ARTICLE IN PRESS

Journal of Magnetism and Magnetic Materials 322 (2010) 342–347

Contents lists available at ScienceDirect

Journal of Magnetism and Magnetic Materials

0304-88

doi:10.1

� Corr

Glasgow

fax: +4

E-m

journal homepage: www.elsevier.com/locate/jmmm

Crystallisation progress in Si-rich ultra-soft nanocomposite alloy fabricatedby melt spinning

Duc-The Ngo a,c,�, Mohamed Sultan Mahmud b, Hoang-Hai Nguyen c, Hong-Gam Duong c,Quang-Hoa Nguyen c, Stephen McVitie a, Chau Nguyen c

a Department of Physics and Astronomy, University of Glasgow, Glasgow G12 8QQ, UKb Department of Physics, University of Asia Pacific, Dhaka-1209, Bangladeshc Center for Materials Science, College of Science, Vietnam National University Hanoi, 334 Nguyen Trai Road, Hanoi, Vietnam

a r t i c l e i n f o

Article history:

Received 16 July 2009

Received in revised form

7 September 2009Available online 20 September 2009

PACS:

75.50.Kj

75.50.Tt

75.50.Bb

70.40.Gb

Keywords:

Amorphous alloys

Nanocrystalline materials

Permeability

Finemet

53/$ - see front matter Crown Copyright & 2

016/j.jmmm.2009.09.054

esponding author at: Department of Physics

, Glasgow G12 8QQ, UK. Tel.: +44141339 88

41413304464.

ail address: d.ngo@physics.gla.ac.uk (D.-T. Ngo

a b s t r a c t

The crystallisation process and the ultras-soft magnetic properties of amorphous/nanocomposite alloy

Fe73:5Si17:5B5Nb3Cu1 fabricated by conventional melt-spinning technique are systematically investigated

in terms of thermal analysis and in-situ measurement of magnetisation dynamics. The thermal analysis

using differential scanning calorimetry showed that crystallisation from Fe-based amorphous state to a-

Fe(Si) started at 535 3C. Further heating the sample leads to a transformation from the a-Fe(Si) to Fe-B

phases at 670 3C. Crystallisation activation energies were determined using two models: Kissinger and

John–Mehl–Avrami (JMA), which were consistent to each other with a value of 2:8170:03 eV. High

resolution transmission electron microscopy investigation revealed an ultrafine structure of a-Fe(Si)

nanocrystallite with mean size of 12.5 nm embedded in an amorphous matrix. At a volume fraction of

86% of a-Fe(Si) phase, optimum soft magnetic properties were obtained with very high permeability of

110,000 and a very low coercivity of 0.015 Oe by annealing the amorphous alloy at 530 3C in 40 min.

Crown Copyright & 2009 Published by Elsevier B.V. All rights reserved.

1. Introduction

The Fe73.5Si13.5B9Nb3Cu1 amorphous/nanocomposite alloy (FI-NEMET) has been extensively studied by many world-wideresearchers [1–4] due to its excellent soft magnetic properties.Commonly, amorphous alloy is firstly produced as a raw material,and the nanocomposite alloy will be subsequently obtained byappropriate heat treatment for crystallising the nanocrystallite a-Fe(Si) with suitable volume fraction. The microstructure and softmagnetic properties of Fe-based nanocomposite alloys are thusstrongly influenced by heat treatment [1,3,5].

The optimum structure of the FINEMET consists of an ultrafinestructure of a-Fe(Si) nanocrystallites embedded in remainingamorphous matrix [2]. A suitable volume fraction of a-Fe(Si)nanocrystallites in the materials leads to compensating the netmagnetostriction of two phases (positive magnetostriction ofamorphous and negative magnetostriction of nanocrystalline

009 Published by Elsevier B.V. All

and Astronomy, University of

55x0895;

).

a-Fe(Si) grains). Furthermore, since the size of nanocrystallites issmaller than the ferromagnetic exchange interaction length(about 35 nm for FeSiBNbCu amorphous/nanocomposite system),ferromagnetically exchange coupling between the nanocrystallinea-Fe(Si) grains through amorphous matrix results in an averagingout of the magnetocrystalline anisotropy [5]. As a result, ultrasoftmagnetic properties with high saturation induction, very highpermeability, low frequency losses and low coercivity areobtained [2,3].

