3388 IEEE TRANSACTIONS ON MAGNETICS, VOL. 36, NO. 5, SEPTEMBER 2000

High Temperature Soft Magnetic Materials: FeCoAlloys and Composites

R. H. Yu, S. Basu, L. Ren, Y. Zhang, Azar Parvizi-Majidi, K. M. Unruh, and John Q. Xiao

Abstract—We have systematically investigated the micro-structural effects including grain size, precipitation, and struc-tural order parameter on the high temperature magnetic andmechanical properties of FeCo-based commercial alloys. At hightemperatures the equilibrium nonmagnetic precipitates signifi-cantly deteriorate the soft magnetic properties. Poor mechanicalproperties are mainly due to the nature of the ordered structureof FeCo alloys. Based on this understanding we have designednew magnetic composites by reinforcing FeCo alloys with highstrength fibers. The magnetic and mechanical properties canthus be improved independently through optimizing the magneticmatrix and fiber network, respectively. These new magnetic com-posites show excellent soft magnetic and mechanical properties.In particularly, negligible creep has been observed at 600C.

Index Terms—Magnetic composites magnetocrystallineanisotropy, soft magnetic materials, soft magnets.

I. INTRODUCTION

T HERE has been an increasing demand in the design ofmotors and generators for use in electric vehicles and

other applications operating at high temperature. The key issueof designing those motors and generators is to develop newsoft magnetic material with excellent magnetic and mechanicalproperties at high temperatures. The Fe–Co alloys exhibit highsaturation magnetization and high Curie temperatures

( C) that make them potential candidates forthe high temperature applications [1]–[5]. It is well knownthat Fe and Co form bcc solid solution (CoFe ) overan extensive range [6]. In the range of , FeCoalloys undergo a phase transformation from a disordered bccstructure to the ordered CsCl type structure below 730C. Theordered FeCo alloys are excellent soft magnetic materials withnegligible magneto-crystalline anisotropy [7]. However,the equiatomic FeCo alloys are extremely brittle and otherelements such as V are usually added to obtain workablematerials. The brittle nature of FeCo binary alloys is attributedto the formation of an ordered B2/L2supperlattice from thebcc solid solution below 730C. Additional elements suchas V induce precipitation of a second phase of L1structure[8], [9], whose volume percent and morphology depend onthe concentration of the added elements. The precipitation

Manuscript received February 12, 2000. This work was supported by US AirForce under Grant MURI F49620-96-1-0434.

R. H. Yu, S. Basu, Y. Zhang, K. M. Unruh, and J. Q. Xiao are with the Depart-ment of Physics and Astronomy, University of Delaware, Newark, DE 19716USA (e-mail: {rhy; sujit; kmu}@udel.edu; yzhang@physics.udel.edu).

L. Ren and A. Parvizi-Majidi are with the Department of Mechanical Engi-neering, University of Delaware, Newark, DE 19716 USA (e-mail: {lren; ma-jidi}@udel.edu).

Publisher Item Identifier S 0018-9464(00)07981-4.

of second phases, however, significantly deteriorates the softmagnetic properties [10]. Besides the second phases, all otherstructural defects such as grain boundaries [5] and dislocationsalso considerably affect the magnetic properties especially thecoercive field .

In this paper, we report a comprehensive and systematic studyto understand the microstructural effects on the magnetic andmechanical properties of FeCo-based alloys. The effects due tograin size, order parameter, and precipitation have been succes-sively separated and analyzed. Based on these studies, we havedeveloped new magnetic composites by reinforcing FeCo mate-rials with fiber or ceramics particles. In the case of FeCo-fibercomposites, the contributions to magnetic and mechanical prop-erties separately come from two interlaced entities, i.e. FeComatrix and fiber network, respectively. One immediate advan-tage of such materials is that we can optimize magnetic and me-chanical properties in each nearly independent entity and thusno comprise between the magnetic and mechanical propertiesis needed, which is always the case in conventional soft mag-netic materials, i.e. the improvement of mechanical propertiesis often achieved at the expense of magnetic properties and viceversa.

