THE UNIVERSITY OF MICHIGAN INDUSTRY PROGRAM OF THE COLLEGE OF ENGINEERING AN ANOMALOUS MAGNETIC TRANSITION IN POWDERED MAGNETITE Dale C. flay January, 1965 IP-692

TABLE OF CONTENTS Page LIST OF FIGURES................................................ iii INTRODUCTION................................................... 1 EXPERIMENTAL.................................................... 2 Preparation of Samples.................................... 2 Magnetic Moment Experiment................................ 3 Screening Experiment.............................. 3 RESULTS AND DISCUSSION........................................ 5 CONCLUSIONS.................................................... 18 FOOTNOTES................................................... 19 ii

LIST OF FIGURES Figure Page 1 Transition Temperature (temperature at peak of transition on DTA warming cycle) versus Mole Percent of Magnetite............................... 7 2 Counter-Force Current versus Temperature for Two Samples of Westrum Magnetite Cooled through Transition without Applied Field and Then Warmed in a Field of 1.04 x 105 A/M.......................... 8 3 Counter-Force Current versus Temperature for Westrum Magnetite Cooled and Warmed in Fields as Indicated....................................... 9 4 AI/mo versus Temperature for Minerville Magnetite..... 10 5 AI/mo versus Temperature for Partially Oxidized Westrum Magnetite..................................... 11 6 AI/mo versus Temperature for S32-80................... 12 7 AI/mo versus Temperature for S35-50............... 13 8 AI/mo versus Temperature for S32-66c.................. 13 9 AI/mo versus Temperature for S32-66b.................. 14 10 AI/mo versus Temperature for S35-64.................. 14 11 Differential-Thermal-Analysis Records Showing Deflection as a Function of Temperature on the Warming Cycle....................................... 15 iii

AN ANOMALOUS MAGNETIC TRANSITION IN POWDERED MAGNETITE INTRODUCTION The presence of the order-disorder low temperature transition for magnetite has been known for a long timei(l) The short-range order for the cubic phase gives way to long-range order in the orthorhombic phase at approximately -153~C for very pure samples of magnetite and at lower temperatures for less pure samples.(2) Verwey et al.(3,4) investigated the feasibility of the control of the order temperature. This idea was suggested by the variance in the recorded ordering temperature for natural magnetites. Controlled oxidation was tried with synthetic magnetites used for thermistors and it was found that both a shift in the ordering temperature and electrical conductivity resulted from the partial oxidation. The theory that electrical conduction resulted from an electronic exchange in the B sublattice between the ferrous and ferric ions is now widely accepted. The short-range order which exists between the order-disorder transition and the Curie point is consistent with magnetite's semiconductor behavior in this range and the long-range order does yield the insulator behavior. The study which will be described started with the assumption that oxidation does alter the characteristics of magnetite but that the oxidation will continue with time and will eventually cause the disappearance of the transition, To examine quasi oxidized magnetite and also to stabilize the'oxidation process, very small percentages of lithium were added to the magnetite system and what resulted is unusual. -1

EXPERIMENTAL Preparation of Samples The samples of lithium doped magnetite were powdered polycrystals prepared in the following fashion: (1) Ferric oxide and varying amounts of lithium carbonate were dry-mixed with mortar and pestle until visually homogeneous. (Samples S35-55, -56a, and -56b, however, were prepared from magnetite and lithium ferrite of varying amounts.) (2) The mixed powder was pressed into slugs at some fixed pressure with standardized dimensions. (3) The slugs were then fired at a fixed rate and held at 1450~C for a length of time which insured complete reaction. (4) The down-leg of the firing process was completed in a reducing atmosphere of dry nitrogen gas. (5) The final step in the material preparation was to grind up the slugs and pass the grains through a 100 mesh screen. Besides the doped samples tests were also run on two natural magnetites (one with an extremely high purity and the other very low purity) and on three synthetic magnetites (one with an extremely high purity prepared elsewhere(5) and two prepared in our laboratory). -2

