EFFECT OF IRON OVER THE MAGNESIA-ALUMINA SPINEL LATTICE

Ortiz Ubaldo, Aguilar Juan, Kharissova OxanaUniversidad Autónoma de Nuevo León, Facultad de Ingeniería Mecánica y Eléctrica,Apartado Postal 076 F, Cd. Universitaria, San Nicolás de los Garza, N.L. 66450, México.

ABSTRACTThe effect of iron over the magnesia-alumina spinel (MgAl2O4) lattice were studied by meansof observing peak positions and their intensities in X-ray charts. Two different methods forproducing magnesia-alumina spinel (MgAl2O4) were used in this work. The first one consistsin heating up magnesia (MgO)-alumina (Al2O3)- hematite (Fe2O3) mixtures in a resistancefurnace. Temperature was 1400°C, which is enough to produce spinel, especially when ironwas added. A second method for achieving temperatures over 2000°C was by means ofmicrowaves as energy source. Patterns from the samples show that indeed there are changes inthe lattice parameter and cationic distribution. It has been observed that a spinel produced athigh temperatures without additives gives a pattern that does not correspond to the perfectspinel structure. When hematite was added, the obtained MgAl2O4 is shows a cationicdistribution that tends to correspond to the normal spinel structure.

KEY WORDSSpinel, lattice, ceramics, iron

INTRODUCTIONThe magnesia-alumina spinel (MgAl2O4) has a crystal structure that shows similarities anddifferences to those of both, magnesia (MgO) and alumina (Al2O3). MgO and MgAl2O4 spinelhave cubic close-packed arrays of oxygen ions, in contrast to Al2O3 which has a distortedhexagonal close packed array of oxygen. The Al3+ ions occupy octahedral sites in both Al2O3and spinel, while the Mg2+ ions are octahedral in MgO but tetrahedral in MgAl2O4 (Figure 1[1] ). This structure is determined by the space configuration of relatively large oxygen ions,with trivalent and divalent cations between them. Due to the relatively large size of theoxygen, it is possible to have changes in the cationic distribution without major changes in thelattice. This situation gives a wide range of compositions for the spinel MgO-Al2O3, rangingbetween 40 through 85% of MgO (All percentages are molar) in a binary system. Mostmagnesia-alumina spinel compositions have approximately the same lattice constant [2]: from0.8 nm, up to about 0.85 nm.The objective of this work focuses on a summary of the results of tests conducted forproducing magnesia-alumina spinel with iron, which bears two charges, either divalent (2+) ortrivalent (3+), and therefore could take either magnesium or aluminum places, and couldproduce changes in the lattice parameter and order degree as function of the iron content.

MgO - Al2O3 SYSTEMSpinel structure has Fd3m symmetry with only the oxygen atom with (u,u,u) independentfraction coordinates and a multiplicity of 32. Eight cations lie in tetrahedral sites with43msymmetry, and sixteen in octahedral sites. Crystallographic data for the spinel structure aregiven in Table I.

A complicating factor in the spinel structure is that the cation distribution over tetrahedral andoctahedral sites may vary giving two extreme cases, which are normal and inverse spinel.Such distributions are described as follows:A2+δB3+1-δ[A2+1-δB3+1+δ]O4 (1)Here the ions placed in the tetrahedral sites are written before the parenthesis, while those inthe octahedral sites are inside the square brackets. The parameter δ defines the disorderdegree, complete disorder gives δ equals 1/3, in normal spinel δ is 1, while for inverse spinel δis zero. Tetrahedral sites in the face centered cubic lattice of oxygen ions have less volumethan the octahedral ones. Actually, the structure is distorted by the movement of the oxygenions, which increase the volume of tetrahedral sites, but decreases the volume of octahedralsites.Distribution of A and B cations depends of several conditions, either geometrical, as atomicradii or energetic as the valence and charge state of the lattice. Tetrahedral sites are smaller thanoctahedral ones, thus it is expected that trivalent ions go to tetrahedral sites, while bivalent ones,which are bigger, accommodates in the octahedral sites (W. Kingery, H. Bowen and D.Uhlmann [1]). Electronic configuration also plays a role in the cationic distribution, because theelectrons from the metals could give a directional bound which provide minimum energylattices. Tetrahedral and octahedral sites are made by the oxygen structure, which is negative,thus metallic ions with the smallest positive charge would be surrounded by just four oxygenions, while ions with the greatest charge would be surrounded by six oxygen ions. This is anattempt for providing local electro-neutrality within the lattice. The competition betweentetrahedral and octahedral sites size and the presence of both, bivalent and trivalent ions,introduce restrictions that produce a spinel where the rule of four and six oxygen ions describedabove is not followed. Therefore spinel is neither normal nor inverse and yet not totallydisordered. Description of this different spinels with the same stoichiometry is given through δ.Jacob et. al. [3] had pointed out that thermodynamic properties are different related to cationdistribution. One expression from the same researchers for cation distribution againsttemperature is:

( )( ) molJ

xxxRT 25540

21ln

2

=

−−

− (2)

Where: δ−= 1xAccording to Singh [4], the parameter δ depends on the conditions for preparing the spinelbecause it is related to the amount of vacancies, which is expected because their relation withtemperature.Equation (2) gives the opportunity to quantify the normality of spinel, notice how as Tdecreases to absolute zero, x tends to zero (normal spinel) and when T goes to infinity x tendsto 2/3, meaning that δ is 1/3 which means a totally disordered lattice.

EXPERIMENTAL PROCEDUREA set of samples with different composition in the system MgO-Al2O3-Fe2O3 (reactive degree)were prepared. MgO was heated prior to use at 800oC because of its hygroscopiccharacteristics. The reagents MgO and Al2O3 were mixed thoroughly to get intimate contactbetween the powders. Samples were milled and weighted, then placed into a crucible. Theoven was at 1400°C and the exposition time was 15 h., used ratios for the mixtures arepresented in the second through fourth rows in Table II which are the atomic percentage ofFe2O3, MgO and Al2O3 employed respectively. In the “Sample composition” row Arabicnumbers correspond to the mixture and letter “E” or “D” means respectively thermodynamic

equilibrium of the given mixture and actual composition of the sample obtained from thatparticular mixture. Asterisks are shown for compounds that were not calculated, eitherbecause a lack on thermodynamic data or a wide variation in the available information. . AsK. Jacob et. al [3] have pointed out that despite the pivotal role of the MgAl2O4 in therefractory industry there is a lack of reliable Gibbs data. Therefore equilibrium was calculatedassuming ideal behavior of the components forming a single phase as Darken and Gurrysuggest for gases [5]. We validated our method by performing gas calculations that arepresented in the same reference, ceramic ones from F. Hummel [6] compilations and diagramsshown by Kingery et. al.[1]. The data were obtained from JANAF tables [7] and O.Kuwaschewski and C. Alcock [8] with good results. Then with the validated method wefound the compositions that are presented in Tables II and III.Microwaves were used for achieving higher temperatures. Here, the sample was placed into acrucible inside of a conventional microwave oven with a magnetron working at 2.45 GHz and800 W. Preheating was aided with 0.5 g. of graphite, which was not in contact with thesample and did not produce any compound. According to previous experience with this kindof material [9], mixtures were exposed to microwave energy for 30 minutes. Ratios arepresented in Table III following the same rules as in Table II.

