T HE U N I V E R S I T Y O F M I C H I G A N COLLEGE OF ENGINEERING Department of Chemical and Metallurgical Engineering Ann Arbor, Michigan Technical Report MgO -MgFe204 MICROSTRUCTURES by Howard McCollister and Lawrence H. Van Vlack ORA Project 05612 under contrac, t with: OFFICE NAVAL RESEARCH ONR Contract No: Nonr-1224(47) administered throurh: OFFICE OF RESEARCHI ADMINISTRATION November 1964

O M0r-Fe 20, MICROSTRUCTURES b; H.toward McCollister and Lawrence H, Van Vlack Periclase (Ms0O) dissolves iron! oxide into its structure at elevated temoeratures and oroduces a solid solution called magnesiowuistitee This soliKd sol:ti:orln is com o.onl7r encountered in Mg0eO-containinrl basic refractories when those refractories are used in contact with iron or iron oxide, Becau ase the solubility lirmit of FoC03 in rooriclase decreases from aoproximately 70 weight per cent at 1700~C to about 10 weight per cent at 1200~oC(1) a two-phase microstructure develo(s during cooling.(2) This microstructure contains an intimate mixture of magnesioferrite (MgFe20)f,) in a matrix of periclase. The distribution of magnesioferrite in the periclase varies considerably: (1) sometimes it exists as a grain boundary network; (2) elsewhere it has a Widmanstatten pattern within the grains, or (3) it may be present as spheroidized particies within the periclase grains. It has been assumed that these three basic inicrostructures are the result of thermal histories however, this nature of their develoement has never been verified, This briof report re: lates the3se microstru", — tures to var-iations in heat treatment and suggests their ori — gi.ns PROC.EDIURE Pi'ne nowdeors of iPeagent srade Mi0 and Fe203 were mixed dr y, 6tlhe}n wi.th alcohol before oressing into 1/4-inch cylindrical pellets. An l8-hour solution treatment at temperatures equal to or greater than 1400~C was used to give a uniform distribution of the two components in the magnesiowustite solid solution.t Subsequent to solution treatment the samples were either water-quenched or silow-cooled, This was followed by annealings at sub-solvus temperatures for times ranging from 0,5 to 27 hours, Standard grinding, mounting, and polishing procedures

mwere used i.n the preparation of samples for reflected light miScroanalysis Lattice dimensions were determined by established diffraction procedureso RESULTS AND DISCUSSION The results arxe presented as photomicrographs of selected samples (Figso 19) Phase relationships are consistent with the MgO-Fe203 diagr"am (Fig~ 10).I For examplee, a 60-40 MgO-Fe203 composition does not dissol.ve all of the Fe203 but leaves excess magnesioferri.te (Fig0 1)o. In contrast, this same composition shows only one phase when quenched from 16000~C All compositions which show only one phase when quenched reveal prec iptated magnesioferrite when reheated to sub-solvus temperatures (Fig~ 2) 0 The size oe the parent magnesiowustite grains varied w.ith the solutionstreating temperature as shown in Figs. 3 and 4o The original grains are faintly outlined with a precipltate of magnesioferriteo At 16000C the "diameter" of the magnesiLowustite grains was about five times greater than for similar treatments at 1400CoC The nature of the magnesioferrite precipitate varies with the annealilng temperature as shown in Figs, 2 and 5. Althcough otherwise s imillar, the lower annealing temperature (11000C) shows a fine Widmanstatten pattern of magnesioferrite within tthe former magnesiowustite grains and has negligible grain boundary precipitation (Figo 2)I When the precipitation oocur.s while anneali ng at 12000C (Fig 5)), intergranular magn.esfi.cerr te appears and the intragranular magnesioferrite spheroidi i1 z e s The magnesi cferrite precipitate agglomerates with extended annealrings o Figure 6 may be compared with Fig0 5 for a 1.200~C anneal, and Fig0 7 may be compared with Fig0 2 for a a1l00C annreal As expetegd, the agglomeration proceeded more rapidly at the higher temperature (Contrast Figo 6 with Fig~ 7)o Magnesicferrite precipitates rapidly from the magnesiowifstit4e solid solution~ This is illustrated by different areas of the same Sl/4linch quench sample0 The microstructure