-3Magnetic Moment Experiment The first test run on the different samples was that of finding the magnetic moment per unit density. The powdered samples were weighed out into 4 to 6 milligram samples, placed in glass capillary tubes, and sealed to insure fixed length measurements. The complete description of the moment measuring apparatus and the following screening experiment can be found elsewhere.(6) The conditions imposed on the samples during the testing were essentially the following: (1) The samples were cooled through the transition temperature region in the absence of a magnetic field (except where noted). (2) The samples were held in the sub-transition temperature region for a period of about one-half hour without a magnetic field and about one-half hour with the desired field applied. (3) The samples were then allowed to warm up through the transition region at a rate of approximately 1~C per five minutes. Force measurements were made during this phase on the sample using the balance and techniques described in the cited report.(6) The test results were reproducible as seen in Figure 2. Screening Experiment The differential-thermal-analysis experiment(7) was adapted to low temperature experimentation to eliminate samples which possessed no apparent order-disorder transition. This experiment is also

-4described elsewhere(6) and will not be detailed here. It should be pointed out that the samples could be prepared and run through this simple test in about one-half hour. The detail thermal fluctuations seen in the magnetic moment test cannot be resolved in the crude version of the DTA test which was run. The samples which exhibited no DTA transition showed no magnetic moment transition. Vestigial transitions were observed for some samples on the DTA which did not show up in the moment experiment.

RESULTS AND DISCUSSION The following table lists the materials that were studied, the nominal mole percents for the magnetite content, the saturation magnetization per unit density at room temperature and the transition temperature measured at the peak of the DTA transition reaction on the warming leg of the test. All the "S" samples were prepared in our laboratory. The high purity synthetic sample is called Westrum(8) and the natural magnetites are Minerville and Kiruna. SATURATION MOMENT AND DTA TRANSITION TEMPERATURE FOR MATERIALS UNDER STUDY Percent Temperature at Material Fe304 a Peak (DTA), ~C Minerville 100(?) 87.0 -153 Westrum 100 90.4 -156 S32-80 100 88.3 -154 S35-50 100 73.3 -157 S32-66c 99.5 78.5 -158 S32-66b 99.0 67.3 -162 S35-64 98.5 50.8 -163 S35-56b 98.5 70.4 -157 S32-66a 98.0 65.8 -164 S35-63 97.0 66.4 -167 S35-56a 97.0 77.5 -163 S35-62 96.0 73.5 -176 S35-55 96.0 80.6 -167 S32-65c 95.0 78.9 No transition S32-65b 90.0 72.5 No transition S32-65aII 50.0 76.8 No transition Kiruna 100(?) 94.4 No transition Note that samples S35-55, 56a and 56b were made from the ferrites rather than from the oxide and carbonate. Minerville and Kiruna are both samples of natural magnetite. -5

-6In Figure 1 is plotted the DTA transition temperature against the nominal mole percent of magnetite. The top curve with triangular indicators arose from making the samples directly from the ferrites. The three samples indicated were all processed at the same time and indicate a loss of lithium due to its volatility(9) in the form used. The circle and x indicators mark the.curve for the materials made in the conventional fashion and fired at the same time. The pluses mark the relative locations of the nominally pure samples prepared on an individual sample basis. In Figures 2 and 3 we have the reduced field magnetization curves for pure magnetite as usually measured(lO) but measured in the system just described and using the testing procedures outlined. These curves are partially described in an earlier paper.( ) The counter-force current, AI, is the balance current required to null the balance against the force on the magnetic sample in the nonlinear magnetic field. The counter force current alone or divided by the mass of the sample will be proportional to the magnetic moment of the sample. The curves shown in Figures 4 through 9 are not like those in Figures 2 and 3 in general shape. The normal behavior for magnetite cooled through the transition without a magnetic field applied and warmed in a moderate(ll) magnetic field is to start with a low magnetization below the transition and reach the saturation magnetization at the transition point. This is shown in Figure 2. The samples shown in Figures 4 through 9 start at approximately the projected saturation value for magnetite in the low temperature range and then exhibit a decreasing magnetization just before the transition occurs. In Curve B in Figure 4 it will be noted that a magnetic history alters this abnormal low temperature behavior,