RESULTS AND DISCUSSIONOnce that tests were completed, the samples were removed from the crucible and analyzed byx-ray diffraction for confirming the spinel and other products formation.Notice how in the case at 1400°C (Table II), low hematite concentrations produced spinelclose to thermodynamic calculations, but at 8.8 % of hematite, MgAl2O4 drops and the mainproduct is Mg(Al,Fe)2O4, AlFeO3 and MgFe2O4, even when thermodynamically, productionof spinel does not get down with that small amount of hematite. At 2000°C (Table III), it waspossible to produce MgAl2O4 closer to thermodynamic equilibrium, however above 15% ofhematite, MgAl2O4 dropped and the main product was now Mg(Al0.91Fe0.09)2O4. Thesereactions take Mg and Al from the mixture and avoid MgAl2O4 formation. Table IV presentssome characteristics of the possible compounds in this system. It is important to note thatnone of these late compounds were present in the mixtures at the beginning, thus if they wereformed it is because they are thermodynamically stable under the conditions that are imposedby iron presence. Calculations were done under the supposition of ideal behavior of species,although deviations between calculated and actual values are difficult to explain from simplechanges in activity of the species. It is clear that thermodynamics is not dealing with kineticsissues, and that any kinetic model evaluated at time equals infinite must coincide withthermodynamics. However, what is often overpassed is that many thermodynamic data arecoming from experiments conducted at normal scale times, and such kinds of constantsinvolve kinetic aspects. In other words, if diffusion of an ion is not possible because theactivation energy for this mechanism is too large that was not reached, then experiments forevaluating thermodynamic constants of this particular system would give an equilibriumcriteria that involves that difficulty. Jacob et. al. [3] explain how thermodynamic data isaffected by cation in MgAl2O4 spinel distribution which is precisely the case in here. Otheraspect is that, as Zhang [10] established earlier, Al3+ diffusion in MgO at high temperatures isrelated with the presence of vacancies created by dissolution of Al3+ in MgO, while theamount of these ones is directly proportional to the Al3+ concentration in the range of 1560-1900°C. Therefore less MgO means less Al3+ dissolution (and diffusion), which incombination with more Fe, which could be either 3+ or 2+ reduces even more the opportunityof Al3+ to move within the lattice. Especially because melting point of hematite and iron

spinels are lower than melting point of alumina, which means that at a given temperaturediffusion of iron is greater than diffusion of aluminum. These interactions between atoms arethe only possible explanation, within this work, for the fact that iron spinels are producedinstead of magnesia-alumina spinel, which is assumed the thermodynamic product. Noticethat simple geometrical suppositions are not enough to explain the diffusion changes, sinceatomic radius of Al+3 is smaller than Fe+2 or Fe+3 (Table V).If the issues were diffusion and cationic distribution, higher temperatures (2000°C) wouldshow more iron products at less hematite contents. In other words, higher temperatures shouldmean sample compositions closer to the thermodynamic equilibrium only if diffusioncompetition of iron and alumina is the same than at low temperature (1400°C), which is notthe case because the iron spinels are liquid at the current temperature (2000°C) and diffusionis quite higher. Therefore, even with these conditions the results would be less magnesia-alumina spinel and more iron compounds. Looking at table III, besides the above discussion,the liquid phase consumes the most of the magnesium and aluminum oxides contrary to thethermodynamic approach. Phenomenological description of the spinel synthesis from oxides-precursors, deals with the bands of oxides that should be broken in order to allow migration ofatoms within considerable distances (in the atomic scale). Such ions as Mg2+ in MgO and Al3+in Al2O3 are usually determined as fixed in their appropriate cell sites, so it is difficult forthem to move to empty sites. It could be assumed that Fe3+ ions that do not enter MgAl2O4structure, but Fe3+ ions change their valence state to Fe2+ forming FeAl2O4.Only at high temperatures, such ions have a sufficient thermal energy, which permit them toleave their normal sites and diffuse through the crystal. To obtain such a sufficient energy, it isnecessary to increase temperature or to introduce melting additives in the spinel precursorsystem, which could decrease the sintering temperature. Hematite performs this task,especially because its melting point is lower. Cation distribution is not just geometrical, hencein the competition between iron and aluminum for either tetrahedral or octahedral sites,structure distortion is created and the lattice parameter is modified.Figure 2 shows some of the experiment that were conducted in a wide range of compositionsbeyond those presented in tables II at 1400°C. Notice the variety of possible phases producedin the system MgO-Al2O3-Fe2O3. When the mixture is free of iron, the products are spinel andunreacted MgO and Al2O3. According to diffraction patterns, the lattice parameter is 8.0831 Åfor the spinel produced either at 1400°C in the electrical resistance furnace or at 2000°C in themicrowave oven. This similitude suggests that the actual energy that went into the sample iscomparable.Successive addition of iron (as hematite) reduces the amount of MgAl2O4 spinel of the samekind. Other product is FeAl2O4, as it was shown in table II.A very important observation was that the MgAl2O4 spinel that was produced at higherhematite concentrations changes its lattice parameter from 8.0831 to 8.08Å, this producesslight shift of the diffraction peaks (Bragg’s law). Other aspect is that the intensity of thepeaks also varies, for instance the plane (1,1,1) located at 19.028° (2θ) which has a relativeintensity of 0.35 in the former spinel changes to 0.04 and 18.987°. Intensity of the peaks isrelated to the structure factor, and this one to the fractional coordinates of each atom in thelattice.One aspect that has to be taken into account is that the oxygen parameter, which is thedistance between the oxygen ion and the vertices of the cube along the diagonals, is affectedby the distortion that is introduced to the lattice by the metallic ions. Therefore the size of thetetrahedral and octahedral sites changes, and cations can switch their positions in such a waythat the parameter of order (δ) of the spinel is modified, giving a structure factor that modifiesthe peaks. Cation distribution that has been presented (Equation 2) corresponds to spinel free

