Jan., 1959
PREPARATIOX O F
FINEPARTICLES
’ A study of the data cited above reveals that, in addition t o r12,the rii appear t o show a linear concentration dependence in ideal liquid systems. For only two systems, however, are sufficient data presented by these workers t o calculate all three friction coefficients. These are the non-ideal benzene-alcohol systems. The possibility of isotope effects in the measurements was minimized by labeling the molecules with C14rather than deuterium. l 2 The accuracy of the diffusion data combined with available thermodynamic figures is nevertheless sufficient to warrant only qualitative discussion of the concentration dependence of the friction coefficients at 25” and 1 atm. in these systems. We have calculated these for the benzene-ethanol system only, for which the following observations are consistent with the accuracy of the results (taking benzene as component 1 and ethanol as 2). (1) The value of r12is lower than that of either rll or rZ2at all concentrations. It ( ~ ~ 2 )shows a negative deviation from linearity, passing through a minimum somewhere between 15 and 30 mole % ethanol.
(12) The agreement among the results reported by Johnson and Babb’o and those of Partington, Hudson and Bagnall [Nature 169, 583 (1952)],who used deuterium-labeled molecules, appears to refute the suggeation of B. Ottar [Acta Chem. Scand., 9, 344 (1955)l that deuterium-labeling gives rise to a strong isotope effect in the tracer diffusion of the lower aloohols.
FROM
BIMETALOXALSTES
83
(2) The value of rll,almost coilstant in the more concentrated benzene solutions, begins a t concentrations below 80 mole % t o show a gradually increasing rate of rise with decreasing concentration of benzene, but remains less than the value of r2p except, perhaps, at concentrations below 25 mole
%*
(3) The value of rZ2 shows a similar behavior, remaining nearly constant as pure ethanol is diluted ’ below which it rises with det o about 80 mole % creasing concentration a t a steadily increasing rate. The discussion immediately preceding summarizes the concentration dependence of molecular fraction coefficients in a single binary liquid system of non-electrolytes, A great deal more information of this type, as well as data on the effects of temperature and pressure, is needed before an understanding of the factors determining the values of these quantities can be developed. We nevertheless share with Onsager (ref. 2, p. 247) the feeling that the theory of transport processes may perhaps be most rapidly advanced by focusing our attention on the coefficients of the dissipation-function rather than on more commonly employed rate constants. A subsequent publication will deal with the extension of this approach to electrolytic systems. l 3 (13) R. W.Laity, J . Chem. Phys., in press.
PREPARATION OF BINE PARTICLES FROM BIMETAL OXALATES1 BY WILL~AM J. SCHUELE The Franklin Institute Laboratories, Philadelphia 3, Pennsylvania Received Julu 81, 1968
Ferromagnetic and ferrimagnetic fine particles were prepared by suitably heat-treating co-precipitated oxalates of Co-Fe, Ni-Fe and Cu-Fe systems in a wide range of compositions. The particles produced by this method included metals, alloys, ferrites and other ouides. The technique is an excellent method for producing fine particles with a wide range of magnetic properties. Ferrite particles were prepared in a manner much simpler than the standard techniques. The existence of permanent magnetic properties in fine particles of normltlly “soft” magnetic materials such as the ferrites was in agreement with predictions of the theory of single domain particles. While the experimental work was confined to magnetic materials, the technique also can be useful in the preparation of other fine particle materials.