Understanding the crystallisation kinetics in these materials isof scientific interest for two crucial reasons. Firstly, for the case ofalloys that exhibit excellent magnetic properties in the amor-phous phase, the crystallisation kinetics represents a limit atwhich these properties begin to deteriorate. Therefore, thermalstability determines the magnetic stability of the materials in theamorphous phase. Secondly, for the case of alloys that exhibitexcellent magnetic properties in the nanocrystalline/amorphousmatrix structure, controlling the crystallisation kinetics providesan ability to tailor the microstructure. The amount of thenanocrystalline phase formed within the matrix can be controlledto achieve the desired magnetic performance. Additionally,models of the crystallisation from the metastable amorphousstate to the stable crystalline state depend on various parameters

rights reserved.

ARTICLE IN PRESS

D.-T. Ngo et al. / Journal of Magnetism and Magnetic Materials 322 (2010) 342–347 343

such as the composition, the concentration of nucleation sites, thediffusion coefficients, the activation energy for diffusion, etc.Investigation of crystallisation kinetics will be fruitful for funda-mental understanding of thermal behaviours and for optimisingproducing process in industrial production.

In this article, the crystallisation evolution as well as theinfluence of heat treatment on the magnetic properties of Si-richFe73.5Si17.5B5Nb3Cu1 alloy fabricated by rapid quenching techniqueis investigated.

2. Experiments

Ingot alloy with nominal composition of Fe73.5Si17.5B5Nb3Cu1

was initially fabricated by induction melting in high vacuum. Thesample is slightly modified from pure-Finemet as enriching Si

content (from 13.5 to 17.5 at%) and lowering B content (from 9 to5 at%). The amorphous ribbons were subsequently produced byrapidly quenching molten alloy on surface of a copper wheelrotating with longitudinal speed of 30 m s�1. The estimatedcooling rate is about 106 K s�1, which is high enough to ensureamorphous state in the ribbons. The ribbons are in 1 cm width and16mm thick. The ribbons were isothermally annealed in Ar

atmosphere at various temperatures ranging from 530 to 580 3Cfor nanocrystallisation. Thermal analysis was performed using aSDT 2960 TA Instruments differential scanning calorimeter (DSC).Microstructure of the sample was examined using a FEI TecnaiTF20 field emission gun transmission electron microscope (TEM)with an accelerated voltage of 200 kV. Crystal structure of thesample was determined by X-ray diffraction (using a BrukerD5005 diffractometer) and electron diffraction on the TEM.Magnetic properties were characterised using a DMS 880 vibrat-ing sample magnetometer (VSM) and an AMH-401A Hysteresis-grapher.

3. Results and discussion

X-ray diffraction and electron diffraction in TEM confirm thefully amorphous state in as-quenched ribbons. Thermal analysisresults performed on the DSC are shown in Fig. 1. The DSC curvesexhibit two clearly exothermal peaks, corresponding to the

Fig. 1. Differential scanning calorimetric curves of the as-quenched ribbon

performed at various heating rates from 10 to 50 K min�1.

crystallisation stages during heating progress. It is well-knownthat the first peak, occurring around 530 3C (so-called onsetcrystallisation temperature) corresponds to the crystallisation ofcrystalline a-Fe(Si) phase from amorphous phase [2,3,5]. It isimportant to note that the onset temperature in the first peak is alittle lower than that of pure Finemet with lower content of Si

[1–3]. Previous studies [2,3] suggested that the second peak(680 3C) related to the formation of a boride phase (Fe–B). A smallamount of boride phase Fe3B was found in the specimen heated to680 3C, confirming the transformation from a-Fe(Si) to Fe3B.