II. EXPERIMENTAL PROCEDURES

Commercial FeCo-based alloys were provided by CarpenterTechnology, Inc. Different grades of samples were used and canbe grouped into two compositions. One contains equiatomic Feand Co in addition to V and other added element, Fe49Co2V andFe49Co2V0.3Nb. The other group contains 27 at.% Co and bal-anced Fe with small amount of C and other elements, Fe27Co.The rolled 0.25 mm thin sheets were cut into samples with dif-ferent shapes and sizes before the heat treatment. All the sam-ples were heat-treated at 820C for 2h then cooled at varyingcooling rates in a flowing H atmosphere. The heat treatmentprovides samples with varying properties and microstructuralfactors.

Fiber or dispersion reinforced magnetic composites were fab-ricated using electrochemical deposition from an aqueous bathcontaining CoSO H O, FeSO H O, HBO , and saccharine[11]. The Al O particles of a diameter m was directlymixed into the solution. Sulfuric acid was added to the bath toadjust the pH value to about 2.0.

Toroidal and long stripe samples were used to measure mag-netic properties using a magnetic loop tracer. Grain size and pre-cipitates were characterized by optical microscopy, SEM andTEM. The magnetic domain structure was observed by Lorentzmicroscopy. Structural order parameters were determined from

0018–9464/00$10.00 © 2000 IEEE

YU et al.: HIGH TEMPERATURE SOFT MAGNETIC MATERIALS 3389

Fig. 1. The change of DC and AC coercivity with grain size for the orderedsample with Fe49Co2V measured at room temperature. The inset shows the plotof the coercivity vs inverse grain diameter for the same sample.

the ratio of integrated intensities of of neutron diffrac-tion patterns. The Vicker microhardness (HV) of the magneticcomposites was measured using a load of 50 g for 10 s. Ten-sile tests were performed using an Instron mechanical testingmachine.

III. RESULTS AND DISCUSSION

Fig. 1 illustrates the relationship between coercivity andgrain size with minor effect from other structural parameters.For all types of samples, regardless of order parameters andcompositions, coercivity isuniversallyfound inversely propor-tional to the grain diameter even at a frequency up to 1600 Hz.A similar relationship between the coercivity and grain sizein DC case has also been found in FeNi [12] and MoCu alloys[13]. According to Mager [14], the coercivity determined bygrain boundaries can be expressed as:

(1)

where is the wall energy, and is the saturation magneti-zation. The wall energy can be estimated by the equation:

(2)

thus

(3)

whereis the Boltzmann constant,is the magneto-crystalline anisotropy,is the Curie temperature, andis the lattice constant.

The smaller grain size and higher values result in a highercoercivity for disordered samples as compared with the orderedsamples. Fitting Fig. 1 with the Eq. (3), the value of forisothermally annealed samples with Fe49Co2V is estimated tobe about J/cm , which is very close to experimentalresults published previously ( J/cm ) [15]. Amore detailed analysis of the relationship between the coercivityand grain boundaries can be obtained by using more accuratemodel proposed by Chui [16].

Fig. 2. Lorenz microscopic photograph showing magnetic domain (md) wallspinned by grain boundaries (gb). (a) in-focus image, and (b) Foucault image.

The linear dependence of as a function of impliesthat grain boundaries serve as pinning sites for the magneticdomain wall. The pinning effect of the grain boundaries on themagnetic domain wall has been investigated by Lorentz electronmicroscopy. Lorenz electron microscopy observations revealedthat some magnetic domain walls lie along the grain bound-aries and are curved at intersecting points of grain boundaries,indicating that grain boundaries act as pinning sites for mag-netic domain wall movement. Fig. 2 shows a typical photographdemonstrating that the domain walls are pinned by grain bound-aries indicated by an arrow.