Introduction The specific purpose of the research described in this paper was the development of a method whereby various ferromagnetic materials (metals, alloys, ferrites and oxides) could be prepared in fine particle form.2 It is well known that ferromagnetic materials in fine particle form exhibit “hard” or permanent magnetic properties, while in large particle form they exhibit “soft” or temporary magnetic properties. a Since fine particle materials sinter a t lower temperatures than larger particle materials, i t was obvious that we could not use a method of preparation which required high temperatures. The usual preparation of a ferrite involves a solid state reaction between the two
metal ~ x i d e s , ~but J since this required the diffusion of the cations over large distances (-lo4 A.) high temperatures are necessary. It seemed possible to eliminate this long range diffusion problem by starting with the two metals in a solid solution, in which case the ions would have to diffuse only the distance of the lattice spacings or a few Angstroms. Wickham6 and Robin’ indicated that some of the bimetal oxalates form solid solutions. Since these oxalates could be decomposed thermally a t moderate temperatures to yield the metal oxides and easily removed gaseous decomposition products, they were selected as the intermediates in our preparation of fine particle materials. We found, experimentally, that we could form solid
(1) This research was supported by the United States Air Force under Contract No. A F 33(616)-5041, monitored by Aeronautical Research Laboratory, Wright Air Development Center. (2) Detailed magnetic studies are being made on the produats reported here and will appear as aaeparate paper at a later date. (3) C. ICittcl, Phiis. Rei,., 70,966 (1946).
(4) J. L. Snoek. “Ferromagnetic Rfaterials,” Elsevier (Press. Houston, Texas, 1949). ( 5 ) E. W. Gorter, Compt. rend.. 230, 192 (1950). (6) D.G . W‘ickham, Tech. Rept. 89,Lab. for Insulation Researrli, MIT, Cambridge, Mass. (7)J. Robin, BuEZ. S O C . clrim., Prance, 1078 (1953).
WILLIAM J. SCHUELE
84
Vol.,63
solutions of ferrous oxalate with the oxalates of zinc, cobalt, nickel, magnesium and manganese. It is interesting to note that these ions have nearly the same ionic radii as the ferrous ion? Experimental
oxalates in the three systems decompose at some tempera: ture below this and such a temperature was low enough to prevent serious sintering of the products. Treatments A and B constituted reducing conditions, treatments C and D oxidizing conditions. A. Reduction in Hydrogen.-Fifteen grams of oxalate was placed in a 21 mm. diameter tube using glass wool plugs to confine the material to the center portion of the tube. Preparation and Characterization of the 0xalates.The sulfates of the two metals desired in the bimetal oxalate The tube then was inserted in a furnace a t 390’ and hydrogen were dissolved in water at 60’ in the proper ratio to give passed through it for one hour. This effected a reduction 1000 meq. in 1500 ml. of solution. The stoichiometric to the metals or alloy. X-Ray spectra for the products of the copper-iron series amount of ammonium oxalate was dissolved in 1500 ml. of water at 60’ and added with rapid stirring to the metal gave separate peaks for the two metals, the height of the ion solution. The precipitate was filtered in a buchner peaks being proportional to the amount of copper or iron funnel, washed with water and then washed with acetone present. No alloying of the two metals was detected. Alloys were obtained for the nickel iron system. A to speed up drying. Of the systems prepared, three were selected for more extensive study: cobalt ferrous oxalate, maximum in the saturation magnetization near 20% nickel nickel ferrous oxalate and copper ferrous oxalate. The found experimentally for the fine particle material agrees Goldschmidt ionic radii for cobalt (0.82) and nickel (0.78) with that reported in the literature for the bulk alloy.9 X-Ray spectra for the cobalt-iron system indicated the are similar to ferrous ion (0.83) while copper (0.70) is smaller. In each of these systems, oxalates were prepared production of owder alloys. The X-ray intensity of the in one mole quantities of the composition A,Fel-,CZO4, characteristic 8nes for the Co-Fe series are presented in where A = Cu, Ni or Co and y = 0.1, 0.2, 0.3, 0.4, 0.5, Fig. 1. In the region 0 to 70% cobalt, the X-ray intensity remained essentially constant for the body centered cubic 0.6, 0.7, 0.8, 0.9, 1.0. In order to determine whether our samples were solid peak. This confirmed the fact that iron and cobalt form a solutions or just a mixture of the two oxalates, X-ray single alloy phase within this composition range. A face spectra were taken for each sample. If the two metal centered cobalt phase appeared experimentally in the oxalates did not form solid solutions, the X-ray spectra region of 75% cobalt. This was in excellent agreement with would consist of the two separate spectra superimposed. the phase diagram for the bulk iron-cobalt alloy which also If the bimetal oxalate was formed, the characteristic peaks indicated that a face-centered cobalt phase should appear on the spectra would lie between those of the pure compo- in that region.’O The lattice spacings for the region in which nents. Data of the nickel ferrous oxalate system shown in the single phase iron-cobalt alloy exists were in o d agreeTable I will serve to illustrate this variation of d values ment with the values obtained by Ellis and breiner for with composition for a solid solution. A similar set was the bulk iron cobalt alloy. B. Reduction in CO COa.