As seen from Fig. 1 the exothermal peaks on DSC curves slightlyshifts to higher temperature with respect to heating rate.According to Kissinger model [6], the temperature at exothermalpeak, Tp, is dependent on the heating rate, b, as follows:

lnbT2

p

¼ �Ea

kBTpþconst: ð1Þ

where b; Tp; Ea are heating rate, exothermal temperature, andcrystallisation activation energy, respectively; and kB is Boltzmannconstant. Fig. 2 shows the dependence of lnðb=T2

p Þ on the inversedexothermal temperature, 1=Tp (Kissinger plot). The lineardependence of the lnðb=T2

p Þ on the 1=Tp confirms that theKissinger model is reasonable for describing the crystallisationin the studied sample. As a result, the activation energy in thiscase can be easily determined by linear fitting the dependence inFig. 2, and shown to be Ek

a ¼ 2:8070:4 eV. This value is calculatedfor the crystallisation at the first peak—for a-Fe(Si) phase.

Crystallisation of the a-Fe(Si) phase at the first peak is alsoconfirmed by using thermomagnetic measurement (Fig. 3) on theVSM. Firstly, the temperature dependence of the magnetisation ofan amorphous specimen was measured by heating the specimenfrom room temperature to 780 3C using flowing Ar gas on the VSM.During the progress, a small magnetic field of 20 Oe is applied tothe sample to detect the magnetisation of the specimen. Thisprogress can be described as follows:

�

Figam

Firstly, the magnetisation rapidly decreases at the Curietemperature of the amorphous phase, TC ¼ 327 3C and thenremains at a small value because the sample is in aparamagnetic state.

� Further increasing temperature to onset crystallisation tem-

perature, the magnetisation drastically increases because of

. 2. Kissinger plot for determining crystallisation activation energy for

orphous specimens.

ARTICLE IN PRESS

Fig. 3. Temperature dependence of magnetisation of the as-quenched specimen

measured during heating and cooling cycles.Fig. 4. Volume fraction of the a-Fe(Si) phase as a function of annealing time (at

530 3C annealing temperature). The inset shows an in situ measurement of

magnetisation dependent on the time at 530 3C in 10 kOe applied field.

D.-T. Ngo et al. / Journal of Magnetism and Magnetic Materials 322 (2010) 342–347344

the formation of the ferromagnetic a-Fe(Si) phase withincreasing volume fraction during the heating process.

� When temperature is continuously increased near the Curie

temperature of the a-Fe(Si) phase, a reduction of themagnetisation is observed.

� In the cooling cycle, a single-phase temperature dependence of

magnetisation is visible because of the unique exchange-coupled behaviour in multiphase-structure a-Fe(Si)/amor-phous system.

Additionally, the Johnson–Mehl–Avrani (JMA) model [7,8] wasalso applied to clarify the crystallisation kinetics during heattreatment process. The JMA model can be expressed that:

�

Nucleation and growth occur at a constant temperature, e.g. atthe isothermal crystallisation temperature. � Nucleation is random throughout the bulk of the sample,

which is assumed to be infinite.

� Growth is isotropic until crystals impinge upon one another.

As a result, the volume fraction of the a-Fe(Si) phase, Xf ðtÞ

depends on the annealing time, t, by the expression:

Xf ðtÞ ¼ 1� e�ðktÞn ð2Þ

where k is the rate coefficient and n is the morphology index. Therate coefficient, which is a temperature-dependent factor, is givenby

kðTÞ ¼ k0e�Ea=RT ð3Þ

Here, Ea;R; T are, respectively, crystallisation activation energy,ideal gas constant and the temperature.

The volume fraction of the a-Fe(Si) phase was determinedusing Leu and Chin method [2,3,9] by comparing DSC curves of as-quenched and annealed samples at a sample heating rate.

Fig. 4 illustrates the volume fraction of the a-Fe(Si) phase as afunction of annealing time as isothermally annealed at 530 3C. Thedata shows that the dependence of the volume fraction on thetime is well consistent with JMA model described in Eq. (2).Hence, values of k¼ 1:04� 10�3 s�1 and n¼ 1:12 are calculatedfor rate coefficient and morphology index, respectively.

