134

IEEE TRANSACTIONS ON MAGNETICS, VOL. MAG-20,NO. 1, JANUARY 1984

REFERENCES
111 P. 1. Bon~hardet al., Large capacityion-implanted bubble devices, IEEE Truns. Magn., vol. MAG-18, no. 2,pp.137-740, Mar. 1982. [2] D. T. Ekholm, et al., NDRO detector for ion-implantedbubble devices, J. Appl. Phys., vol. 53, no. 3, pp. 2525-2527, Mar. 1982.

[3] P. I. Bonyhard et al., Progress on ion-implanted bubble memory chips, presented at 31M3 Conf., Montreal, Paper FB-3, July 1982. [41 T. J. ~~l~~~ al., Effects of N - H ~ et doses inion implanted hubble memory test chips, presented at 31M3 Conf., Montreal, Paper FB-4, July 1982, and IEEE Truns. Magn., vol. MAG-18, pp. 1358-1360, NOV. 1982.

Interaction of Magnetic Domain Walls with Microstructural Features in Spinel Ferrites

I-NAN LIN, RAJA K. MISHRA,
AND

GARETH THOMAS

Abstract-Lorentz microscopy studiesof (Mn,Zn)FezO4 and LiFe50s show that grain boundaries, pores, microcracks, stacking faults, nonand magnetic second phase particles act as pinning sites for moving domain walls. Secondphaseparticles,pores, and some grain boundariesalso act as favorable sites for reverse domains to be nucleated. In addition, it is observed that domain walls cannot move smoothly as they approach a grain boundary.

I.INTRODUCTION HE MAGNETIC properties of ferromagneticand ferrimagnetic materials depend on theirchemical composition, crystal structures, and microstructures. The recognition of the effect of microstructure on some of the magnetic properties has led to the development of various processing techniques and thermomagnetic treatments so as to optimize the microstructural variables for the desired combination of properties [ l ] -[3] . The actual studies of the effects of grain boundaries, grain sizes, porosity, second phases, etc. on coercivity, permeability, maximum energy product and other properties have also received considerable attention [3] . Development of various experimental techniques, such as the Bitter technique and the Kerr effect have made it possible to directly examine the effect of microstructures on the domain configurations [4]. The development of Lorentz microscopy (in transmission electron microscopy)hasenabled imaging of magnetic domains and domain walls to be done at higher resolutions and also has

made it possible to study the interaction of magnetic domains and microstructure [5] . In the latter technique,if a magnetic field can be applied in situ, then the dynamics of the domain wall motion and its interaction with the microstructures can be examined directly. Since domain wall motion is related to many of the magnetic properties, such dynamic studies are of interest in understanding themagnetic behavior of the materials. In this paper a study of the interaction of magnetic domain walls with grain boundaries, stacking faults, second phase precipitates, grain boundary segregates, pores, and internal microcracks in some spinel ferrites is described. The motion of the domain wall is studied in situ and the possible influence of such motion on the propertiesis discussed.

11. EXPERIMENTAL Sintered (Mn, Zn)Fe,04 withand without CaO impurities and flux grown single crystals of LiFe,O, were used for the present study. The detailed composition and processing history of (Mn,Zn)Fe,O, can be found elsewhere [6] . The single crystal LiFe,08 was heat treated in air at 1200C for 1 h to obtaina two-phasematerial [7] . Thinelectrontransparent specimens, suitable for transmission electron microscopy, were prepared fromthe bulk material by ionthinning[SI.The Lorentz microscopy was done on a Philips EM301 microscope operating at 100 kV. Using the four lenssystem ofthe EM301, the Lorentz image could be focussed with the diffraction lens. A magnetic field was applied to the specimenby exciting Manuscript received April 25,1983; revised August 3,1983. This the objective lens of the microscope with a very low current work was supported by the Director, Office of Energy Research, Office of Basic Energy Sciences, Materials Sciences Division of the U.S. Depart- [9] . The strength and direction of the magnetic field in the ment of Energy under Contract DE-AC03-76SF00098 and by INER plane of the specimencould be controlled by appropriately (R.O.C.). tilting and rotating the specimen (using a tilt-rotation holder) The authors arewith the Department of Materials Science and Mineral Engineering and theMaterialsand Molecular Research Division, Lawrence so that it is at an angle with the magnetic flux. A strong field normal to the specimen surface was thus always present and Berkeley Laboratory, University of California, Berkeley, CA 94720.
0018-9464/84/0100-0134$01.00 0 1984 IEEE