-7-150 MINERVILLE -155 + S32- 80 + WESTRUM o S35-50 z g -160- MADE WITH S FERRITES -165 0. -170- MADE WITH LITHIUM CARBONATE Z AND FERRIC OXIDE 0 C)' -175 i-, I I I I 95 96 97 98 99 100 MOLE PERCENT OF MAGNETITE(Fe304) Figure 1. Transition Temperature (temperature at peak of transition on DTA warming cycle) versus Mole Percent of Magnetite,

'0 z 1.9 1.8 z W 1.5 TEPERTURE IN DEGREES CENTIGADE Applied Field and Then Warmed in a Field of 1.04 x 105 A/m. Data obtained on materials ust after acquisition. A/m. Data obtained on materials Just after acquisition.

-90 Z 2.9 J'< 0 2.8 2.8 ~ FIELD OF 1.52 x 10 A/m w 2.7 z 0 2.0 z o 1.9 0 1.8 12. ar a 1.7 - FIELD OF 1.04 x 10 A/m z Wcr1.2 C.) ~a a WI. LL' 0.9 ~ 0-9 - 7P1 CcC =^ FIELD OF 0.64 A /m5 ^^ o -180 -170 -160 -150 -140 -130 -120 TEMPERATURE IN DEGREES CENTIGRADE Figure 3. Counter-Force Current versus Temperature for Westrum Magnetite Cooled and Warmed in Fields as Indicated. Data obtained on material just after acquisition.

-10- _so Z 0.50 w 0.49 0 o 0.48 A w 0.47 - z 0.460 0.45 z o 0.44 - Ir~ B ~ 0.43 0 a.I I I I I I I I ___ o - 170 -160 -150 -140 - 130 -120 E TEMPERATURE IN DEGREES CENTIGRADE <1 Figure 4, AI/mo versus Temperature for Minerville Magnetite. Curve A - freshly ground natural magnetite, Curve B - same sample as for curve A but with magnetic history. Curve C - same sample as curve B after randomization by heating. Sample in all cases cooled through the transition region in the absence of a magnetic field and measurements taken at 1.38 x o05 A/m. AI is the magnetic balance counter force current and mo is the mass of the sample.

Z 5 u 0.46 1.42x105 A/m: 0.40 ( 0.45 I — z 0.34 074 1.05x10 A/m 2 032 0 0.33 < 0.32 0.3 0.32 i. -170 -160 -150 -140 -130 -120 *Mo~o ~TEMPERATURE IN DEGREES CENTIGRADE E ~ ~Figure 5. AL/mo versus Temperature for Partially Oxidized Westzrm Magnetite. Both curves the result of cooling through the transition in the absence of an external field and then warming in fields as indicated.

- z 2 0 _ 0.46 - o 4 o~,o 0 0.10 E- 000~-6 10Oo LU,0.452' 0.44 3j 0.43z z 0.42 0 0 0.41 0t E -170 -160 -150 -140 -130 TEMPERATURE IN DEGREES CENTIGRADE Figure 6. AI/mo versus Temperature for S32-80. Sample cooled in the absence of an external field and then warmed in a field of 1.38 x 105 A/m.

-13Iz W 0.39 o o o ~D \ —.-_... 0. 0.38 z- 0.34 0o c -170 -160 -150 -140 -130 -120? TEMPERATURE IN DEGREES CENTIGRADE 1-z Figure 7. AI/mo versus Temperature for S35-50. Sample cooled in the absence of an external field and then warmed in a field of 1.38 x 105 A/m. z LJ 0 0.40 - 0.393 0 0fi- - 170 - 160 - 150 - 140 -130 -z 0 TEMPERATURE IN DEGREES CENTIGRADE <1 Figure 8. A I/mo versus Temperature for S32-66c. Sample cooled in the absence of an external field and then warmed in a field of 1.38 x 105 A/m.