in Fig. 1 was near the surface, and that in Fig~ 8 was near the center, Incipient precipitation from the solid solution occurred in a matter of seconds. This rapid precipitation from solid solution is closely related to the crystal structures of the two phases. Each has the same fcc pattern of oxygen ions; as a result, the precipitation occurs simpiy by movement of small.er c ations, With slower cooling (33 minutes from 1400rC to 1100~C) a di ferent microstructure is developed as shown in Fig. 9. Both the grain boundary and the Widmanstatten precipitate are coarser and better defined, Furthermore, there is a magnesioferrite-fr-ee zone arorlund each of the former magnesiowustite gr arns This microstructure is best analyzed by concluding that the grain boundary precipitate was the first- to form after the temperature was lowered below the solubility limit. The excess Fe203 from the adjacent grain diffused toward the boundary as cooling continued, thus depleting the rim of the g; ain of extra iron oxide, The center of the grain remained supersaturjated with Fe203 however, until a Widmanstatten precippitate was finall y initiated, Of course this Widmanstatten structure could not grow into the outer zone of the grain which had lost Fe203 Zo the boundary. This slow-cooled sample had fewer buxt coarser Widmanstatten plates than a comparable quenched and annealed sample (Fig. 2), indicating that most of the preclp itation occurred before 1100)C was reached, The above microstructure (Fig. 9) shows that the Widmanstatten pattern was initi ate d within the grain rather than at the boundaries~ A Widmanstatten structure implies a crystali-graphic correliatilcn beftween the two phases, In these microstructures the maximum number of plate orientations ever observed with any single periclase grain was three. This is signific ant since the maximum number of plate orientations cortresponds with the number of distinctly oriented planes within the form {hkl}. Since all cubic forms other than 100} have mo.,e than three distinctly oriented planes, this strongly suggests that the magnesioferrite (MgFe204) precipitates along the f!0O0 planes of the periclase (MgC). If sc, it is probably safe to conclude that the matching plane in the magnesioferrl.te spinel is also of the {lOO) form since each phase has

the same fcc pattern of oxygen ions, differing only in the interstitial locations of part of the cations. CONCLUSIONS Magnesioferrite-periclase microstructures can be correlated to their thermal history, Grain boundary precipitation occurs at higher temperatures, whereas a Widmanstatten precipitation is favored at lower temperatures. Magnesiowustite which is cooled slowly through the solubility limit will first produce magnesioferrite at the grain boundary, followed by a Widmanstatten precipitate of magnesioferrite within the grains~ Prolonged heat treatments at any temperature favor the spheroidization of the magnesioferrite within the peril clase matrixe ACKNOWLEDGEMENTS This work was performed under a grant from the Office of Naval Research0 Its support is gratefully acknowledged.

R.ziKFEREN CES 1. Phillips, B., Somiya, S. and Muan, A., "Melting Relations of Magnesium Oxide - Iron Oxide Mixtures in Airtt J, Amero Ceram. Soc., Vol. 44, [4], pp. 167-169, (1961). 2a, Wells, R. G. and Van Vlack, L. H,, "Mineral and Chemical Changes in Periclase Brick Under Conditions of Steel Plant Furnaces," J, Ameri Ceram. Soc., Vo, 34, [2], ppo 64-70, (1951). 2bo Tro jer, F', and Konopicky, K,, "Die Kristallisationsfor-men von Magnesiurnferrit bei Ausscheidung aus dem festem Zustand," Radex Rundshau, [7/8], pp. 149-153, (1948), 2co Homer, P. No and Richardson, Ho M., "The Reaction of Some Synthetic Spinels with Magnesia at High Temperatures," Trans. Brito Ceram. Soco, Vol. 63, [8], pp. 389-415, (1964),