of iron. Figure 3 shows δ (calculated from equation 2, lower curve) against temperature,showing that spinel is not normal (In normal spinel δ equals 1). The lattice parameter for thespinel is 8.08 Å. We have found that normal spinel is not produced when hematite is in smallamounts in the mixture, but as hematite was increased this “normal” spinel was betterproduced. The net effect of iron is that the presented curve would be shift to the right, or inother words, apparent temperature is greater when hematite is part of the system, which isreasonable because melting point of iron compounds are lower. Global effect is represented bythe constant in equation 2 (25540 J/mol), which is related to the energy for the diffusionprocess. According to the explanation related with vacancies and melting points that has beengiven above, more iron means that diffusion is harder for aluminum. In Figure 3 there areother two curves calculated with the same equation, but considering 1.5 and 2 (upper curve)times the energy value (38310 and 51080 J/mol). Lines are shifted in such a way that it isshown that order parameter is about the same at higher temperatures when iron is present.That is, the curves and the results are in good agreement. This is also supported on the factthat the standard pattern for the X-ray diffraction data bank was taken at 25°C from a samplefurnished in an electric arc furnace, and the excess MgO was removed with hot HCl [11].Spinel obtained at low temperature by processes such as solid state reaction tends to havenormal structure. In this diffraction pattern, peaks do not correspond to normal spinel. At lowiron concentrations, the obtained spinel is just like this one, no matter if it was produced byconventional heating at 1400°C or microwaves at 2000°C. When the process was conductedin a binary system, the formed phase is mainly spinel. Figure 4 shows several compositionswere this spinel was found.Presence of spinel Fe (Fe1-xAlx)2O4 was earlier described by Moon and Philips [12] at 1500°C,and this study is in agreement with our results.

CONCLUSIONSThis work gives a general idea of the phases that can be found in the alumina-magnesia-hematite system, which is not greatly reported. Thermodynamic calculations depend on thereliability of data, and in this work is noticeable that there are several kinds of spinel, whichexplain the wide values for thermodynamic constants. Addition of hematite is not againstmagnesia-alumina spinel production until certain critical amount (8% at 1400°C and 15% at2000°C) of hematite is added and then equilibrium tends to iron spinels, this effect is notreduced with the temperature, proving that the change of phase to liquid plays a role in thediffusion of the iron ions against the aluminum ones. The presence of iron induces productionof magnesia-alumina spinel in normal structure at high temperatures, when the mostcommonly found is the structure with a lattice parameter of 8.0831 Å instead of the 8.08 Å.Iron induces order into the magnesia-alumina spinel, when MgAl2O4 is obtained at hightemperature with iron the order is similar to those prepared by solid state processing at lowertemperatures. Results confirmed that aluminum ions reduce their mobility when iron ispresent, explaining why MgAl2O4 spinel is not produced above mentioned critical amounts ofhematite.

