lable at ScienceDirect

Intermetallics 16 (2008) 1073–1082

Contents lists avai

Intermetallics

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

Mechanical alloying of Fe100�x�yNbxBy (x¼ 5, 10; y¼ 10, 15): From purepowder mixture to amorphous phase

J.J. Ipus, J.S. Blazquez, V. Franco, A. Conde*

Departamento de Fısica de la Materia Condensada, ICMSE-CSIC, Universidad de Sevilla, P.O. Box 1065, 41080 Sevilla, Spain

a r t i c l e i n f o

Article history:Received 25 April 2008Received in revised form 10 June 2008Accepted 13 June 2008Available online 26 July 2008

Keywords:A. Nanostructured intermetallicsA. Magnetic intermetallicsC. Mechanical alloying and milling

* Corresponding author. Tel.: þ34 95 455 28 85; faE-mail address: conde@us.es (A. Conde).

0966-9795/$ – see front matter � 2008 Elsevier Ltd.doi:10.1016/j.intermet.2008.06.006

a b s t r a c t

The mechanical alloying process of Fe75Nb10B15 and Fe85Nb5B10 systems has been studied from an initialmixture of elemental powders. The amorphization process is monitored by X-ray diffraction andMossbauer spectrometry. An amorphous phase is formed after 400 h milling only for Fe75Nb10B15 alloy,whereas a bcc supersatured solid solution is the final product after milling Fe85Nb5B10 alloy. For bothcases, a dispersion of w10% in the Fe content of the powder particles persists after 400 h milling. Powderparticle size, Cr content and lattice parameter of bcc phase are larger for the alloy with the highest Nbcontent.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Nanocrystalline materials are defined by a crystal size below100 nm. As a limit, amorphous materials are solid systems wherethe structural long-range order is lost. These materials have re-ceived much attention due to their physical properties (mechanicaland magnetic), which are clearly different from those exhibited byconventional microstructures and, in some cases, improve theirtechnological applicability [1–3]. One way to obtain nanocrystallinematerials is by partial devitrification of a precursor amorphousalloy during controlled thermal annealing. This technique controlsthe microstructure of materials and thus optimizes the propertiesof the final product. Another possibility is mechanical alloying,which has become a very versatile technique to directly producemetastable microstructures (amorphous, nanocrystallines, super-saturate solid solutions, etc.) [1] from elemental powders or alloys.During this milling process the material is submitted to fractureand cold welding phenomena, as well as intensive plastic de-formation, which define the powder morphology, microstructureand properties. The continuous storing of defects in the crystallinephase during milling process destabilizes it, leading to nano-crystalline and/or amorphous structures [1].

Nanocrystalline Fe–M–B type alloys (M¼ Zr, Nb, etc), so-calledNanoperm [4], are attractive due to their soft magnetic propertiesafter optimum thermal treatment and are used in commercial

x: þ34 95 461 20 97.

All rights reserved.

applications such as telecommunications, micro-devices andpower electronics [5,6]. Although these systems are generallyobtained by rapid quenching and subsequent annealing, nano-crystalline alloys of these compositions can be directly obtained bymechanical alloying of elemental powders. As soft magneticproperties depend on the structure of the material [7–9], itsstructural characterization is a very important task to understandthe system behavior and to predict its possible technologicalcapabilities.

In this study, two Fe100�x�yNbxBy (x¼ 5, y¼ 10 and x¼ 10,y¼ 15) alloys were produced by mechanical alloying from a mix-ture of pure elements and their morphological, compositional andmicrostructural evolution, as well as their thermal stability, werestudied as a function of milling time. The amorphization of theternary FeNbB system by rapid quenching methods has beenstudied previously [10]. Whereas compositions similar to x¼ 10 canbe obtained in amorphous structure, those similar to x¼ 5 cannotbe obtained as amorphous.

2. Experimental

Fe100�x�yNbxBy (x¼ 5, y¼ 10 and x¼ 10, y¼ 15) compositionswere prepared by ball milling in a planetary mill Fritsch Pulveri-sette 4 Vario from elemental powders (�99% purity), with particlesize d<200 mm for Fe and Nb and d<1 mm for B. For simplicity, thestudied alloys will be named in the following by their Nb content:Nb10 for Fe75Nb10B15 and Nb5 for Fe85Nb5B10. The initial powdermass was 30 g and the ball to powder ratio was 10:1. The rotationalspeed of the disk which supports the vials was 150 rpm and that of

Fig. 1. SEM images of both alloys after selected milling times.