-Fifteen grams of oxalate obtained for cobalt ferrous oxalate. However, for the copper ferrous oxalate, solid solution was not obtained and was packed in a tube which was closed a t one end and conthe peaks of both copper and iron oxalates were present in tained a one hole stopper a t the other end. The tube was the spectra, the intensity of the respective peaks being inserted in a furnace a t 390’. As the oxalate decomposed, proportional to the concentration of the metal oxalate. the gases produced (CO and COZ) were passed through The X-ray data further supported the idea that similarity tubing into a wash bottle partially filled with water. The of ionic radii greatly aids in the formation of solid solutions. progress of the reaction was indicated by the bubbling in the wash bottle. When the decomposition reaction was over, the tube was sealed and permitted to cool. The product TABLE I was then protected from air by the addition of benzene. PRINCIPAL “d” VALUESOFOR NI-FE OXALATESYSTEM This treatment resulted in products consisting of several phases and is the least understood of the four treatments. NiCaOc 4.78 3.94 3.01 2.53 2.07 1.94 1.87 One would anticipate that the decomposition of the oxalate Nio.oFeo.tCaO4 4.78~ 3.93 3.01 2.53 2.07 1.94 1.87 would proceed according to the equations Nio.aFeo.nCnO4 4.80 3.92 3.02 2.55 2.08 1.95 1.87 Nio.rFeo.1CnO4 4.81 3.92 3.03 2.57 2.09 1.95 1.88 FeC204 -+ FeO COZ CO Nio.aFeo.& a 0 4 4.83 3.91 3.04 2.58 2.09 1.96 1.88 cuczo4 + CUO coz co Nio.rFeo.sCnO4 4 85 3.90 3.07 2.59 2.10 1.98 1.88 CUO co +c u coz Nio.deo.sCaO4 4.87 3.90 3.11 2.59 2.11 1.99 1.88 Nio.sFeo.iCzO4 4.89 3.89 3.12 2.60 2.11 2.00 1.88 X-Ray examination of the products obtained from the Nio.iFeo.aCaO4 4.89 3.89 3.14 2.61 2.12 2.01 1.88 copper ferrous oxalate system revealed copper and magneNio.iFea.oCaOc 4.90 3.88 3.15 2.62 2.12 2.02 1.89 tite lines. The high saturation magnetization values 2.02 1.89 FeC104 4.89 3.87 3.10 2.63 2.13 appear to require that about one fourth of the total iron be present as free iron and the remainder as magnetite. Simia Interplanar spacinIgs. larly, Franklin, et al.,l* showed that a mixture of magnetite The oxalates were crystalline precipitates whose particle and iron was obtained from the decomposition of ferrous size depended on the metal ions and the ratio of metals formate. They suggested the disproportionation reaction Fes04. This would correlate well with involved. The size was determined from measurements of 4Fe0 + Fe electron micrographs. I n the nickel ferrous oxalate system our data. However, in our samples careful scans of the there was a steady decrease in the average particle size as X-ray spectra at 0.04’/minute and the highest sensitivity, the nickel concentration increased, ranging from 6 p for could not verify the presence of free iron. Its presence 10% Ni to 1 p for 100% Ni. I n the cobalt ferrous oxalate in extremely small (100 A , ) particle size is suggested system, there was no gradual change as the composition as a possible explanation, since the X-ray lines would be varied. The partides were ellipsoidal, with an average too broad to be detected. This hypothesis was further major axis of 4-6 p , and an a.verage minor axis of 2-3 p . supported by the fact that the coercive forces of these In the copper ferrous oxalate system, two types of particles samples were found to be four to five times greater at were obtained, one spherical, the other a parallelopiped. -196’ than they were at room temperature, which is The spherical particles were found to be the copper oxalate consistent with the magnetic behavior of very small parand measured about 1 p in diameter. The ferrous oxalate ticles .I8 particles were about 3 p in width and depth, and up t o 10 p (9) R. M. Bozorth, “Ferromagnetism,” D. Van Nostrand Co.. in length. Treatment of the Bimetal Oxalates.-Each of the three New York, N. Y.1951.p. 247. (10) “Metala Handbook,” American Society for Metals, 1948, p. oxalate systems was given four treatments: (A) heated in a stream of h drogen a t 390’; B)heated in its own decom- 1192. (11) W.D. Ellis and E. 8. Greiner, Trans. Ant. Soo. Metols, 29,415 position pro&cts (CO and Cb,) at 390’; (C) l p t e d in (1941). a stream of air a t 390’; ( D ) heated in air a t 1100 (12) A. D.Franklin, L. Muldawer and P. J. Flandera,THISJOURNAL, The temperature of 390” was chosen because all the
+
+
,
+ +
+
+ +
+
.