ACKNOWLEDGMENTSAuthors express their gratitude to CONACYT (Mexican Council for Science and Technology)and PAICYT (Autonomous University of Nuevo León Research Program for Science andTechnology) for its financial support.

REFERENCES

[1] W. Kingery, H. Bowen, D. Uhlmann, “Introduction to Ceramics”, John Wiley andSons, (1976)

[2] D. Maschio, B. Fabbri, C. Fiori, “Industrial applications of refractories containingmagnesium aluminum spinel”, Industrial Ceramics, 8 (1988) 3

[3] K.T. Jacob, KV. Jayedevan, Y. Waseda “Electrochemical determination of the Gibbsenergy of formation of MgAl2O4”, Journal of the American Ceramic Society, 81(1998) 209

[4] V.K. Singh, R.K.. Sinha “Low temperature synthesis of spinel” Mater. Lett. 31 (1997)281-285

[5] L. Draken, R. Gurry, “Physic Chemestry of metals”, McGraw Hill Company (1953)[6] F.A. Hummel, “ Introduction to phase equilibria in ceramic systems”, Marcel Dekker

Inc. (1984)[7] D. Stull, Proplet, “JANAF Thermochemical Tables, Second edition, National Standard

reference Data (1971)[8] O. Kubaschewski, C.B. Alcock “Thermodynamic thermo-chemestry”, Fifth Edition,

Maxwell Macmillan International Editions (1979)[9] J. Aguilar, M. González, I. Gómez, “Microwaves as an energy source for producing

magnesia-alumina spinel”. Journal of Microwave and Electromagnetic Energy, 32(1997)

[10] P. Zhang, T. Debroy, S. Seetharaman,”Interdiffusion in the MgO-Al2O3 spinel withand without dopants”,. Metallurgical and Mater.Transac. A., 27 (1996) 2105

[11] JCPDS-International Centre for Diffraction Data (1997)[12] A.K. Moon, M.R. Philips, “Iron and spinel precipitation in iron-doped sapphire”,

Journal of the American Ceramic Society, 74 (1991) 865-868[13] Boudias, D. Monceau, CaRline Crystallography Version. 3.1 (1989-1998), France

Table I: The spinel structure: crystallographic dataSpace group: Fd3m, No. 227, face centered cubicAtomic coordinates for MgAl2O4.

Coordination number:Mg - tetrahedral, MgO4Al - octahedral, AlO6 O - tetrahedral, OMgAl3

Atom Position Fractional coordinatesO 32e uuu;uuu;uuu;uuu;

1/4-u,1/4-u,1/4-u; 1/4+u,1/4-u,1/4+u;1/4-u,1/4+u,1/4+u;

1/4+u,1/4+u,1/4-u + face centeredAl 16d 5/8 5/8 5/8; 5/8 7/8 7/8; 7/8 5/8 7/8;

7/8 7/8 5/8; + face centeredMg 8a 000;1/4 1/4 1/4+ face centered

Table II: Description of the mixtures and samples obtained at 1400°C. Roman numberscorrespond to mixture compositions, while Arabic numbers correspond to the composition(atomic) of each particular mixture after being processed. Letter “E” is for thermodynamicequilibrium and “D” for composition estimated from its x-ray diffraction chart. Compoundsthat were not calculated are represented by an asterisk.