~;iil~~iifY~iti;J F i1. Mg0 6Fe O Fig. 1. MgO-Fe2O mic rost ruc ture. Eiixampe: 60 g-40 Fe203 heated for 18 hours at 14000C and quenched. The two-phase microstructure was retained. In all..?,microstructures the bright phase is magnesioferrite (MgFe204) and the dark phase is an FeO-contain-.Ej,:,.ing periclase. X 1000. ~~~:..j Fig. 2. MgO-Fe20 microstructure. Example: 70 MgO-30 Fe2O0 heated for 18 hours at 1600od quenched in water, reheated to 11000C one hour. A Widmansto'tten precipitate of magnesioferrite (bright) formed within the origi- nal single-phase magnesiow'istite. X 1000. — 4 Z Fig. 3. MgO-Fe2% microstructuro. Example: 80 MgO-20 Fe2O3 heated 7~~~~~~~~~~~~~~~~~Ali~~~~~~~~~~~~~~~~~~~~:,:. ~.~./::i:,......:......... 18 hours at 16000C, quenched in water, reheated to 120000 one hour. Compare the initial magnesiowustite grain size with Fig. 4. X 1000. ~:;.:: Ff:.r':~'.~::~~:~..;:~.~~::::::f:~:..:..:;[:~ -':.::'::'~.:::::;:x:?;~~:~:~%~~ ~~~.:~~.~: I-lf~ w t r eeae o 20~ n.........,::~ ~~:.~~........~:.'~"'':'::'":::~:~~':~::~:::~~~:::i~::~...~~:~::"':".............. e i n t i a l m g

^.tiii,:~: Fig. 4. MgO-Fe 203 mic rostruc ture. F; Exampl e: 80 MgO- 20 Fe203 heated:8or 18 hours at 140o0C quenched kgdset<*0;in water, reheated to 12000C for 4>:"' one hour. There ias less growth of the initial magnesiowlistite grains than in a 1600~C anneal Fig. 5. <:gF,:e2. i. r:ost r~ re.:X.: cmaale3 F S. microstructure. trWea eFdl%*>i..:at..... (Fi:g.':...i':/,... 2').. X 10'0. 0Example 70 0 Fe2 heatedht for 18 hours at 1600a~Ct uenched qe L~xamp to cin watered to 100C for.. e~27 hours. As c ntrated o the one hour. There is more gratin rme i.. 5 sn~hrheroidization of been a m. jr....t granular precipitate than in the mr comparable microstructure treat-i boundaies "I~~~~~~~~~~~~~~~~~~~r"'':.:'::[.~:/.:%...:-~

...... Fig. 7. MgO-Fe203 microstructure. Example: 70 Mg0-30 Fe203 heated for 18 hours at 160000, quenched..::,.'.in water, reheated to 11000C for for>' 18''i...............:....::- 2 7 our s, ndoed tim e s at this h s quenched U k F..1h(cf. Fig. 2) but did not produce s. Tr tsa relocation o- the magnesioferi eferrite to the boundaries (cfo hea;ti 9 tFig 6)e, X 01000 -.....:~.....:~:::.':':::::-.~:-:.:.~.........:..... ~:.~~:~'::~........!.;............'':St9'9*F;9' f... Fg.: 8. Mg~O-Fe 0 microstuFig 9e mr m_:e 60xgpeFe7 Mg -30 Fheee hea u for 18 hours at 1400a 1 0, 9g9',. *Xquenchedgn Unlike Fil from thelL same 1/4 i. sample, this microstructure occurred at the center. where the quenchng rheated was sower.e nTherefore thi magnesioerein the supertingterio w{ stite developed some magnesio-o temri ferrite precip itate without a heat treatment. X 1000. Example: 70 M.-30 Fe203 heated for 18 hours at 1400~C, cooled slowly (33 minutes) from 14000C to 1100~C, held one hour at 1100~C. A coarse boundary precipitate of magnesioferrite formed by depleting Fe203 from the adjacent'part of the magnesiowustite grains. A a-iiWx manst tten precipitate formed in the supersaturated interior of the grains. This latter precipitate could not grow into the declase. X lO00.,

2000 la. L MW w, 1500 4 MF WT. PERCENT 2 5 Fig. 10. MgO-Fe O phase diagram (in air). After PEi~lips, et al, Reference 1. MW = ate MF = magnesioferrite, H = hematite. magniesioferrite,. H = hematite.