Jan., 1955
Low TEMPERATURE REDUCTION OF IRON OXIDES
I n this case a 4-10 potential law results. This law has been introduced into the expression for V , and the apparent area calculated.la An area for the saran of 700 m.2/g. results. C r ~ w e l lhas ~ ~ suggested that his calculations can be extended to our case and it will be of interest to see the area that his exact summation will yield. I n the absence of exact information concerning the structure of the charcoal, it is of interest to compare the present measurements with our earlier res u l t ~on~ a saran charcoal of lower area. There the B.E.T. and our areas were in substantial agreement with a value of 800 mS2/g. (13) W.A. Steele, Thesis, University of Washington, 1954. (14) A. D.Crowell, private communication.
65
The higher surface material is prepared by “burning out” the lower surface material. If such a procedure increases the capillary size, and a t the same time destroys some of the capillaries, the various areas can be reconciled. If the nitrogen a t “point B” in the B.E.T. measurements fills the pores rather than just forming the monolayer, it is clear that the volume so accommodated (and thus the B.E.T. area) will rise as the pores get larger. However, the area measured by our method is not influenced by condensation of gas, and so does not rise to the same extent. I t would appear then that, on the basis of the arguments presented here, the apparent B.E.T. area of the higher surface area saran is unreasonably large.
LOW TEMPERATURE REDUCTION OF IRON OXIDES1 BY A. D. FRANKLIN AND R. B. CAMPBELL Franklin Institute Laboratories for Research and Development, Philadelphia, Pa. Received July 90, 106.4
Comparison of the diameters of iron oxide crystallites before reduction with those of the iron particles produced at temperatures less than about 200” indicates that each oxide cryatallite produces one iron particle. At higher temperatures, sintering occurs. These results suggest that for low temperature reduction, nucleation of the new phase is much slower than growth.
Introduction The recent development of single-domain ferromagnetic particle^^-^ has focussed attention upon the low temperature reduction of metal oxides to produce metallic particles. In the literature on this type of reaction7-12 there appears to be little discussion of the essential factors controlling the particle size of the final metal particles. I t is well known that finer particles are produced a t lower reduction temperatures, but the question of the relation between the geometric properties of the initial oxide and of the final metal remains unexplored. This paper presents the results of a study of the particle and crystallite sizes of some iron oxides arid of the iron powder produced by reduction of these oxides a t temperatures between 126 and 450”. Experimental Ferrous formate was prepared by dissolving General Aniline and Film Corporation Type “L” carbonyl iron in Merck Reagent formic acid. Excess formic acid waR added to bring the pH of the solution t o 2-3 to minimize oxidation of the ferrous ion, and the ferrous formate precipitated by evacuation of the Polution. The hydrated ferrous formate crystals possessed diameters of the order of several microns. This ferrous formate was dehydrated below 150°, and then (1’ Supported by the OEce of Naval Research. (2) L. Neel, Conapt. rend., 284, 1488 (1947). (3) C. Kittel, Phya. Rev., 70, 965 (1946). ( 4 ) F. Bertaut, Conpl. rend., 229, 417 (1942); Thesis, University of Grenoble (1953). ( 5 ) C. Guillaud, Thesis, University of Strasbourg (1943). (6) W. C. Elniore, Phys. Rev., 64, 1092 (1938). 1 7 ) F. Lihl, Acto Phgs. Austriaca, 4 , 360 (1951). ( 8 ) F. Lilil, Metall., 6, 183 (1951). (9) N. I. Ananthanarayanan and J. F. Libsch, J . Metals, 6, Trans., 79 (1953). (10) J. Robin and J. Benard, Cornpt. rend., 232, 1830 (1951). (11) V. A. Roiter, V. A. Yusa and A. N. Kuznestsov, Zhur. Pia. K h i n . , 26, 960 (1951). (12) F. Olmer, Rev. Met.. 88, 129 (1941).
thermally decomposed at temperatures from 220 to 255”
in vacuo.