J.J. Ipus et al. / Intermetallics 16 (2008) 1073–10821074

the vials was 300 rpm in opposite direction. After selected times,some powder was taken out from the vials to characterize themorphology, composition, microstructure and thermal evolution.The opening and closing of the vials were done under argonatmosphere in a Saffron Omega glove box to avoid oxygen andhumidity contamination.

Particle size distribution and morphology were studied byscanning electron microscopy (SEM) using secondary electrons(SEs) and backscattered electron (BSE) images in a Jeol JSM-6460 LVoperated at 30 kV. Compositional evolution was studied by energydispersive X-ray (EDX) analysis using an Incax-sight of Oxford In-struments (stationary spots at the center of each probed particle).Phases present and microstructure were studied from X-ray dif-fraction (XRD), using Cu Ka radiation, and Mossbauer spectrometry(MS). Mossbauer spectra were recorded at room temperature ina transmission geometry using a 57Co(Rh) source. Values of thehyperfine parameters were obtained by fitting with NORMOSprogram [11]. The isomer shift, I, was quoted relative to that of a-Feat room temperature. Thermal characterization of the samples wasstudied by differential scanning calorimetry (DSC) using a Perkin–Elmer DSC7 in Ar atmosphere.

3. Results

3.1. Morphology and composition

Fig. 1 shows SEM images of both alloys obtained after differentmilling times. For both alloys at short milling times, t< 20 h, theparticles are formed by joint layers and inclusions, which presentdifferent composition. In order to appreciate this heterogeneity inmore detail, Fig. 2 shows SE (Fig. 2a) and BSE (Fig. 2b) images takenon a typical particle after 2 h milling for Nb10 alloy. Fig. 2c showsEDX spectra taken on the different points marked in Fig. 2a. The Aspectrum, on a dark inclusion, only shows the emission line ofboron besides the typical background at low energy, so this zone isrich in this light element. The B and C spectra show zones rich in Feand Nb, respectively. Therefore, the heterogeneity of the individualpowder particles is evidenced for short milling times in the con-ditions used in this study.

After 20 h milling, the powders show pores and cracks on theirsurface for Nb10 and Nb5 alloys but no clear layers and, for longermilling times, t� 100 h, agglomeration of powder is observed. Thisprocess seems to depend on milling conditions, as other authors

Fig. 2. a) SE image, (b) BSE image of a typical particle of Nb10 alloy after 2 h milling and (c) EDX spectra of selected points indicated in (a).

J.J. Ipus et al. / Intermetallics 16 (2008) 1073–1082 1075

have detected agglomeration of particles much earlier for similarcompositions (e.g. after 5 and 15 h for Fe84Nb7B9 alloy with moreenergetic milling conditions [12]). The particles that form theseagglomerates are larger for Nb10 than for Nb5 alloy. After 400 hmilling, the agglomerates are not observed for Nb10 alloy but, forNb5 alloy, the particles are still agglomerated.

From SEM images, a statistical analysis of the average particlesize evolution, CdD, has been performed over w100 particles persample (Fig. 3) for both alloys. After 20 h milling, CdD increases overCdD> 200 mm for both alloys. As milling time increases, a decrease inCdD is observed and, about 100 h milling, this value is stabilized,being smaller for Nb5 alloy (w25 mm) than for Nb10 alloy(w50 mm). This indicates that a stationary situation between coldwelding and fracture has been achieved.

The compositional evolution was studied by EDX from a statisticalset of w20 particles of each sample. As B content cannot be quanti-tatively measured by EDX, compositional analysis of the systems is

0 100 200 300 4000

100

200

300

Nb10 Nb5

iteration step (x103)

<d>

[m

]

milling time [h]

0 10 15 20 25 30 35

simulated

5

Fig. 3. Experimental and simulated data of the average powder particles’ size, CdD, asa function of the milling time for both alloys.

referred to the relative amounts of Fe and Nb. In Fig. 4, histograms ofFe content of the powder particles are presented for Nb10 and Nb5alloys after different milling times. For short milling times, t� 5 h, it ispossible to find Nb rich particles and a high fraction of Fe rich particlesin both alloys. After 10 h milling, in Nb5 alloy a strong reduction in thebroadening of the compositional distribution, DCFe, is observed(calculated as the difference in Fe/FeþNb ratio between the Ferichest and the Fe poorest powder particles found). However, thisreduction occurs for 20 h in the Nb10 alloy, being a wider composi-tional distribution for the alloy with a higher Nb content. For longermilling times, t> 20 h, the broadening of the distribution of Fecontent is almost constant and, at the end of the studied range, thereis not a unique composition but a certain broadening is found(DCFe¼ 16 at.% of Fe for Nb10 and 8 at.% of Fe for Nb5).