SB, 340 (1955).
(8) L. Pa.iling, “The Nature of the Chemical Bond,” 2nd Ed., Cornell University Press, Ithaca, N. Y., 1940, p . 343-350.
(13) L. Meel, Compl. rend. S28, 664 (1949); Ann. Geoghys., 6, 99 (1949); Reu. M o d . Phys., 26, 293 (1953).
Jan., 1959
PREPAR.4TION OF
FINEPARTICLES
FROM
85
BIMETAL OXALATES
TABLE I1 EXPERIMENTAL AND CALCULATED MOMENTS FOR CUFE A N D NIFE OXALATE SYSTEMS HEATEDIN AIR Oxalate system
Moment, calcd.
9.0 Cuo.lFe0.sczo4 CUO.ZF~O.~CZO~ 18.0 Cuo.~Feo.GOl 27.0 ~ U o . ~ ~ e o . ~ ~ z 0 4 27.0 Cuo.aFe0.&~04 22.5 18.0 Cuo.sFe0.4C~O4 13.5 Cuo.deo.rCz04 Cuo.sFeo.tCzO4 9.0
Heated at llOOo, Heated at 390°, obsd. obsd.
9.3 19.5 28.8 25.8 23.5 17.1 9.7 8.1
Oxalate system
Nio.lFeo.G O , Nio.eFeo.aCzO4 Nio.sFe0.7CzO4 Nio.rFeo.sCzO4 Nio. ~Feo. sCZO~ Nit 6Fe0.& 2 0 4 Nio.beo.aCzO4 Nio.Peo.zCz04
40.0 36.8 29.0 16.0 26.5 19.1 13.2 3.6
The interpretation of the nickel-ferrous system is also complicated by the presence of several phases. Additional studies will be necessary to elucidate the composition of these products. X-Ray spectra obtained from the cobalt-ferrous system indicate the presence of magnetite in the high iron end of the system and COO and Co in the high cobalt end of the system. The presence of cobalt indicates the reduction of COOby carbon monoxide. C. Oxidation in Air at 390'.-Fifteen grams of the oxalate was placed in a tube and held in position with glass wool plugs. The tube was inserted in a furnace at 390' and air passed through the tube for one hour a t this temperature. The products obtained indicated clearly that when the starting material was not a solid solution this temperature was too low to permit the ions to diffuse distances of the particle diameters ( p ) . Table I1 presents the calculated and observed moments per gram for the copper-ferrous system at 390 and l l O O o and the nickel-ferrous system at 390'. The calculated values for copper are based on the production of stoichiometric cop er ferrite with the excess metal present as CuO or aFeZOa,&pending on which ion was present in excess of the ferrite composition. The nickel ferrous oxalate system, which was a solid solution, produced the ferrites a t 390". The calculated moments for the nickel system were based on the production of nickel ferrite with the excess metal present as NiO or as the magnetic iron oxide y-FetOa. No consideration was made for the solubility of any phase in the others, but the agreement of experimental values with the calculation on our basis would suggest that such considerations are not required. At present it is not clearly understood why the y-FeaOs is stable in the Ni-Fe system and not in the Cu-Fe and Co-Fe system. There are some indications, however, that certain ions do increase the stability of y-FezOa.14 A possible explanation for the deviation which occurred at the high nickel end is proposed: in the nickel ferrite, the nickel ions tend to occupy the B sites of the spinel structure, while the iron occupies both the B and the A sites." When the number of nickel ions is low, there is a good probability that they occupy only B sites but as the concentration of nickel ions increases the probability of all nickel ions going into the B sites is decreased. The presence of nickel in the A sites increases the moment. An alternate explanation is that in the initial decomposition of the oxalate some of the material was reduced to metallic nickel, and r e oxidation was not complete at this relatively low temperature. The cobalt-iron system also produced the ferrites a t 390". X-Ray s ectra for this system were made and the intensity of the ciaracteristic lines vs. composition are plotted in Fig. 2. This plot would also suggest that there is very little mutual solubility of the various phases. D. Oxidation at llOOO.