Chemical analysis of the resulting oxides showed them to be principally FeO, containingless than2yo carbon by weight, and small, variable amounts of ferric ion. Their X-ray diffraction patterns indicated a spinel structure, in agreement with Lihl’s** results on similar material. However, other evidence to be reported in a subsequent paper indicated that this FeO was really an intimate mixture of Fe and Fes04. The distribution and particle size of the Fe was such that the corresponding X-ray diffraction lines were scarcely observable in the pattern of the mixture. When data was taken from the spinel part of this pattern, it applied only to the FesOc phase. It seems reasonable also to suppose that the Fe phase initially present made only a negligible contribution to the diffraction pattern observed after reduction. Three other iron oxide specimens were used. Fer04 and ?-Fezor powders of the type used for magnetic recording tapes were included. The a-FepOl specimen was Baker C.P. ferric oxide. Reduction of the oxides a t temperatures as low as 125” was accomplished by mixing with excess CaHp powder (Met,al Hydrides, Inc.) and heating at the desired temperature in a low (2-5 cm.) pressure of hydrogen. Reduction with CaHa has been reported14 to involve a two-step reaction 2Fe0 2€Iz 2Fe 2Hz0 CaHz 2Hp0 Ca(OH)2 2H2 (1) Ca(0H)z CaHa 2Ca0 2Hz (2) Preliminary experiments showed that under these conditions reaction 2 proceeded to completion. For each FeO molecule reduced, one excess molecule of Hz wfts produced. Since the reaction velocity increased strongly with increasing Ha pressure, the excess Ha was removed from the reaction through a by-pass manometer, which allowed the hydrogen pressure in the reaction chamber to remain constant. The excess Ha was pumped into a system of known volume, and the system pressure followed as a measure of the extent of reduction. The extent of reduction of the final product was also determined inde endently by dissolving a small sample in dilute HzS04 an$ collecting the Hz produced over mercury as a measure of the free metal present. The two methods showed reasonably good agreement.
+
+
+
Jr
+
+ +
(13) F. Lihl, Monalsh., 81, 632 (1950). (14) H . Flood, Kul. Noreks Videnskab. Selakab. Forh., 1, 66 (1936).
A. D. FRANKLIN AND R. B. CAMPBELL
VOl. 59
of the ferrous formate to the oxide, and dl the oxide crystallite diameter. Tz is the temperature a t which the oxide was reduced to iron, and dz the iron crystallite (and particle) diameter. The column headed % Free Iron gives the degree of reduction, determined by hydrogen evolution. TABLE I REDUCTION OF FeO PARTICLES
100
200
300
REDUCTION TEMPERATURE
('a,
Oxide sample
"C.
di>6.
'C.
da, 1.
FeO-1
255
350
400
536 382 275 250 250 300 255 260 230 195 174 165 170 440 265 145 130 95
Ti
Fe0-2
235
210
Fe0-3
220
100
400
Fig. 1.-Effect of reduction temperature on diameter of iron particles. Particle sizes were determined with the electron microscope. Crystallite sizes were determined from the broadening of X-ray diffraction lines, using a G. E. XRD-3 spectrometer and Fe K, radiation. The observed broadening was ~ using Jones16 corrected for the effect of the ~ I - C Ydoublet method, and for instrumental broadening using WarrenW approximations. The instrumental broadening was determined at several values of 28 from an annealed Armco iron specimen and from annealed zinc and tin. Crystallite diameters were determined using the (110) line for iron, the (311) line for the spinel oxides, and the (211) line for aFez03. The particles in some of the oxides appeared to be polycrystalline. The iron particles were single-crystal, as indicated by rough agreement between electron microscope and X-ray sizes. Estimation of the crystallite diameter by the method used here involves considerable uncertainty. Aside from measurement errors, the chief source of uncertainty lies in the choice of the constant K in the Scherrer equation'? used to relate line breadth and crystallite diameter d = KA/p COS 0 where d is the crystallite diameter, A is the X-ray wave length, p is part of the breadth of the diffraction line due to crystallite size, and 0 is the Bragg angle. The value of K depends upon the crystallite shape and structure. Since there is a change in structure upon reducing the oxide to iron (especially in the case of &Fez03,where a drastic change in symmetry from rhombohedral to body-centered cubic occurs), the appropriate value may be different before and after reduction. Even for relatively simple shapes and structures, however, the value of K is in doubt by about 2O%.lS In this work, a value of 0.94 was chosen for all materials. This uncertainty in the crystallite diameters must be considered in the interpretation of the data.