Repeating this process for specimens annealed at temperaturesof 540, 550, 560, 570, and 580 3C, the crystallisation activation

energy is again estimated. The activation energy in the JMA modelis 2:8370:03 eV, which is closed to that obtained by the Kissingermodel mentioned above. Furthermore, it is obviously seen that ahigher Si content results in a lower crystallisation activationenergy in comparing with original Finemet (3.2–3.4 eV [3,10,11]).This is crucial for tailoring the magnetic properties of thematerials. The lower activation energy, the lower annealingtemperature is required, and as a result the better properties ofthe materials (soft magnetic, mechanical, etc. properties) shouldbe remained. For example, using lower annealing temperature, itis indeed that crystallisation rate would be slow down, and bettergrain structure (finer structure, more homogeneous structure,etc.) could be controlled. On the other hand, a lower annealingtemperature could prohibit the hardness of the alloys, and keepthe flexibility of the annealed alloy. Furthermore, an enhancementof Si and a reduction of B content will prohibit the hardness ofannealed alloy [11], an important feature for using in variouspurposes. By the same ways, activation energy of phase transfor-mation at the second peak on the DSC curve is also determined bythe same methods and shown to be 4:770:07 eV.

The magnetisation evolution at onset crystallisation tempera-ture, 530 3C, is depicted in the inset of Fig. 4. In this measurement,an amorphous specimen was heated to 530 3C and kept at thattemperature. Time dependence of the magnetisation was mea-sured in an applied field of 10 kOe, which is high enough tosaturate the sample. The result in the inset of Fig. 4 shows that thetime dependence of the isothermal magnetisation is again inagreement with the JMA model given by Eq. (2). It is well-knownthat the saturation magnetisation in the specimen is the totalmagnetisation of two phases: the crystalline a-Fe(Si) phase andamorphous phase, which is given by [5]

M¼ Xf MFeðSiÞ þð1� Xf ÞMamor ð4Þ

At high temperature, the Mamor � 0, therefore the total magnetisa-tion is approximately proportional to the volume fraction of thea-Fe(Si) phase:

M� Xf MFeðSiÞ ¼MFeðSiÞ½1� e�ðktÞn � ð5Þ

Eq. (5) expressed the time dependence of the magnetisation asshown in the inset of Fig. 4. Grain structure of the materials wasdetermined by a transmission electron microscope using dark-

ARTICLE IN PRESS

Fig. 5. Bright-field TEM micrographs of the specimens annealed at 530 3C (a), 560 3C (b), 580 3C (c) (in 40 min keeping time). The frame (d) shows the temperature

dependence of grain size. The inset shows selected area electron diffraction pattern, and dark shading on the images is due to the thickness variation of the TEM specimens.

D.-T. Ngo et al. / Journal of Magnetism and Magnetic Materials 322 (2010) 342–347 345

field and bright field imaging. Fig. 5 shows a series of bright fieldTEM micrographs of the samples annealed at varioustemperatures. No grain structure was observed in as-quenchedspecimen due to the amorphous structure. By annealing, the grainstructure of the a-Fe(Si) phase was formed. Increasing annealingtemperature results in an increase of grain size. Namely, at 530 3C,mean grain size is 12.5 nm, and rising to 14.0 nm when annealingtemperature is 560 3C. And at 580 3C annealing temperature,average grain size becomes 37.2 nm (see detail in Fig. 5d).Diffraction analysis in TEM confirms a polycrystalline structureof bcc-Fe(Si) phase in the materials. As increasing annealingtemperature from 530 to 580 3C, the lattice constant slightlydecreases from 0.299 to 0.285 nm. This indicates that increasingannealing temperature enhances the concentration of Si in thebcc-Fe(Si) phase. Namely, at 530 3C, lattice constant ofa¼ 0:299 nm of bcc-Fe(Si) phase corresponds to the bcc-Fe87.6Si12.4 phase [12]. Lattice constant decreases to 0.285 nm at580 3C confirming the existence of the bcc-Fe85.5Si14.5 phase [12].