LIN e t a l . : MICROSTRUCTURAL FEATURES IN SPINEL FERRITES

135

Fig. 1. Interaction of domain wall with an edge-on grain boundary (dotted line g-g) in (Mn, Zn)Fe204 sample. Domain walls parallel to grain boundary (1-1, 2-2) are stopped completely while the domain wall normal to the grain boundary (2-3) can move up and down. Top photo is the focussed image of a grain boundary. Lorentz image pairs u l , ol, and u2, o2 are taken with-different applied fields.

boundary and the domain wall. As shown in Fig. 1, when the moving wall is parallel to the grain boundary, the boundary acts as a pinning site for the wall. On the other hand, if the wall 111. RESULTSAND INTERPRETATION makes an angle with the grain boundary, the motion of the wall is retarded, there being least retardation for walls normal to Under the application of a magnetic field, some of the domains will grow and some will shrink as a result of the rotation the grain boundary (Fig. 1). When the grain boundary is inclined to the specimen surface (whereas normally, due to magof the magnetization vectors in the domains. The motion of the domain walls and their interaction with the microstructure netostatic effects, the magnetization lies in the plane of the surface), the effect can thus be examined. In the following figures, the direction sample and the domainwall is normal to the of the applied field is from top to bottom of the figure (the is similar. In addition to the pinning effects, the grain bounddirection is reversed in figures labeled with a negative number). ary region acts as a site for the nucleation of new domains. Inthe case of materials containingan intergranular phase A pair of Lorentz images in both underfocussed (u) and overfocussed ( 0 ) conditions ineach stage of magnetization are [6], as the domain wall approaches the boundary, its motion shown in Figs. 1-5 along with a sketch of the domainwall con- is retarded and the wall stops at a distance from the boundary. As the applied field is increased, the wall suddenly jumps as figuration in each case. a result of magnetization reversal by rotation in the region beA . Effect of Grain Boundaries and Grain Boundary Segregates tween the grain boundary and the impeded wall. Domains are The interaction of a domain wall with a grain boundary de- also nucleatedat these grain boundariesandjunctionsconpends,amongotherthings,onthe angle betweenthe grain taining the second phase. All these are shown in Fig. 2. It can could not be avoided inthis study.Themicrostructure was examined under normal operating conditions the microscope. of

136

IEEE TRANSACTIONS ON MAGNETICS, VOL. MAG-20, NO. 1, JANUARY 1984

Fig. 2. Underfocussed (u) and overfocussed ( 0 ) Lorentz images taken from the same area of a (Mn, Zn)Fe204 sample show how the domain walls move and get pinned at the grain boundary containing CaO segregates (dotted line). The domain wall 1-1 is stopped at a distance from the grain boundary in stages l and 2 and jumps to the boundary in stage 3 as the field is increased. When the field is reduced (stage 4), new closure domain E is created. Domain C is nucleated in stage 2 at grain boundary.

Fig. 3. BF image showing LiFe02 precipitates (d, e, g) in LiFe508 and Lorentz micrographs 1 and 2 showing domain wall pinning by the precipitates. Reverse domain (B) is nucleated as applied field is increased.