Fig. 3 shows the variation of the yield stress with grain sizefor Fe49Co2V and Fe27Co alloys. The yield stresslinearlydepends on , i.e. follows the well-known Hall–Petchrelationship:

(4)

whereis the yield stress,is the grain size, and

and are constants.The following relationships are obtained for Fe27Co andFe49Co2V, respectively:

for Fe27Co (5)

and

for Fe49Co2V (6)

3390 IEEE TRANSACTIONS ON MAGNETICS, VOL. 36, NO. 5, SEPTEMBER 2000

(a)

(b)

Fig. 3. The effect of grain size on the yield stress for (a) Fe27Co, and(b) Fe49Co2V alloys.

Fig. 4. Change of volume percent (a) and diameter (b) of the precipitates withaging time for two alloys.

In addition, we have also found that the Fe49Co2V alloy hasa higher yield stress in the disordered state than in the orderedstate, which is consistent with previous results [17].

Aging these FeCo alloys at 600C yields the phaseprecipitates whose size and volume percent increase withaging time. Fig. 4 illustrates the change of the volume percentand size of the precipitates with aging time for Fe49Co2V

Fig. 5. Variation of magnetic properties with aging time at 600C.

and Fe49Co2V0.3Nb alloys aged at 600C in a flowing Aratmosphere. TEM observation indicates that most of thepre-cipitates are globular particles, with twins and defaults insidethe particles. The crystalline orientation relationship betweenthe matrix and the precipitates areand . From X-ray diffraction spectra, thelattice parameters of the matrix andphases were determinedto be 2.85 and 3.58 Å, respectively. The detailed analysissuggests that the precipitates correspond to a phase of(FeCo) M (M-added elements), i.e., one of the four interpen-etrating simple cubic sublattices with Llstructure. Ashbyetal. [8] have suggested the same structure for the precipitatesin Fe Co V samples. The size and density of thephasesare much higher in alloys with Nb additions, suggesting thatthe Nb addition promotes the nucleation and growth of thephase. Pittet al. [9] found that the addition of Ni increases theprecipitation rate of the phases, and the precipitates areNi-rich phase. EDX analyzes suggest thatprecipitates maybe also Nb-rich phase.

The change of the magnetic properties with aging timeis shown in Fig. 5. The soft magnetic properties deterioratesignificantly with aging time. The influence of a second phaseprecipitation on the coercivity has been treated theoretically inprevious investigations. According to Néel [18] a linear relationbetween and the volume percent of the second phase ()as a rough approximation should be valid, resulting from thepinning of magnetic domain wall due to precipitates. Dijkstraet al. [19] proposed that the pinning is most effective when thediameter of the precipitatesis of the order of the Bloch-wallthickness , resulting in substantial increase in the magneticcoercivity . For and , the effect becomesless pronounced. A considerable increase in the coercivityis seen for the samples annealed at 600C for less than 10 h.We have measured the size of the precipitates which is about0.01–0.3 m that is comparable to the magnetic domain wallthickness (about 0.26m).

YU et al.: HIGH TEMPERATURE SOFT MAGNETIC MATERIALS 3391

(a)

(b)

Fig. 6. Neutron diffraction patterns for the disordered (top) and ordered(bottom) samples, respectively.

(a)

(b)

Fig. 7. Change of magnetic properties with order parameterS. (a) CoercivityH , and (b) SaturationM .

The equiatomic FeCo alloys undergo the disordered to or-dered phase transformation at about 730C. The high temper-ature disordered phase can be quenched to room temperaturewithout order. By changing the quenching rate samples with dif-ferent long range structural order parameters,can be readilyobtained. The order parameter can be determined from neutrondiffraction study based on the formula described by Smith andRawling [20]. Fig. 6 shows typical neutron diffraction patternsfor the disordered and ordered samples for Fe49Co2V alloys.

The relationship between magnetic properties and order pa-rameter is shown in Fig. 7. The plots were generated after theeffects due to grain size were extracted. The increase in satura-tion magnetization with order parameter was also foundpreviously [21]. Coercivity is found to decrease with in-creasing order parameter. Microstructural analyzes indicate thathigh density of dislocations is always associated with less or-dered structure that may be the reason why samples with disor-dered structure show higher than ordered samples.