Results Table I lists all of the data for the FeO reduction. I n this table TIis the temperature of decomposition (15) F. W. Jones, Proc. R o y . Hoc. (London), 166A, 16 (1938). (16) B. E. Warren, J . A p p l . P h y s . , 12, 375 (1941). (17) See for instance R. W. James, "The Optical Principles of the Diffraction of X-rays," G Bell and Sons, London, 1948, p. 536. (18) L , Alexander and H. P. Klug, J. S p p l . P l r ~ a .21, . 137 (1950).
Ta,
300 220 205 150 350 f 10 285 285 260 220 200 185 150 450 300 200 150 125
Free iron,
%
95 75 82 92 91 80 70 68 72 80 60 68 85 72 86 62 70 68
Table 11, with the same notation, gives the data for the lowest temperatures of reduction for the FeO, and for three other oxides reduced a t temperatures sufficiently low to avoid sintering. TABLE I1 OBSERVEDAND COMPARISONOF CALCULATED LOWER LIMITING Fe PARTICLE DIAMETERS Ta,
Oxide sample
di, A.
'C.
FeO-1 Fe0-2 Fe0-3 Fer04 7-Fe20a a-FetOa
350 210 100 280 356 430
200 200 125 225 210 210
da, A. (Obsd.) (Calcd.)
250 170 95 200 275 330
273 164 78 221 264 331
Oxide density, g./cm.a
4.8 4.8 4.8 5.2 4.6 5.2
The calculated values for dz, the iron particle diameter given in the fourth column, were obtained from the oxide crystallite diameters, assuming that each oxide crystallite produces one iron particle without change of shape. The oxide densities listed in the fifth column were used in this calculation, together with a value of 7.9 g./cme3for metallic iron. Discussion In Fig. 1, the iron particle diameters listed in Table I are plotted for the three sizes of FeO crystallites as a function of reduction temperature. Each curve terminates a t the lower left in a horizontal dotted line. This line represents the iron particle diameter expected if each oxide crystallite produced one iron particle. For low reduction temperatures, the observed iron particle diameters agree with these expected diameters. This result is confirmed by the data in Table 11. The correspondence between oxide and iron diameters is repeat,ed for the FeO, i-uid is also shown for some
Jan., 1955
ADSORPTION OF WATERVAPORON GERMANIUM AND GERMANIUM DIOXIDE
67
For higher reduction temperatures, an abrupt Fe304, y-Fez03, and cr-FezOa specimens. For the latter three, the oxide particles were polycrystalline. change in slope of the curves in Fig. 1 occurs. I n Each particle contained several thousand crystal- the high temperature region, some form of sintering lites. The iron particle diameter is clearly deter- apparently takes place. This sintering occurs a t mined by the crystallite rather than particle di- surprisingly low temperatures, boeginning in the ameter of the oxide. neighborhood of 100" for the 100 A. diameter FeO. I n terms of nucleation-and-growth, it appears This result is in conformity with observations on that only one nucleus is formed per oxide crystal. thin films. Wheeler20 states that evaporated iron This in turn implies that nucleation may be slow, and nickel films sinter a t temperatures as low as and that once a nucleus is formed, it grows rapidly 25". until it has consumed the entire oxide crystallite. Conclusions The role of nucleation as the slow process is borne out by the kinetics of the reduction. T a t i e v ~ k a y a ' ~ The following conclusions are drawn from this and co-workers have shown that in the temperature work: 1. For reduction temperatures from 125 to region from 350 to 600" the reduction of a-FeO, 450", iron oxides are reduced in such a fashion that FesOl and a-Fez03is autocatalytic. According to each oxide crystallite produces one single-crystal their data, and also to observations made during iron particle. 2. Sintering of iron particles smaller this work, the reaction velocity passes through a than 300 8. in diameter begins a t temperatures maximum as reduction proceeds. In the nucleation- as low as 200". 3. I n the reduction process, nuclegrowth picture, the slow initial portion corresponds ation of the new phase is much slower than growth. to the period of nuclei formation, while the rapid This view also explains the general form of the reaction a t maximum velocity corresponds to the kinetics of reduction. particle growth. (20) A. Wheeler, in the chapter "Chemisorption on Solid Surfaces," (19) E. P. Tatievskaya, G . I. Chufarov and V. I