Small Fe contamination is hard to measure in such Fe richcompositions. However, Cr is easily quantified as the initial powdermixture of this study is Cr free. Therefore, Fig. 5 shows the Crconcentration as a function of the milling time. For both alloys,a linear increase with the milling time is observed for t> 20 h in theexplored range, but for t< 20 h the increase in Cr content is faster(the extrapolation of the linear behavior to t¼ 0 h does not lead tothe actual initial Cr content). At the end of the explored range, theamount of Cr is higher for Nb10 (w2 at.%) alloy than for Nb5(w1 at.%) alloy, in agreement with the expected increase of hard-ness as Nb increases in Fe based alloys [13]. Similar Cr contami-nation has been found in other ball-milled systems [14].

3.2. Structural evolution

3.2.1. X-ray diffractionFig. 6 shows the XRD patterns of both alloys as a function of

milling time. For short milling times, t� 20 h, a slight broadening of(110) diffraction peak of the a-Fe phase can be observed for bothalloys. The full width at half maximum (FWHM) is approximatelythe same for both alloys at this stage; FWHM increases from 0.26 to0.52� 0.10� and from 1 to 20 h, respectively. This effect can be re-lated with the decrease of the crystalline size and an increase ofmicrostrains.

4

8

12

popu

latio

n

2

4

6

2

4

2

4

6

Nb10

0.2 0.4 0.6 0.8 1.0

2

4

Fe fraction0.2 0.4 0.6 0.8 1.0

Fe fraction

4

8

12

Nb5

2 h

4

810 h

2

4

620 h

4

8

1250 h

4

8400 h

Fig. 4. Histograms of the percentage of Fe content in total FeþNb content for both alloys at selected milling times.

J.J. Ipus et al. / Intermetallics 16 (2008) 1073–10821076

After 50 h milling, a strong broadening of (110) peak is ob-served, as well as a shift to lower values of 2q position of thispeak. Moreover, the different maxima of bcc-Nb are no longerdetected. These effects are related to the Nb incorporation intothe bcc-Fe lattice and to the formation of the supersaturatedsolid solution. For longer milling times, t> 200 h, the XRD pat-tern of Nb10 alloy shows a halo centered at 2q w 44� and the a-Fe (200) peak is reduced, being unappreciable for t¼ 400 h. Thisis related to the formation of the amorphous phase in thepowder particles.

3.2.2. Mossbauer spectrometryMossbauer spectra along with the hyperfine magnetic field

distributions of the different samples are shown in Figs. 7 and 8

0.0

0.6

1.2

1.8

Cr

[%]

0 100 200 300 400milling time [h]

N5B N10B

(7±2)* 10-4 at. % Cr/h

(3.6±0.3)* 10-3 at. % Cr/h

Fig. 5. Cr content as a function of milling time for both alloys. The slopes of the dif-ferent fitted lines are also indicated.

for Nb10 and Nb5, respectively. Mossbauer spectra were fittedusing a ferromagnetic site contribution with HF¼ 33 T for pure a-Fe phase (site-F) and two hyperfine magnetic field distributions;one for low field contributions, D1 (from 0 to 10 T) and otherfor high field contributions, D2 (>8 T). Furthermore, other sitecontribution but paramagnetic with quadrupolar splittingw0.5 mm/s (site-P) is necessary to fit the spectra of Nb10 for longmilling times. It is worth mentioning that, for such complexsystems as studied here, there is ambiguity between low fieldferromagnetic sites (<5 T) and paramagnetic ones. For t� 10 hmilling, the spectra only show a sextet with narrow absorptionpeak (width w0.30 mm/s) and hyperfine magnetic field 33 T forboth alloys, indicating that the a-Fe lattice has not been signifi-cantly affected by milling, in agreement with EDX (heterogeneous

40 50 60 40 50 60

2θ [degree] 2θ [degree]

inte

nsity

[a. u

.]