LIN e t a l . : MICROSTRUCTURAL FEATURES IN SPINEL FERRITES

137

Fig. 4. BF image shows the cation stacking faults in LiFe,OB. Domain walls (solid lines) are pinned at stacking faults in stage 1 when they lie parallel to the fault plane, but wall 1-1,normal to the faults, canmove. Stacking faults act as a site for nucleation of reverse domains (a) and @) in stages 2 and 3.

also be seen in the figure that if the applied magnetic field is reduced, the domainwall motion is not reversed. For example, domain wall 1-1 in Fig. 2(3) does not retreat; instead, itssplits and a new closure domain E is nucleated. The domain wall 4 - 4 in Fig. 2(3) cannot move at all until a new domain nucleates.

the semicoherent interfaces act as sites for reverse domains to be nucleated. As to their influence on the properties, thesmall precipitates will enhance the coercivity while the larger ones will reduce coercivity and may increase permeability [ 101.

C. Effect of Stacking Faults

The most common stacking faults found in spinel ferrites are the cation faults [ 111 and several of these faults can beseen in Fig. 4. The Lorentz images of an area containing these faults show that the magnetic domain walls are pinned to the faults on either side and the faulted area is a thin 180" domain. The domain walls pinned to the faults do not move easily. Also it can be seen in Fig. 4 that when a field is applied, new domains such as (a) and (b) are created at the stacking faults and the domain wall away from the faultsweeps through the unfaulted matrix easily.

B. Effect of the SecondPhase

The size and nature of the second phase determine the effect they may have on the motion of the domain wall [2]. As can be seen in Fig. 3, the nonmagnetic LiFeOz precipitatein LiFe, OB [7] can act as a barrier to domain wall motion or as a site for nucleation of a reverse domain. It is observed that if the precipitate is small (< 300 nm), it acts only as a barrier to the wall motion and if the precipitates are distributed closely, domain walls show very little mobility. For larger precipitates,

138

IEEE TRANSACTIONS ON MAGNETICS, VOL. MAG-20, NO. 1 , JANUARY 1984

Fig. 5 . Domain Wall 1-1 moves to the left or right as the field is increased or decreased (stages 2 and 3 in Lorentz images), but is pinned by pore P.

sintered with some CaO, it has been shown that the Ca rich intergranular phase strains the spinel lattice in the vicinity of the Pores and microcracks are regions where the material has a As shown in Fig. 5, pores act as boundary [ 6 ] . The present observation that the domain wall small enclosed free surface. pinning sites for domain walls in MnZnFe,04. Similarly, Fig. in such materials moves to the boundary by a sudden jump 6 shows that a crack can stop a moving domain wall. It is ob- in the vicinity of the boundary can be understood by assuming served that, unlike the case of grain boundaries, a domain wall that the stress in the spinel lattice leads to a local change in the moves continuously towards the crack until it is finally pinned. anisotropy and the magnetization vector rotates only when a Since small pores can be found in many ceramic ferrites not sufficiently large field is applied.Such a materialinevitably only at the grain boundaries, but also inside the grains, the ef- will have a lower permeability compared to materials with n o fective distance between domain wall barriers in such ferrrites intergranular phases or strain centers near the grain boundaries t61. will be reduced [ 121 . The formation of reverse domains at pores, second phases, and grain boundaries/junctions are due to the demagnetizing IV. DISCUSSION fields at these inhomogeneities. When the demagnetizing field The typical microstructure of a ceramic ferrite contains grain is large enough, aregion of the material near the inhomogenity boundaries, cracks, pores, etc. Ithas been recognized that grain is magnetized in the reverse direction. Materials with manysuch boundaries act as barriers to domain wall motion [ I ] . As the sites will typically have a lower coercive force. Since many ceramic materials,due to their processingschemes, above studyshows,the pinning behavior ofdifferent grain boundaries is different depending on whether the boundaries contain small pores, microcracks, etc., and since these features act as barriers to the domain wall motion, they must be taken have any segregates or not. In the case of (Mn, Zn)Fe,04

D. Effect of Pores and Cracks

LIN etaZ.: MICROSTRUCTURAL FEATURES IN SPINEL FERRITES

139

Fig. 6. Domain wall 1-1 moves continuously towards crack C as the field is increased. Finally walls 1-1and 2-2 annihilate each other and material is uniformly magnetized across the crack.