Fig. 8. The relationship between the yield stress and order parameter forFe49Co2V alloys.

Fig. 9. Magnetic hysteresis loops for the as-deposited and annealedFe Co –W samples with a diameter of 50�m. The diameter of the fiber isabout 12�m.

Fig. 8 shows the relationship between the yield stressandorder parameter . The results indicate that the yield stress in-creases with decreasing degree of order. An ordered structurecontains superdislocations, which consist of two unit disloca-tions separated by a strip of anti-phase boundaries (APB). Theenergy of the APB, , is dependent on the degree of longrange order , by , where is the en-ergy of the APB for completely ordered structure, i.e. at[22]. So, the lower the order parameter, the lower the energy ofAPB. At low value, the energy of the APB is so weak that thesuperdislocations dissociate into their constituent ordinary dis-locations, which then can move independently. These glidingunit dislocations will create wrong bonds in their wake thusleading to a hardening. On the other hand, whenis increased,unit dislocations more tend to associate into pairs, namely su-perdislocations, which glide in an ordered matrix without cre-ating a trail of wrong bonds. The yield strength thus decreasesas the proportion of superdislocation increases. As the order pa-rameter decreases the yield strength is then controlled by shortrange order. The higher the short-range order parameter, thehigher the yield strength.

Based on above studies, we have designed and fabricatedfiber reinforced soft magnetic composites FeCo–W andFeCo–C. Pure FeCo alloys without additional elements showexcellent high temperature magnetic properties since thereis no precipitation, and the inclusion of fibers offers supe-rior mechanical properties. Fig. 9 shows room temperaturemagnetic hysteresis loops for the as-deposited and annealed

3392 IEEE TRANSACTIONS ON MAGNETICS, VOL. 36, NO. 5, SEPTEMBER 2000

Fig. 10. The mechanical strengths for the annealed FeCo –W compositesas a function of W fiber volume percent.

Fe Co –W composites. In the as-deposited state, the samplesare not magnetically soft and show low magnetic permeabilityand coercivity of about 16 Oe [Fig. 9(a)], whereas the annealedsamples show good soft magnetic properties [Fig. 9(b)].The improvement of soft magnetic properties is related tomicrostructural change during thermal annealing process.First, the grain sizes increase from about 50 nm to 200 nmafter heat-treatment, which reduces coercivity as previouslydiscussed. In addition, thermal annealing relieves the internalstress in the samples induced during the electrodepositionprocess. This internal stress which has caused radial dependentpermeability, results in giant impedance effects in as-depositedsample [23]. There are also other microstructural changes suchas an increase in the structural order parameter and a decrease inthe defects, which further enhance the soft magnetic properties.

By varying the relative concentration of Fe and Co ions inthe deposition bath, FeCo alloys of the whole compositionrange can be readily fabricated. X-ray diffraction experimentsindicate that the FeCo –W samples are of bcc structure at

at.%. For at.%, an hcp structure graduallydevelops, A single hcp structure is observed at .

Significant improvement of mechanical properties has beenobserved in these composites, as shown in Fig. 10 the mechan-ical strength as a function of the W fiber volume percent forthe annealed FeCo –W samples. Both yield strength andtensile strength increase linearly with fiber volume percentas seen in other common composite materials [24]. In oursample geometry, the stress is applied on matrix material andtransferred to W-fiber through the interface, the observation ofsuch a composite behavior suggests good adhesion betweendeposited materials and W fibers. Magnetic measurementsillustrated that FeCo thickness does not significantly alter themagnetic properties of the samples. More significantly, asshown in Fig. 11, a negligible high temperature creep (600C)has also been seen in FeCo–W composites in contrast withcommercial FeCo-based alloys in which substantial creep takesplace after about 50 hours, causing detrimental failures at600 C operating temperature.