-The samples were treated as in treatment C. They were then placed in Vycor crucibles and heated in a Lindberg furnace a t 1100" for one hour. This treatment produced ferrites in all three systems. This was as expected, since the temperature was high enough to permit diffusion of the ions between particlea, even in the Cu-Fe system where no solid solution had been formed. Unfortunately, there has also been considerable sintering of the particles resulting in the loss of the fine particle nature of the product. I n all cases, at least two phases (14) P. W. Selwood, "Magnetoohemistry," 2nd Ed., Interscience Publiihers, New York, N . Y . .p. 807. (15):L. Neel, Ann. Phus. Paris, 3, 137-9 (1948).
0
10
20
Moment, calcd.
Heated at 390' obsd.
67.7 59.1 50.5 43.1 36.1 29.2 21.9 14.7
30 40 50 60 Atomic % ' cobalt. Fig. 1.
67.3 60.8 49.0 42.4 34.9 28.5 22.3 17.2
70
80 90 100
70 20 _ COMPOSTION _
60
50 rcI
0
53 7
D
50
a v
ini 611 4
I
co FO ,. . rn FaZO, CO~CO~O,I
coo
40
.e8 O
30
Y
2 20 10
0 0 10 20 30 40 50 60 70 80 90100 Co 100 90 80 70 60 50 40 30 20 10 OFe Atomic per cent. Fig. 2.
0
10
20
30 40 50 60 Atomic yo cobalt. Fig. 3.
70
80
90
100
86
PETERJ. DUNLOP -4ND LOUISJ. GOSTING
were produced, the ferrite and the oxide or oxides of the metal ion present in excess. Figure 3 is a plot of the X-ray intensities of characteristic lines vs. composition for the Co-Fe system heated a t 1100”. Note that two curves are given for the ferrite and that these are in good agreement with each other.
Conclusion Metals, alloys, ferrites, metals in metal oxides and ferrites in metal oxides have been prepared as fine particles by the solid solution oxalate technique. By virtue of their particle size, it was possible to produce the ferrites as magnetically hard substances. Excellent control of the composition of the oxalate was possible because the metal ions were mixed in solution before precipitation. This method offers the possibility of “doping” a sample in the initial precipitation with very small amounts of a foreign metal ion. The low solubility of the oxalates involved eliminated the necessity of making solubility corrections.
Vol. 63
General metallurgical techniques for the preparation of alloys in bulk form are well established but not applicable to the preparation of fine particle materials. Some of the techniques useful for powder metallurgy purposes produce particles several microns in diameter. The technique presented in this work permits the production of fine particles down to single domain sizes and should have general application in the production of fine particles of any alloy whose metals have atomic radii of the proper size to form the bimetal oxalate. Because of their unusually small particle size, the alloys prepared by the above method should be of great interest in nuclear and electron magnetic spin resonance measurements as well as the ferromagnetic studies for which they were intended. These alloys also would be useful in the study of the penetration of radio frequency fields into fine particles to determine electron mean free paths. -
USE OF DIFFUSION AND THERMODYNAMIC DATA TO TEST THE ONSAGER RECIPROCAL RELATION FOR ISOTHERMAL DIFFUSION IN THE SYSTEM NaCl-KCl-H20 AT 25’ BY PETERJ. DUNLOP AND LOUISJ. GOSTING Contribution from the University of Wisconsin, Madison, Wisconsin, and the Institute of Physical Chemistry, Uppsala, Sweden Received August 4,1968 A test of the Onsager reciprocal relation has been made for four compositions of the system NaC1-KCl-HZO, using existing data for the diffusion and activity coefficients. The Onsager relation is satisfied within 5% at all four compositions. Significant errors which affect these tests have been considered in detail; for each composition the observed deviation is about half the maximum error which is estimated from uncertainties in the experimental data.