400 h

(110) O bcc FeX bcc Nb

50 h

20 h

Nb5Nb10 OO

XX

OX1 h

OX

Fig. 6. XRD patterns of both alloys after selected milling times.

0.15

0.30

0.06

0.12

0.08

0.16

0.5

1.01 h

20 h

50 h

rela

tive

tran

smis

sion

400 h

probability

35-8 -6 -4 -2 0 2 4 6 8 0 5 10 51 20 25 30velocity [mm/s] Bhyp [T]

Fig. 7. Mossbauer spectra and hyperfine magnetic field distributions for Nb10 alloy after selected milling times.

J.J. Ipus et al. / Intermetallics 16 (2008) 1073–1082 1077

powder particles, not really alloyed) and XRD (lattice parameterclose to pure a-Fe) results.

For Nb10 alloy spectra at t� 20 h, two new contributions appear,D1 and D2. This shows the existence of Fe atoms in three differentmain environments: first (site-F), pure bcc-Fe phase environment;second (D1), Nb-rich environments; and third (D2), Fe-rich envi-ronments. This latter contribution can be related with Fe atoms inthe a-Fe phase but in the presence of impurities of Nb, B and/or Cr(due to contamination by the grinding media) or to Fe atoms at theinterface region of nanocrystals [15]. However, in the Nb5 alloy,along with the crystalline contribution, site-F, only the D2 distri-bution is observed (D1 is negligible). This means that, in Nb5 alloy,Nb rich environments are not as significant as in Nb10 alloy, as itwould be expected.

-8 -6 -4 -2 0 2 4 6 8 0

rela

tive

tran

smis

sion

velocity [mm/s]

Fig. 8. Mossbauer spectra and hyperfine magnetic field di

4. Discussion

4.1. Simulation of powder size evolution

As it was previously mentioned, for 20 h milling, cracks areobserved in the particle surfaces, which may facilitate the fractureof the particles when the milling continues. The change in theevolution trend of CdD could be ascribed to changes in the me-chanical properties of the system due to rapid accumulation ofdefects into the particles [16]. This changes the balance betweenthe two processes responsible for the evolution of CdD: fracture andcold welding. In fact, a qualitative change in the mechanical be-havior of the powder is clearly observed for milling times longerthan 50 h. Until 50 h milling, the powder sticks on the vial wall and

0.1

0.2

0.6

0.06

0.12

5 10 15 20 25 30 35

0.1

0.2

0.5

1.0

400 h

50 h

1 h

20 h

probability

Bhyp [T]

stributions for Nb5 alloy after selected milling times.

0

50

100

0 2 4 6 8 10

0 100 200 300 400

90

100

Nb10

iteration steps (x 107)

a

simulatedNb5

b

milling time [h]

CF

e [%

at.

Fe]

C

Fe

[at.

% F

e]

Fig. 9. Experimental and simulated values of (a) DCFe and (b) CFe as a function of themilling time (experimental) and iteration steps (simulation). Iteration step axes havebeen conveniently rescaled to show the agreement with the experimental data.

0 100 200 300 400

0.00

0.02

0.04

0

10

20

30

40

milling time [h]

<D>

[nm

]

Nb10

Nb52.86

2.87

2.88

2.89

2.90

2.91 a

b

c

a [Å

] [%

]

Fig. 10. a) Lattice parameter, a; (b) minimum crystal size, CDD; and (c) maximum mi-crostrain, 3; as a function of milling time for both alloys.

J.J. Ipus et al. / Intermetallics 16 (2008) 1073–10821078

ball surface. However, for longer milling times the powder detachesfrom milling media surfaces, noticeably increasing the amount ofloose powder.

Based on these two processes, a basic simulation algorithm wasperformed to obtain a first approximation of the tendency followedby CdD during milling. The initial system consist of 500 equally sizedparticles. Once a random particle is chosen, its probability to be coldwelded with another particle is defined by Pcw¼ exp(�di/dc). Thisexpression depends on the size of the i particle, di, and a criticalsize, dc. Particles with di> dc will tend to fracture, while if di< dc,the particle will tend to join with another. As can be inferred fromthe experimental evolution of CdD, dc is not constant during themilling process. In our simple simulation, only two different valuesof dc were used (supported by the abrupt change in the mechanicalproperties detected between 20 and 50 h). A dc value 100 times theinitial size was used to reproduce the rapid increase in CdD at shortmilling times. For long milling times, the constant experimentalvalue of CdD was used as dc. Simulation results reproduce the ex-perimental ones after properly rescaling the iteration steps withthe milling time (see Fig. 3).