High resolution imaging of transmission electron microscopy(HRTEM) confirms the nanostructure of the nanosized bcc-Fe(Si)single crystals embedded in remaining amorphous matrix (Fig. 6).At high annealing temperature, some lattice dislocations wereobserved in the bcc-Fe(Si) crystallites. This suggests that highannealing temperature is not suitable for creating a perfectmicrostructure of the materials, which directly influences on themagnetic properties of the samples.

Fig. 7 displays hysteresis loops of the studied sample beforeand after heat treatment. Amorphous as-quenched specimenexhibits a highly rectangular-shape hysteresis loop indicating that

the magnetisation reversal is governed by pinning of domain walldisplacement because of high mechanical stress in amorphousstructure. A quite large coercivity, Hc ¼ 0:20 Oe and low maximumpermeability, mmax ¼ 6500 are measured for as-quenchedspecimen. In the annealed specimens, the mechanical stress isreasonably reduced because of the formation of nanostructure ofbcc-Fe(Si) crystallites, resulting in a narrow S-shape hysteresisloop of domain wall movement of magnetisation reversal. Theformation of the bcc-Fe(Si) phase with appropriate volumefraction leads to satisfying the criteria for the formation ofultrasoft magnetic properties. Excellent soft magnetic propertiesare indeed obtained as shown in Fig. 8 with a very small coercivityof 0.013 Oe, a very high maximum permeability up to 110,000 atoptimum annealing condition (530 3C for 40 min). These are betterthan other ultrasoft Finemet-based alloys studied previously[1–4]. As seen from Fig. 8, as increasing the time from 5 to40 min, the volume fraction of the a-Fe(Si) phase raises from 18%to 86% leading to rapid reduction of net saturation magneto-striction of nanocomposite system due to the combination of twophases, whereas the particle size increases very slowly. This effectleads to decreasing the coercivity from 0.20 Oe (for as-castspecimen) to 0.015 Oe (annealing in 40 min). In longer annealingtime (50 and 60 min), the higher volume fraction and large grainsize of the a-Fe(Si) phase make an increase of coercivity. Contraryto the variation of the coercivity, as increasing annealingtemperature, the permeability increases and goes through amaximum peak and finally decreases when the keeping time istoo long (see Fig. 8). It is noticed that very high permeability (adetail of magnetic characteristics is also shown in Table 1) was

ARTICLE IN PRESS

Fig. 6. HRTEM micrographs of the specimens annealed at 530 3C (a) and 580 3C (b) (in 40 min keeping time).

Fig. 7. Hysteresis loops of studied sample: as-quenched specimen and optimally

annealed specimen.

Fig. 8. Magnetic characteristics coercivity and maximum permeability as a

function of annealing time as annealed at 530 3C.

Table 1Volume fraction, coercivity (Hc), initial permeability (mi), maximum permeability

(mmax) and saturation induction ðBsÞ as a function of annealing temperature

(annealed in 40 min).

As-

quenched

530 3C 540 3C 550 3C 560 3C 570 3C 580 3C

vol% Fe(Si) 0 86% 87% 89% 93% 95% 98%

Hc (Oe) 0.200 0.015 0.018 0.022 0.034 0.082 0.120

mi 1300 22,000 22,500 19,000 13,200 10,000 8500

mmax 43,000 110,000 100,000 90,000 62,800 50,800 48,000

Bs (kG) 3.0 12.5 12.6 12.3 12.2 12.4 12.0

D.-T. Ngo et al. / Journal of Magnetism and Magnetic Materials 322 (2010) 342–347346