-14z 0 I< 0.35 Z 0.34 o_ 0 -170 -160 -150 -140 -130 aC..L.TEMPERATURE IN DEGREES CENTIGRADE 0 E Figure 9. AI/mO versus Temperature for S32-66b. Sample cooled in the absence of an external field and then warmed in a field of 1.38 x 105 A/m. z 0 -170 - 160 - w 01 O.3 -170 -160 -150 -140 -130 -JM < 0 0 0.29 Figure 10. LI/mo versus Temperature for S35-64. Sample cooled in the absence of an external field 0and then aed in a field of 1.38 x 105 A -1 a. TEMPERATURE IN DEGREES CENTIGRADE Figure 10. AI/mo versus Temperature for S35-64, Sample a-d then warmed i a field of 1,38 x 105 A/m,

<0 10 S 32-80 CY -(D 10o CM 0 3 I I I I I I ALUMINA S 35-50 MINERVILLE S5-66c S 35-66 b WESTRUM an i^ < so do C o Figure 11. Differential-Thermal-Analysis Records Showing Deflection as a Function of Temperature on the Warming Cycle.

0 0D t n 00 0I O S35-56a, I I, I I I I 1 I I I' \ S35-64 S35-55 S35-56b S32-65c S32-66Qa \ S35-63 NOTE SCALE CHANGE S35-62 FROM PRECEEDING 2:1 DEFLECTION D 1 0000 tn 0 I I I I I I oFie O1. (Co ) o Figure 11. (Continued)

-17Now it should be recalled that the study of lithium doped magnetite was begun in order to study the effect of quasi oxidation without actually oxidizing the sample. In Figure 5 we see what happened when the very pure synthetic magnetite began to oxidize. It should also be pointed out that the powdered samples are more prone to oxidize than samples studied by other investigators.(l0) Figures 6 and 7 show additional evidence of oxidation of a synthetic magnetite. Figures 8, 9 and 10 show effects similar to oxidation brought on by doping with lithium. The effect of doping on the transition as seen through the magnetic moment experiment disappears when the mole percent of magnetite drops below 98.5%. The DTA results which are shown in Figure 11 indicate a trace of a transition down to 96% mole percent of magnetite. One last comment should be made about the saturation magnetizations per unit density given in the table. Small amounts of non-interacting impurities in the samples will cause what could be called "mass loading." Mass loading amounts to a measured quantity of material larger than the ferrimagnetic portion. This mass discrepancy makes the measured a smaller than the true a

CONCLUSIONS First, I must point out that I have no positive explanation for the observed anomaly. The anomalous magnetic transition observed in experimenting with partially oxidized and lithium doped magnetite may simply be a property of the powdered polycrystalline samples used, but the results on the high purity magnetite before oxidation appear to rule out this possibility. What we might be observing is a double transition such as barium titanate(l2) undergoes. We might be observing a transition from the orthorhombic to the cubic via a tatragonal intermediate phase. The anomalous magnetic transition does exist and should be investigated in greater detail. -18~1

FOOTNOTES 1. P. Weiss and R. Forrer, Ann. Phys., 12, 279 (1929). 2. E. J. W. Verwey, Nature, 144, 327 (1939)o 3. E, J. W. Verwey and P W. Haayman, Physica, 8, 979 (1941). 4. Eo J. W. Verwey, P. W. Haayman and F. C. Romeyn, J. Chem. Phys., 15, 181 (1947). 5o Supplied by Professor E. F. Westrum, Jro, Dept. of Chemistry, The University of Michigan. 6. D. C. Ray, The University of Michigan Industry Program of the College of Engineering, IP-684 (1964). 7. R. C. Mackenzie, The Differential Thermal Investigation of Clays, {Mineralogical Society (Clay Minerals Group), London, 1957}. 8. D. C. Ray, Doctoral Dissertation, The University of Michigan, 1962. 9. E. W. Gorter, Philips Res. Rept., 9, 295, 321 and 403 (1954). 10. See for instance, C. A. Domenicali, Phys. Rev., 78, 458 (1950). 11. D. C. Ray, Phys. Rev., 135, A436 (1964). 12. L. R. Bickford, Jro, Phys. Rev., 76, 137 (1949). -19