REFERENCES into account while studying the behavior of magnetization of the ceramic ferrites. For example, if pores, etc., are present inof permanentside the grains and not only at the grain boundaries, the inter- [ 11 J. D. Livingston,Microstructureandcoercivity magnet materials,Bogr. Mater. Sci., vol. 7, p. 243, 1981. face distance and not the grain size defines the distance over [2] A. L.Stuijts,Ceramicmicrostructures, i Proc. 6th Int. Mat. n which a wall can travel without any obstacle. Symp., R. M. Fulrath and J. A. Pask, Eds. Boulder, CO: Westview Press, 1977. The formation ofa nonmagnetic second phase can be utilized [3] H. Watanabe, S. Iida, and M. Sugimoto, Eds. Ferrites, Proc. 3rd in some ceramic systems, particularly the hard ceramic magnets, Int. Conf. on Ferrites, Center for Academic Publications, Tokyo, Japan, 1981. to enhance their coercivity. Since precipitates of varying sizes [ 4 ] M. S . Cohen, Handbook on Thin Film Technology, L. I. Maissel and distributions can be formed by manipulating the composiand R. Glang, Eds. New York: McGraw-Hill, 1970, ch. 17. tionandthermaltreatments (fromknowledge of phase and [ 5 ] P. J. Grundy and R. S . Tebble, Lorentz electron microscopy, Advances in Physics, vol. 17, p. 153, 1968. kinetic diagrams), it should be possible to manipulate the micro[ 6 ] A. H.Tsunekawa, A. Nakata, T. Kamijo, K. Okutani, R. K. Mishra, structure to control the properties. and G. Thomas, Microstructure of commercial grade (Mn, Zn) Finally it must be emphasized that the above observations Fe204,IEEE Trans. Magn.,vol. MAG-15, p. 1855,1979. [7] R. K. Mishra, 0. van derBiest, and G. Thomas,Materialsloss are qualitative. In order to make a quantitative study of doand phase transformations in LiFe50a, J. Amer. Ceram. SOC., main wall motion versus applied magnetic field, a special specivol. 61,p. 121,1978. men stage withproper field compensation [13] must bede[8] D. J. Barber, F. C. Frank, M. Moss, J. W. Steeds, and I.S.T. Tsong, Prediction of ion bombardment surface topographies using signed and then one measure magnetizing can field,etc., directly Franks kinematic theory of crystal dissolution, J. Mater. Sci., by observing the Lorentz images. Nevertheless, the qualitative vol. 8,p. 1030, 1973. observationsdescribedin thepresent papershedsomelight [9] PhilipsScientific and Analytical Equipment Service Manual PW 6002. onto the interaction of microstructural features with the domain wall. Caution must be used in extrapolating the results [ 101 R. K. Mishra and G. Thomas, Electron microscopy of ferrites, J. Amer. Ceram. SOC.,vol. 62, p. 293,1979. of these observations on thin foils to the case of bulk magnets. [ l l ] 0. van der Biest and G. Thomas, Cation stacking faults in lithACKNOWLEDGMENT Thanks are due to TDK Electronics Co. of Japan for supplying the MnZn ferrite samples.
ium ferrite spine1,Phys. Stat. Sol. (a), vol. 24, p. 65, 1974. [ 121 G. C. Jain, B. K. Das, and S . Kumari, On the origin of core losses in a manganese zinc ferrite with appreciable silica content, J. Appl. Phys., vol. 49, p. 2894, 1978. [ 131 R. Taylor, Oxford University, private communication.