Just as classical thermodynamics provides equati6ns relating certaiu measurable properties of a system at equilibrium, the thermodynamics of irreversible processes provides equations relating measurable properties of a system which is somewhat removed from equilibrium. Of basic importance to the theory of irreversible processes are the reciprocal relations derived by Onsagerl -4 from the assumption of microscopic reversibility. The Onsager reciprocal relations and their connection with the theory of irreversible processes have been discussed in several monographs (see for example, ref. 5-8) which also list original articles in this field. For isothermal diffusion in a ternary system there is a single reciprocal relation which provides one restriction on the values of the four diffusion coefficients that can be measured experimentally for (1) L. Onsager, Phya. Rev., 37, 405 (1931). (2) L. Onsager, ibid., 88, 2265 (1931). JOURNAL,36, 2689 (1932). (3) L. Onsager and R. M. Fuoss, THIE (4) L. Onsager. Ann. N . Y . Acad. Sci., 46, 241 (1945). (5) 8. R. de Groot, “Thermodynamics of Irreversible Processes,” Interscience Publishers, Inc., New York, N. Y., 1951. (6) K. G. Denbigh, “The Thermodynamics of the Steady State,” John Wiley and Sons, Inc., New York, N . Y.,1951. (7) J. 0.Birschfelder, C. F. Curtias and R. B. Bird, “Molecular Theory of Gases and Liquids,” John Wiley and Sons, Inc., New York, N. Y.,1954. (8) I. Prigogine, “Introduction to Thermodynamics of Irreversible Processes,” Charles C Thonias, Springfield, Illinois, 1955.
any composition of the system; thermodynamic quantities also appear in this equation. Here we attempt t o test this Onsager relation by using data for diffusion a t four compositions of the system NaC1-KC1-H20 a t 25”. The diffusion coefficients used were measured with the Gouy diffusiometer by O’Donnell and Gostings and the activity coefficients employed were derived from isopiestic data reported by Robinsonlo and by Robinson and Stokes.’l Although the four diffusion coefficients for a few other ternary systems have been measured a t one or two compositions12-16 we will not consider them here. Activity coefficients have been measured only for those systems containing two electrolytes, and of these only the system NaClKCl-H20 obeys Harned’s r ~ l e ’ ~ -(PI ’ ~ = PZ = (9) I. J. O‘DonnelI and L. J. Gosting, a paper presented in a Symposium at the 1957 meeting of the Electrochemical Society in Waahington, D. C.; the Symposium papers are to be published as a monograph by John Wiley and Sons, New York, N. Y. (10) R. A. Robinson, in “Electrochemical Constants.” National Bureau of Standards Circular 524, 1953. (11) Appendix 8.3 of ref. 18. (12) P. J. Dunlop and L. J. Gosting, J . Am. Chem. Soc., 77, 5238 (1955). (13) H.Fujita and L. J. Gosting. ibid., 78, 1099 (1956). 61, 994 (1957). (14) P. J. Dunlop, THISJOURNAL, (15) P. J. Dunlop, ibid., 61, 1619 (1957). (16) F. E. Weir and M. Dole, J . Am. Chem. Soc., 80, 302 (1958). (17) H.S. Harned and B. B. Owen, “The Physical Chemistry of Electrolytic Solutions,“ Third Edition, Reinhold Publ. Corp., New Yorli, N. Y , , 1958.
I