Mixture I II III IV VFe2O3 3 5 8.8 10 20Al2O3 50 50 50 50 50MgO 47 45 41.2 40 30Samplecomposition

1E 1D 2E 2D 3E 3D 4E 4D 5E 5D

MgO 10 11 9 5 7 0 6 0 3 0Al2O3 14 13 16 7 20 0 21 0 30 0MgAl2O4 70 64 67 57 60 0 57 0 40 0Mg(Al,Fe)2O4 * 12 * 30 * 56 * 58 * 55MgFe2O4 0.9 0 0.1 0 0.1 42 0.1 40 0.1 27AlFeO3 * 0 * 1 * 2 * 2 * 0Fe3O4 0.5 0 1 0 2 0 2.4 0 5 18Al2FeO4 0.6 0 0.8 0 1.1 0 1.4 0 2 0Fe2O3 4 0 6 0 10 0 12 0 20 0

Table III: Description of the mixtures and samples obtained at 2000°C. Roman numberscorrespond to mixture compositions, while Arabic numbers correspond to the composition(atomic) of each particular mixture after being processed. Letter “E” is for thermodynamicequilibrium and “D” for composition estimated from its x-ray diffraction chart. Compounds thatwere not calculated are represented by an asterisk.

Mixture VI VII VIII IXFe2O3 5 8.8 15 20Al2O3 50 50 50 50MgO 45 41.2 35 30Sample composition 6E 6D 7E 7D 8E 8D 9E 9DMgO 13 22 11 4 7 4 6 3Al2O3 18 14 20 2 24 2 27 0MgAl2O4 61 28 56 48 47 47 40 0Mg(Al0.91Fe0.09)2O4 * 30 * 36 * 36 * 53MgFe2O4 0 0 0 0 0 0 0 0AlFeO3 * 1 * 0 * 0 * 0Fe3O4 3.5 0 6.6 0 12 0 15 0Al2FeO4 3.5 6 4.8 10 7 11 9 44Fe2O3 1.2 0 1.8 0 2.6 0 3 0

Table IV: Lattice parameters of different species found in the MgO-Al2O3-Fe2O3 system.International Centre for Diffraction Data [11]. Melting points are from references [1] and [8]

Name Compound Lattice parameter (Å) Melting point(°C)

Iron aluminum oxide(hercynite)

FeAl2O4 Cubic, a=8.1534 1780 [8]

Magnesium iron aluminumoxide (ferrospinel)

Mg(Al,Fe)2O4 Cubic, a=8.1905

Magnesium iron aluminumoxide

Mg(Al0.91,Fe0.09)2O4

Cubic, a=8.1

Magnesium aluminumoxide (spinel)

MgAl2O4 Cubic, a=8.08(Prepared by solidstate reaction)

2135 [1]

Magnesium aluminumoxide (spinel)

MgAl2O4 Cubic, a=8.0831(Prepared in anelectrode arc furnace)

2135 [1]

Hematite Fe2O3 Hexagonal,a=5.0317, c=13.737

1594 [8]

Corundum Al2O3 Hexagonal, a=4.751,c=12.97

2054 [1]

Magnesioferrite MgFe2O4 Cubic, a=8.366 1700 [8]Magnetite Fe3O4 Cubic, a=8.391 1597 [8]Periclase MgO Cubic, a=4.203 2825 [8]Wustite FeO Cubic, a=4.296 1400 [8]

Table V: Atomic radii of the different ions [13]

Name Ion Atomic radius (Ǻ)Aluminum Al+3 0.51

Iron Fe+2 0.74Iron Fe+3 0.64

Magnesium Mg+2 0.66Oxygen O-2 1.32

Figure 1: Lattice of the spinel structure, after W. Kingery[1]

Figure 2: Ternary equilibrium diagram for the magnesia-alumina-hematite system showing the chosen compositions for this work and thephases found

Figure 3: Parameter of order δ as function of temperature. Lower curve is for equation 2, upper one is twice the constant of equation 2,and the last curve is with one and a half. x = 1 - δ

Figure 4: X rays diffraction patterns (2θ) of several samples obtained at 1400°C that exhibit Mg(Al,Fe)2O4.Samples compositions are (mol.%): 50) 35 MgO-55 Al2O3-10 Fe2O3, 46) 50 MgO-40 Al2O3-10 Fe2O3, 33) 40 MgO-50 Al2O3-10 Fe2O3,and 32) 40 MgO-40 Al2O3-20 Fe2O3. Intensities are given at arbitrary scale, just for comparing the results, target was made of cooper.