obtained in very low applied field (below 1 Oe). It allowssuggesting that the studied material is a suitable candidate forapplications of sensitive response e.g. sensitive magnetic sensor,small transformation. Moreover, at optimally annealed condition,a high saturation induction up to 12.5 kG is measured, slightlyhigher than that of pure Finemet [1]. It is noted that lowerannealing temperature (e.g. 520 3C) had been tested. Attemperatures lower than 530 3C, the crystallisation processbecame slowly, resulting in a unexpected mechanical hardnessof the alloy. These temperatures are lower than onsetcrystallisation temperature of the amorphous phase (530 3C,hence, the heat treatment required long keeping times to obtaina proper volume fraction of the a-Fe(Si) phase (e.g. 120 min toobtain 80% vol. of the a-Fe(Si) at 530 3C annealing temperature).This causes the alloy to be brittle-fracture and start oxidisingslightly. Therefore, lower annealing temperature ðo530 3CÞ is notinteresting so far for tailoring the magnetic properties of the alloy,and annealing at 530 3C in 40 min (either at vicinity of onsetcrystallisation temperature) should be considered as optimalcondition to obtain the best soft magnetic properties.

4. Conclusions

Crystallisation evolution, microstructure and magnetic proper-ties of Finemet-like Fe73.5Si17.5B5Nb3Cu1 nanocomposite materialhave been systematically investigated. Differential scanningcalorimetry and thermomagnetic measurements revealed thecrystallisation kinetics in the amorphous specimens relating tothe crystallisation of the a-Fe(Si) phase with the crystallisationactivation energy of 2:8170:03 eV, which is lower than pureFINEMET containing a lower Si content. The crystallisationevolution can be reasonably described by Kissinger’s and John–Mehl–Avrami (JMA) models. Diffraction analysis on TEM indicated

ARTICLE IN PRESS

D.-T. Ngo et al. / Journal of Magnetism and Magnetic Materials 322 (2010) 342–347 347

that the lattice parameter of the a-Fe(Si) in the annealedspecimens increases by increasing annealing temperature due tothe enhancement of diffusion of Si atoms into the crystal lattice ofthe bcc-Fe. By optimally annealing at 530 3C in 40 min, excellentsoft magnetic properties are obtained with very high permeabilityof 110,000 and a very low coercivity of 0.015 Oe. Ultrasoftmagnetic softness made the studied material is suitable for highlysensitive applications such as sensitive magnetic sensor, low-power transformer.

Acknowledgements

This work is completed by the financial support from theVietnam Fundamental Research Program for Natural Sciencesthrough the Grant 406506. Authors from Glasgow would like tothank Engineering and Physical Sciences Research Council andOverseas Research Students Awards Scheme (ORSAS) for support-ing for the work. We would like to express our sincere thanks to

Dr. S. McFadzean and Mr. B. Miller (Kelvin NanocharacterisationCentre, University of Glasgow) for their technical support of TEMmeasurements.

References

[1] Y. Yoshizawa, S. Oguma, K. Yamauchi, J. Appl. Phys. 64 (1988) 6044.[2] N. Chau, N.X. Chien, N.Q. Hoa, P.Q. Niem, N.H. Luong, N.D. Tho, V.V. Hiep,

J. Magn. Magn. Mater. 282 (2004) 174.[3] N. Chau, N.Q. Hoa, N.D. The, L.V. Vu, J. Magn. Magn. Mater. 303 (2006) e415.[4] E.Y. Kang, Y.B. Kim, K.Y. Kim, Y.H. Chung, J. Appl. Phys. 99 (2006) 08F111.[5] G. Herzer, IEEE Trans. Magn. 25 (1989) 3327.[6] H. Kissinger, Anal. Chem. 29 (1957) 1702.[7] M. Al-Haj, J. Barry, J. Mater. Sci. Lett. 16 (1997) 1640.[8] J. Bigot, N. Lecaude, J.C. Perron, C. Milan, C. Ramiarinjaona, J.F. Rialland,

J. Magn. Magn. Mater. 133 (1994) 299.[9] M.S. Leu, T.S. Chin, MRS Symp. Proc. 577 (1999) 557.

[10] A.C. Hsiao, M.E. McHenry, D.E. Laughlin, M.R. Tamoria, V.G. Harris, IEEE Trans.Magn. 37 (2001) 2236.

[11] C.-Y. Um, F. Johnson, M. Simone, J. Barrow, M.E. McHenry, J. Appl. Phys. 97(2005) 10F504.

[12] F. Richter, W. Pepperhoff, Arch. Eisenhuettenwes. 45 (1974) 107.