4.2. Simulation of compositional evolution

EDX results indicate that the composition of particles is notunique, even at long milling time but compositional distributionsachieve a stationary situation. The parameter DCFe was simulatedconsidering a system of particles defined by a 100 componentsbinary array, being 0 for Nb and 1 for Fe (1% in compositionalresolution). Two random particles will interact changing a portionof their arrays with the same aleatory size. Therefore, powderparticle’s size and number remain unchanged and only compo-sitional evolution is simulated in this very simple simulation.Results are shown in Fig. 9a along with experimental ones forcomparison. Experimental and simulation results agree evenquantitatively (once iteration steps are conveniently rescaled),showing a stationary situation of the system with a compositionaldistribution and, therefore, the existence of a certain degree ofheterogeneity (DCFe¼ 18 at.% of Fe for Nb10 and 14 at.% of Fe forNb5).

Other parameter which enables to follow the compositionalevolution is the most probable Fe content in the particles, CFe,

(shown in Fig. 9b). This parameter rapidly decreases with themilling time close to the nominal composition for both alloys. After20 h milling, CFe¼ 89 and 95� 2 at.% for Nb10 and Nb5, re-spectively. An increase with respect to the nominal compositions(88 and 94 at.% for Nb10 and Nb5, respectively), although into theexperimental error, would not be surprising and could be expecteddue to Fe contamination from milling media. The evolution of CFe

was also obtained from the simple simulation described above (alsoshown in Fig. 9b, using the same iteration step/milling time ratiothan for Fig. 9a) and a good agreement with the experimental datawas also found.

4.3. Microstructural evolution

4.3.1. X-ray diffractionThe broadening observed from 20 to 50 h milling in the a-Fe

(110) peak is bigger for Nb10 (FWHM¼ 1.55� 0.10�) than for Nb5(FWHM¼ 1.08� 0.10�). This fact can be related to several effects,for example, a smaller grain size or a higher value of microstrainsfor Nb10 than for Nb5 alloy. However, another effect can be con-sidered after EDX results, where a wider distribution of particlecomposition was found for Nb10 than for Nb5 alloy. Therefore, theformed crystals are expected to present a wider distribution oflattice parameters. The shift of the position of the a-Fe (110) peak tolower values of 2q is also bigger for Nb10 (2q¼ 44.35� 0.05�) thanfor Nb5 alloy (2q¼ 44.45� 0.05�). This result agrees with thehigher amount of Nb in the Nb10 alloy, thus the crystals of the a-Fephase must incorporate more Nb atoms in the lattice.

0.0

0.4

0.8

0 100 200 300 4000.0

0.4

0.8

Nb5

Nb10

Fe fr

actio

n

Site-F

D1

D2

Site-P

milling time [h]

Fig. 11. Area fraction of the different Mossbauer contributions to the total fitting asa function of milling time for both alloys.

400 500 600 700 800 900 1000temperature [K]

dH

/dt

400 h

400 h 50 h

50 h

20 h

20 h

5 h

5 h Nb10

Nb5

0.2

W/g

(exo

)

Fig. 12. DSC scans at 40 K/min for both alloys after selected milling times. Symbols aresuperimposed over the curves to distinguish them.

J.J. Ipus et al. / Intermetallics 16 (2008) 1073–1082 1079

For Nb10 alloy an amorphous halo is observed for milling timeslarger than 200 h. This amorphous phase should be formed becausethe Nb incorporation in the bcc-Fe lattice destabilizes it, due to itsvery low solubility, allowing disordering and/or amorphous phaseformation. Moreover, Nb content affects the velocity and energynecessary to form the amorphous phase [17]. For example, theFe82Nb6B12 alloy transforms to amorphous phase using a processcontrol agent (PCA) and more energetic conditions [18], and theFexNbx�1 system transforms to amorphous phase only for a com-position range between x¼ 0.3 and 0.7 [19].