To further enhance the mechanical properties of the soft mag-netic materials, fine AlO particles were uniformly dispersedin FeCo alloys during deposition. Fig. 12 shows the dependenceof on the volume percent of AlO particles of diameter of37 nm for as-deposited and annealed FeCo(AlO )–W samples.A reduction in coercivity is seen after thermal annealing, butthe heat treatment does not influence the trend of coercivity

Fig. 11. High temperature creep for FeCo-composites in comparison withcommercial FeCo bulk alloy.

Fig. 12. Relationship between the coercivity and AlO content in theas-deposited and annealed FeCo –W samples.

Fig. 13. Vicker hardness versus the cube root of the volume percent of AlOcontent for the as-deposited and annealed FeCo –W samples.

as a function of the AlO content. Kersten [26] developed atheory to explain the effect of inclusions on the magnetic prop-erties. In his model, he assumed that spherical nonmagnetic par-ticles of average radiuswere uniformly imbedded at the cornerof unit cells of an adopted simple cubic lattice with lattice con-stant . In addition the size of inclusions is assumed to be com-parable to the domain wall thickness. The coercivitycan beexpressed as:

(7)

where is the volume percent of the inclusions. This linearrelationship between and has indeed been observed asshown in Fig. 12. Using , where is theBoltzmann constant, we estimated the domain wall thickness forFe Co alloy is about m which is comparable toAl O particle size.

Fig. 13 shows the Vicker hardness of FeCo–AlO com-posites as a function of AlO content. About 100% increase

YU et al.: HIGH TEMPERATURE SOFT MAGNETIC MATERIALS 3393

in hardness value was found for FeCo samples with 12 vol.%Al O particles in comparison with pure FeCo sample. Areduction of the hardness in the annealed samples is due tolarge grain size and relief of the internal stress. A roughlylinear relationship between the hardness and the cube root ofthe volume percent of the dispersed AlO phase has beenobserved for both as-prepared and annealed samples. Disper-sion hardening commonly follows the Orowen–Ashby model[27]. For the case of noncoherent spherical particles of size,Orowen [27] proposed the mechanism where the yield stressis determined by the shear stress required to bow a dislocationline between two particles separated by a distance. Theincrease in yield stress is given by:

(8)

where is elastic modulus, is Buger vector of a dislocation.For dispersion reinforced composites, the spacingbetweenparticles can be estimated using: . The hard-ness as a function of is shown in Fig. 13. The linear rela-tionship between hardness and suggests that the dispersionhardening in our samples follows Orowen–Ashby model.

IV. CONCLUSION

We have systematically investigated high temperaturemagnetic and mechanical properties of FeCo alloys. The mi-crostructural factors such as grain size, precipitates, defects, andstructural order parameter significantly change both magneticand mechanical properties in FeCo-based alloys, especiallyin samples with small additions such as V and Nb. We haveexperimentally found that grain boundaries, precipitates, anddisordered phases act as pinning sites for the magnetic domainwall movement, which increase the coercivity , whereasthe precipitates and order parameters alter the saturationmagnetization . The precipitation of second phase createsmost adverse effects on magnetic properties. The mechanicalproperties of bulk commercial FeCo-based alloys are poor yieldstrength, ductility, and creep behavior at high temperatures.

To significantly improve the mechanical properties withoutconsiderably sacrificing the magnetic properties, we havedesigned and fabricated fiber and ceramic particle reinforcedsoft magnetic composites. Excellent mechanical propertiesincluding high yield strength and low creep rate have beenobserved in the fabricated fiber composites. These magneticcomposite materials show promising potential in applicationsand open a new avenue for the exploration of the interestingphysical phenomena.

ACKNOWLEDGMENT

The authors would like to thank Dr. M. Chen and Prof. W.Yelon of the University of Missouri for the neutron diffractionstudy, and Dr. L. Li and S. Masteller of Carpenter Technologyfor providing samples.

REFERENCES

[1] F. Pfeifer and C. Raddeloff, “Soft magnetic Fe–Ni and Fe–Co Al-loys—Some physical and mettallurgical aspects,”J. Magn. Magn.Mater., vol. 19, pp. 190–207, 1980.

