as basic aluminum ions and that these are adsorbed on the negative surfaces and com pete with or displace the colloidal alumina particles. On the other hand, adsorption is also difficult to demonstrate if the pH is higher than 5 or 6, because the colloidal alumina is flocculated and it is difficult to distinguish adsorbed alumina particles from mechanically trapped and flocculated aggre- Fig. 2. Crystalline hydrated silica platelets coated with colloidal alumina. gates. For this reason, adsorptions are carried out in the pH range 3.5 to 4.5. (A) Crystalline Hydrated Silica: The crystalline sodium polysilicate described by McCulloch6 was prepared and ex tracted with acetic add to remove the sodium and then coated with colloidal alumina. In Fig. 2 there are two extremely thin rectangular sheets of this silica that are transparent to the electron beam, so that the adsorbed layer of colloidal alumina fibrils can be deaily seen. It is evident that there is orientation of the alumina fibrils, many of which lie parallel to the edges of the silica sheets, suggesting that the negative ionic charges are regularly arranged on the surface of the crystal lattice of the silica. {B) Kaolin: Four hundred grams of day* were suspended in 3320 g of water aridified with 30 g of acetic add. When 2S0 g of a 7% colloidal alumina sol was added with intensive mix ing and the mixture was diluted tenfold and again adjusted to pH 4, the clay was so well dispersed that the resulting sol could be passed through E and D No. 615 filter paper under suction. It was only- with difficulty that the coated day could be separated from the excess colloidal alumina by centrifugmg. As shown in Fig. 3, the kaolin is still not quite free of un adsorbed alumina. It is dear, however, that the surfaces of the day platelets are covered with a crisscrossed layer of fibrils of colloidal alumina. It was found that on some sam ples of kaolin, adsorption of colloidal alumina was quite poor. Possibly in these days the surface was already partly covered with positively charged basic aluminum ions from the original natural environment. 1 Leon McCulloch, "New Highly Siliceous Soda-Silica Com pound, " J. Am. Chem. Soc., 74 [10} 2453-56 (1952); Ceram. Abstr.. 1953, July, p. 119b. * "Hydrite"-UF, Georgia Kaolin Company, Elizabeth, N. J- April 1964 . Adsorption of Colloidal Silica on Alumina and of Colloidal Alumina on Silica 197 Fig. 3. Kaolin platelets coated with colloidal alumina. Fig. 4. Uncoated dtrysotile asbestos fibers and fibers coated with. colloidal alumina (right). (O Chrysolite Asbestos: This mineral was first washed with 2% nitric acid to remove magnesium ions from the sur face so that a suspension could be prepared in water with a _ pH of 4. After treatment with colloidal alumina, a small '>r amount of the coated fibers was repeatedly washed by dispersing about 0.1% in water at pH 4 and centrifuging, v Coated and uncoated asbestos fibers are shown in the elec; tron micrographs of Fig. 4; the fine tangled layer of adhering J: colloidal alumina can be seen on the coated sample. V (D) Graphite: In Fig. 5, some thin graphite platelets* permit the adsorbed fibrils of alumina to be observed. It is likely that the surface of graphite contains sufficient car boxylic add groups from oxidation to provide the negatively charged sites necessary for adsorption of colloidal alumina. (V. Discussion The surfaces of most siliceous substrates, induding fused silica, glass, and silicate minerals, bear a negative ionic charge in water at pH 4 and will adsorb a layer of colloidal alumina partides. This surface layer then bears a positive charge and the original charge of the substrate is thus reversed. The positive charge on metal oxide partides such as alumina can be similarly reversed by the adsorption of colloidal silica. Tp achieve charge reversal, however, the colloid being ad sorbed must be present in excess and the partides to be coated must be added to the sol with violent agitation at the point of mixing. The colloidal partides otherwise will merely form bridges between thelarger partides and flocculation win occur. It is probably for this reason that the adsorption of a single layer of colloidal partides on the surfaces of larger partides of opposite charge has been difficult to observe. Specificapplications for this phenomenon will become appar ent from the changes observed in the behavior of the coated materials. Coating alumina partides with colloidal silica provides a negatively charged alumina slip which then is not thickened or flocculated when mixed with other negatively charged ceramic components such as clays. A silica slip ordinarily will thicken or flocculate when mixed with an acidic alumina slip, but pretreatment of the silica with colloidal alumina will pre vent this. Possibilities are thus suggested for the use of col loids in varying the degree of dispersion and thixotropic be havior of ceramic slips and plastic masses. * "Air-Spun'' graphite, Joseph Dixon Crucible Company, . Jersey City, N. J. Fig. 5. Graphite flake* with adsorbed eolloidoi alumina.