In Fig. 10a, the lattice parameter of the a-Fe, a, is presented asa function of milling time. Before 20 h milling, the lattice param-eter for both alloys (a¼ 2.864� 0.003 Å) is close to pure a-Fe(a¼ 2.8665 Å) and, after 50 h, an increase is observed (due to theincorporation of Nb and B atoms in the bcc-Fe crystals). The valueof a is higher for the alloy with more Nb content. However, thevalue obtained for Nb5 alloy is higher than that found forFe85Nb6B9 supersaturated solid solution [20]. The lower Fe con-tamination observed in the present study could explain thisdifference.

From these results, it is possible to consider that the actualsystem after long milling times is formed by powder particles (ina certain compositional range) containing bcc-Fe(Nb,B) crystallites(which composition may depend on that of the powder particle).Taking into account the difficulty of separating the different effectscontributing to the broadening of diffraction peaks (small crystalsize, microstrains, existence of a lattice parameter distribution) theanalysis has been simplified to calculate a minimum average grainsize, CDD, (Fig. 10b) and a maximum value of microstrains, 3,(Fig. 10c), considering each individual effect as the only responsiblefor the broadening of the XRD peaks. CDD reaches a minimum stable

value after 50 h milling, being smaller for Nb10 alloy (CDD¼ 5 and7 nm for Nb10 and Nb5, respectively). This result is comparablewith those reported in other studies for similar alloys [1,20,21], aswell as its dependency with the Fe content [20]. The maximummicrostrains (just a limit value) reach w2% after 100 h milling forboth alloys.

4.3.2. Mossbauer spectrometryAs milling times increases, the single site with HF¼ 33 T is not

enough to fit the spectra and new contributions are needed. Theappearance of these new contributions, D1 and D2, is related to thepresence of Nb and B in the neighborhood of Fe crystalline sites (inagreement with the increase of the lattice parameter from XRDresults after t¼ 50 h) and with Fe atoms outside crystals.

Fig. 11 shows the fraction of Fe atoms in the different contri-butions. For Nb10 alloy, the disappearance of site-F after 100 h isobserved, as well as the increase of the paramagnetic contributions(Nb rich environments) with the appearance of site-P after 200 hmilling. However, no significant changes for Nb5 alloy are observedafter 50 h milling. Therefore, microstructural changes are notexpected in agreement with XRD results.

Mossbauer and XRD results allow us to distinguish two mi-crostructures at the end of milling experiment performed(t¼ 400 h). On the one hand, for Nb5 alloy, there are a-Fe nano-crystals with pure bcc environments (site-F) and the presence ofimpurities taken into account by D2 contribution. This contribu-tion also considers the interface effect. On the other hand, for Nb10alloy, the pure bcc environment is no longer detected, and it is onlypossible to distinguish three Fe atomic environments: one ferro-magnetic (possibly ascribed to residual crystallites) and twoparamagnetic environments related with the amorphous phaseformation observed by XRD. It is worth mentioning that all thepowder particles of the Nb10 sample for 400 h milling moved inthe vicinity of a small magnet, indicating that there are no pureparamagnetic powder particles but every particle must be com-posed of a mixture of ferromagnetic crystalline and paramagneticamorphous phases.

The amount of amorphous phase can be estimated from bothMossbauer (low field contributions) and XRD results (aftera deconvolution of the experimental maximum using two Pseu-dovoigt functions). Both techniques agree showing an increase ofthe amount of amorphous phase with milling for t> 100 h. FromMossbauer results, the area fraction of the low field contributionincreases from 38 to 77% from 100 to 400 h milling. In the case of

40 50 60

600 K

800 K

as-milled

2 [degree]

1000 K

500 1000

0.286

0.288

T [K]

a [n

m]

Fig. 13. XRD patterns for Nb5 alloy after different heated temperatures. Inset showsthe lattice parameter, a, as a function of the heating temperature.

40 50 60

600 K

800 K

as-milled

2θ [degree]

1000 K

500 10000

50

100

T [K]

XC [%

]

Fig. 14. XRD patterns for Nb10 alloy after different heated temperatures. Inset showsthe crystalline volume fraction, XC, as a function of the heating temperature.

J.J. Ipus et al. / Intermetallics 16 (2008) 1073–10821080

XRD measurements, the amorphous volume fraction, increasesfrom 29 to 80% and from 100 to 400 h milling.