[2] R. H. Yu, S. Basu, Y. Zhang, and J. Q. Xiao, “Magnetic domain andcoercivity in FeCo soft magnetic alloys,”J. Appl. Phys., vol. 85, pp.6034–6036, 1999.

[3] L. Li, “High temperature magnetic properties of 49%Co–2%V–Fealloy,” J. Appl. Phys., vol. 79, pp. 4578–4580, 1996.

[4] R. V. Major and V. Samadian, “Physical metallurgy and properties of anew high saturation Co–Fe alloy,”J. Mater. Eng., vol. 11, 1989.

[5] R. H. Yu, S. Basu, Y. Zhang, A. Parvizi-Majidi, and J. Q. Xiao, “Pinningeffect of the grain boundaries on magnetic domain wall in FeCo-basedmagnetic alloys,”J. Appl. Phys., vol. 85, pp. 6655–6659, 1999.

[6] T. Nishizawa and K. Ishida, “Fe–Co phase diagrams,”Met. Progress,vol. 57, 1986.

[7] R. M. Bozorth,Ferromagnetism. New York: IEEE, 1991.[8] J. A. Ashby, H. M. Flower, and R. D. Rawling, “Gamma phase in an

Fe–Co–2%V alloy,”Metal Science, pp. 91–96, March 1977.[9] C. D. Pitt and R. D. Rawling, “Microstructure of Fe–Co–2V and

Fe–Co–V–Ni alloys containing 1.8–7.4 wt.%Ni,”Metal Science, vol.15, pp. 369–376, August 1981.

[10] R. H. Yu, S. Basu, Y. Zhang, and J. Q. Xiao, “Precipitation and hightemperature magnetic properties of FeCo-based alloys,”J. Appl. Phys.,to be published.

[11] R. H. Yu, L. Ren, S. Basu, K. M. Unruh, A. Parvizi-Majidi, and J. Q.Xiao, “Novel soft magnetic composites fabricated by electrodeposition,”J. Appl. Phys., to be published.

[12] E. Adler and H. Pfeifer,IEEE Trans. Magn., vol. 10, p. 172, 1974.[13] F. Pfeifer and W. Kunz,J. Magn. Magn. Mater., vol. 4, p. 214, 1977.[14] A. Mager,Ann. Phys., vol. 6F11, p. 15, 1952.[15] R. V. Major and C. M. Orrock,IEEE Trans. Magn., vol. 24, p. 1856,

1988.[16] S. T. Chui, “Private communication,” unpublished.[17] L. Zhao and I. Baker, “The effect of grain size and Fe:Co ratio on the

room temperature yielding of FeCo,”Acta Metall. Mater., vol. 42, pp.1953–1958, 1994.

[18] L. Néel, , Ann Univ. Grenoble, vol. 22, p. 299, 1946.[19] L. J. Dijkstra and C. Wert,Phys. Rev., vol. 79, p. 979, 1959.[20] A. W. Smith and R. D. Rawling,Phys. Stat. Sol. (a), vol. 34, pp. 117–123,

1976.[21] C. W. Chen,J. Appl. Phys., vol. 32, p. 3485, 1961.[22] N. S. Stoloff and R. G. Davis, “The plastic deformation of ordered FeCo

and FeAl alloys,” Acta Met., vol. 12, pp. 473–485, 1964.[23] R. H. Yu, G. Landry, S. Basu, and J. Q. Xiao, “Magneto-impedance ef-

fect in soft magnetic tubes,”J. Appl. Phys., to be published.[24] A. G. Metcalfe,Interfaces in Metal Matrix Composites. New York,

NY: Academic Press, 1974.[25] R. F. Finger, “Private communication,” unpublished.[26] C. W. Chen,Magnetism and Metallurgy of Soft Magnetic Materials:

North-Holland, 1977, p. 132.[27] G. E. Dieter,Mechanical Metallurgy: McGraw-Hill, 1986, p. 212.