4.4. Thermal stability

Fig. 12 shows the DSC scans of different as-milled samples takenat a heating rate of 40 K/min for both alloys. For short millingtimes, relaxation phenomena [20,22] start at w700 K. However,for samples milled t� 50 h for Nb10 and t� 20 h for Nb5, a de-viation from the baseline is detected at w410 K. For Nb5 alloymilled 50 h or more, (after a supersaturated solid solution is

-8 -6 -4 -2 0 2 4 6 8 0

rela

tive

tran

smiti

on

velocity [mm/s]

Fig. 15. Mossbauer spectra and hyperfine magnetic field

formed), an exothermic peak is observed at w900 K. For Nb10 alloy,after 400 h milling a clear exothermic peak is clearly observed atw850 K.

In order to obtain activation energies, Q, by Kissinger method[23], scans at 5, 10, 20, 40 and 80 K/min were recorded on Nb5 andNb10 samples milled 400 h, leading to an activation energy of1.3� 0.1 eV for the broad exotherm at w750 K of both alloys. Forthe DSC peaks detected for Nb5 alloy at 900 K and Nb10 at 850 K,Q¼ 3.6 eV and 3.8 eV, respectively. Crystallization temperaturesand activation energies of these exothermic processes are inagreement with values found in the literature for devitrification ofamorphous/partially crystalline ribbons [6,24].

0.1

0.2

0.08

0.16

0.24

0.2

0.4

5 10 15 20 25 30 35

0.2

0.4

0.6

as-milled

600 K

probability

800 K

Bhyp [T]

1000 K

distribution for Nb5 alloy after annealed treatment.

0.1

0.2

0.1

0.2

0.03

0.06

-8 -6 -4 -2 0 2 4 6 8 0 5 10 15 20 25 30 35

0.1

as-milled

probability

velocity [mm/s]

rela

tive

tran

smiti

on 600 K

800 K

Bhyp [T]

1000 K

Fig. 16. Mossbauer spectra and hyperfine magnetic field distribution for Nb10 alloy after annealed treatment.

J.J. Ipus et al. / Intermetallics 16 (2008) 1073–1082 1081

In order to clarify the nature of the different transformationpeaks detected by DSC, XRD and Mossbauer techniques were ap-plied to samples heated up to different temperatures (600, 800and 1000 K at 40 K/min). Figs. 13 and 14 show XRD patterns ofheated samples along with the corresponding as-milled ones forNb5 and Nb10 alloys, respectively. Figs. 15 and 16 show Mossbauerspectra and hyperfine field distributions of heated samples alongwith the corresponding as-milled ones for Nb5 and Nb10 alloys,respectively.

For Nb5 alloy heated up to 600 K, a slight decrease of the latticeparameter and a decrease of 17% in FWHM of (110) a-Fe maximumis observed. After heating up to 800 and 1000 K, a continuous de-crease in the lattice parameter down to that of pure a-Fe (see insetin Fig. 13) and in FWHM (22 and 28%, respectively) is observed withrespect to the as-milled sample. This can be ascribed to a pro-gressive enrichment in Fe of the a-Fe phase and the decrease in themicrostrain and crystal grain coarsening. However, it is worthmentioning that the FWHM of the sample heated up to 1000 K,1.3� 0.1�, yields a CDD w7 nm. This indicated that the crystallinestructure is stabilized by the Nb atoms rejected from the crystals,even after this temperature. Mossbauer technique confirms theseresults, with a progressive increase of the 33 T contribution asheating temperature increases.

For Nb10 alloy, neither XRD nor Mossbauer show significantchanges for the sample heated up to 600 K with respect to the as-milled one. Heating up to 800 K, yields a sharpening of the centralregion of the halo detected in the XRD pattern, as well as, an in-crease in the high field contributions (33 T sextet appears). Thiscould be ascribed, analogously to Nb5 alloy, to the purification ofthe remaining a-Fe(Nb,B) phase, as the crystalline volume fractiondoes not change significantly (see inset Fig. 14). After heating up to1000 K, crystallization of the amorphous phase occurs, evidencedby a clear (110) peak of the a-Fe phase. The lattice parameter of thisphase is close to that of pure a-Fe (2.864 Å) and the average crystalsize is w9 nm. This indicated that the amorphous structure is al-most still up to 800 K and the crystalline structure is achieved afterthe DSC peak detected at 850 K. These results are in agreementwith Mossbauer spectra (see Fig. 16), where high field contributionsincrease.

5. Conclusions

From the present study on the mechanical alloying of the ter-nary FeNbB system, several conclusions can be remarked.

� The average powder particle size at long time milling is smallerfor the alloy with the lowest Nb content.� At long time milling, the composition of powder particles are

not the same but a dispersion of w10% in their Fe content stillexists.� At long milling times, Cr contamination linearly increases with

the milling time for both alloys, being higher for the alloy withthe highest Nb content.� For milling times t� 50 h, the only phase present in the sam-

ples is a bcc-Fe(Nb,B) supersaturated solid solution for bothalloys. After 400 h, only for the alloy with the highest Nbcontent, the presence of amorphous phase is evidenced by anamorphous halo observed by XRD, as well as the detection ofan exothermic process in DSC ascribed to crystallization. Thisamorphous phase is stable up to w850 K.� From Mossbauer results, no new Fe atomic sites appear for the

alloy with the lowest Nb content after 50 h milling, being themore important contribution at 33 T (pure bcc-Fe site). How-ever, for the alloy with the highest Nb content, the Fe atomicenvironment continuously evolves increasing the fraction ofparamagnetic environments.

Acknowledgments

This work was supported by the Spanish MEC and EU FEDER(Project MAT2007-65227) and by the PAI of the Regional Govern-ment of Andalucıa (Project P06-FQM-01823). J.J.I. acknowledgesa fellowship from the Spanish Ministry of Education and Science.J.S.B. acknowledges a research contract from the RegionalGovernment.

References

[1] Suryanarayana C. Prog Mater Sci 2001;46:1–184.

J.J. Ipus et al. / Intermetallics 16 (2008) 1073–10821082

[2] Gleiter H. Prog Mater Sci 1989;33:223–315.[3] Gleiter H. Acta Mater 2000;48:1–29.[4] Suzuki K, Kataoka N, Inoue A, Makino A, Matsumoto T. Mater Trans JIM 1990;

31:743–6.[5] Makino A, Hatanai T, Inoue A, Matsumoto T. Mater Sci Eng A 1997;226–228:

594–602.[6] McHenry ME, Willard MA, Laughlin DE. Prog Mater Sci 1999;44:291–433.[7] Hernando A, Vazquez M, Kulik T, Prados C. Phys Rev 1995;B51:3581.[8] Chicinas I, Jumate N, Matei Gh. J Magn Magn Mater 1995;140–144:1875–6.[9] Szabo S, Beke DL, Harasztosi L, Daroczi L, Posgay Gy, Kis-Varga M. Nanostruct

Mater 1997;9:527–30.[10] Landolt-Bornstein, New Series, Vol. III/37A, vol. 62–63. Berlin: Springer–Ver-

lag; 1997. pp. 173–4.[11] Brand RA, Lauer J, Herlach DM. J Phys F Met Phys 1983;13:675–83.[12] Lu W, Yang L, Yan B, Huang W, Lu B. J Alloys Compd 2006;413:85–9.

[13] Stephens JR, Witzke WR. J Less-Common Met 1976;48:285–308.[14] Ipus JJ, Blazquez JS, Franco V, Conde A. Intermetallics 2007;15:1132–8.[15] Greneche JM, Slawska-Waniewska A. J Magn Magn Mater 2000;215–216:264–7.[16] Zhang BQ, Lu L, Mai MO. Physica B 2003;325:120–9.[17] Jartych E, Oleszak D, Zurawicz JK. Hyp Int 2001;136:25–33.[18] Caamano Z, Perez G, Zamora LE, Surinach S, Munoz JS, Baro MD. J Non-Cryst

Solids 2001;287:15–9.[19] Yang JY, Zhang TJ, Cui K, Li XG, Zhang J. J Alloys Compd 1996;242:153–6.[20] Sunol JJ, Gonzalez A, Saurina J, Escoda Ll, Bruna P. Mater Sci Eng A 2004;375–

377:874–80.[21] Blazquez JS, Franco V, Conde CF, Conde A. Intermetallics 2007;15:1351–60.[22] Oleszak D, Shingu PH. J Appl Phys 1996;79:2975–80.[23] Henderson D. J Non-Cryst Solids 1979;30:301–15.[24] Makino A, Suzuki K, Inoue A, Masumoto T. Mater Sci Eng A 1994;179–180:

127–31.